PDF: drop files first, choose a tool next
A shared PDF workspace with contextual tool choices, thumbnails, page editing and clearer downloads on mobile and desktop. Files stay on your device.
Open PDF →Changelog
Allt vi släpper på Krymp — verktyg, förbättringar, datadelning och säkerhet.
A shared PDF workspace with contextual tool choices, thumbnails, page editing and clearer downloads on mobile and desktop. Files stay on your device.
Open PDF →Quick starts, inline editing, portrait frames, unified exports and larger on-set camera notes.
Plan a film →Automatic frame strip, channel presets, copying and direct downloads with simpler mobile controls.
Create a thumbnail →Create QR codes for links, WiFi and contacts in a focused workspace. Pick a palette, copy the image or download PNG, SVG, PDF and EPS. Logos are protected automatically and detailed settings open on demand.
Create QR →After download, Image, Social media and Moodboard offer an optional next step with Hasty, and Video with Fenny. Sister services and expert tips stay discreet, with no partner-link tracking or account requirements. Ellary is no longer shown.
Drop images onto the canvas for an automatic composition and shuffle layouts. Open styling only when needed, swap images with two clicks and download a double-resolution PNG.
Create a moodboard →Channel choices and adjustments appear after upload. See variants together, adjust focus and choose which ones to include in the ZIP.
Open Social Media →Drop a video and start with Web, MP4 and 1080p. Settings stay collapsed, audio can be muted per file and finished videos can be previewed before download.
Compress video →Balanced compression and auto format work immediately. Open settings, rename and sharing only when needed, compare from the thumbnail and see savings next to the download.
Compress images →Tell us what an image is for and we suggest format and size. Audio mastering can suggest a mode, transcripts get tips before export, big drops get a suggested tool and failed files get a way forward. Files never leave your device.
Try it in Image →In the PDF tool you can now type what the file is for (e.g. "email it" or "archive it") and AI suggests the right compression level — and honestly warns when the file probably can't get noticeably smaller. Only your text and facts about the file are sent, never the file.
Try the PDF tool →When you drop a file you can now type what you want, e.g. "make it smaller for email" or "remove the sound", and we suggest the best tool. Only your text is sent, never the file.
Opening a shared link to e.g. the English home page could previously bounce you to Swedish if you had ever picked Swedish in the language switcher. The address you open now always wins, regardless of your saved language.
English home →Switching from an English page like /en/compress-video to Swedish now takes you to /video instead of back to the same English page or the home page. The same applies to German, French and Spanish.
Compress video →Previously you could be bounced to an English info page if you had English saved and clicked a Swedish URL like krymp.io/bild. You now stay on the page you asked for — automatic language redirect only happens on the bare home page.
Compress image →Every compressed PDF is reopened and checked page by page before you get it. If anything goes wrong you get a safer version or the original with an explanation. Stencil and colour-masked images are left alone, and signed documents are never rewritten.
Compress PDF →Light and Balanced now leave hidden image transparency masks untouched. Logos and cut-out images keep their transparency without black boxes or damaged PDF files.
Compress PDF →Print-ready (CMYK) images and images with extra encoding layers now shrink too, and identical images are counted once. If the file still can't shrink you get the reason — whether the weight is in fonts, vector graphics, attachments or images we can't touch without damaging them.
Compress PDF →If your PDF has clickable links they stay in the compressed file — even with Hard, where the pages are redrawn. Both web addresses and jumps to other pages inside the document keep working.
Compress PDF →The two gentler options used to barely change anything. Now the images inside the PDF are shrunk — Light re-saves them at high quality, Balanced scales them down further. The text is never touched, so you can still select and search the document. If the file doesn't get smaller you get the original back instead of a fake win.
Compress PDF →The tool now measures how much background noise your recording has and tunes the cleanup to it, so hum and room noise in the pauses almost disappear while the voice stays. You choose light, normal or max. On phones there are big buttons, a draggable timeline in the before/after player and a download button right below.
Master audio →You can now choose whether the finished video is saved as MP4 or AVI. On phones the timeline is taller, the trim handles are bigger, and there are buttons to fine-tune the start and end exactly.
Trim video →Drop in an audio file, pick Voice & podcast, Music or Recording — Krymp removes background noise, evens out the volume and makes the sound crisp. Compare with the A/B button and download. Nothing is uploaded.
Master audio →Drop in a video, drag the handles on the timeline and delete the parts you don't want. Split into several parts, undo anything, play the result and download the finished file. Nothing is uploaded.
Trim video →A blank document used to be created on every visit, burying your saved notes behind empty tabs. The editor now waits for your saved documents to load and reopens the one you last worked in.
Markdown →On mobile, every tool – open file, clean up, formatting, view options, fullscreen, focus mode and theme – now lives in a single Tools menu, giving the editor more room.
Markdown →Write tables right in the text with a dedicated table bar and jump between cells with Tab. On mobile, switch between Files, Write and Read with three buttons. You can also open Word (.docx) files and PDFs straight in the editor.
Markdown →Double-click a tab in Markdown to rename the document right there, without using the title field.
Markdown →The Markdown document name now sizes to its content and the saved checkmark sits right next to the title instead of far to the right.
Markdown →We now load only the language you actually use and split icons per page. The startup file shrank by about 1.8 MB, so the tools — especially the Markdown editor — appear faster.
Markdown →Cmd/Ctrl+K now opens the command menu from inside the editor, and the close button on document tabs is visible on mobile.
Open Markdown →The editor and reading view now adapt to the length of your document, so you scroll the page instead of a small box.
Open Markdown →Tool icons are now visible at the top, tabs sit below them, and the document title sits right above the editor. The saved icon sits next to the title.
Open Markdown →Removed clutter in the top bar, added simple formatting icons and a fullscreen button.
Open Markdown →The editor now spans the full available width, giving you more room to write and preview.
Open Markdown →Bullet lists now show real bullets in normal text colour, and the file name follows the document's first heading.
Open Markdown →The editor sometimes felt like the page had frozen. Now a clear loading indicator appears and the editor starts faster.
Open Markdown →The start screen is gone — you land directly in a fresh document and can start writing right away.
Open Markdown →Text is styled as you type, selecting it brings up bold, italic, link and heading right there, and ⌘K opens a search box with everything the tool can do. Pasting a table from Excel turns it into a table.
Open Markdown →Set your own background, text colour, accent colours, font and an icon (pick or upload) right in the audiogram. After download the film plays inside the tool with a checklist of what is actually visible.
Create audiogram →Audiogram now has five complete visual directions coordinating colour, typography, waveform, captions and motion. Preview and export use exactly the same graphics.
Create audiogram →All downloads now happen through the Export button, where you pick PDF, Markdown, CSS or everything in a zip. Images and icons are included, and the PDF uses the system's own colours.
Design system →The PDF now includes everything: a table of contents with page numbers, numbered sections, ready-to-paste CSS variables and a closing page about the document.
Design system →The design system preview has three new views: Colours (all colours, custom colours and shades), Images and Icons — so everything you pick is visible right away.
Design system →Upload images to your design system — they are compressed automatically in your browser — and pick icons from our icon library. Everything is included in the PDF, Markdown and zip.
Design system →You now get ten-step shade scales, ready-made dark mode colours, your own extra colours, full control of the type scale, plus icon sizes, breakpoints, motion and layers. The preview shows a page, buttons, a form and a table — and everything is included in the PDF, Markdown and CSS.
Open design system →Paste a link or drop an image — Krymp reads colours, fonts, sizes, spacing, corners and shadows and builds a complete design system you can download as PDF, Markdown or CSS.
Open Design system →There is now a press page where you can download our media kit, logo and color codes. Every common color also has its own page with all formats and readability, and color pairs show right away whether text is readable.
Open the press page →Website analysis now picks up more colors, grouped into main colors, text/background and other tones — click the ones you want. Fonts are detected automatically, you get matching color suggestions, and the preview follows you as you scroll.
Open Brand Kit →The tool is now called Brand kit and lives at /brandkit (old /farger redirects here). You get a ready-made kit right away and are guided through three steps: pick a starting point, adjust colors and fonts, download. Everything extra sits under More.
Open brand kit →The color tool remembers your recent colors, lets you save favorites, pick a color straight from an image and test contrast against any background. Sliders are now hue, saturation and lightness, with R, G and B under Advanced.
Open the color tool →Media kit, brand kit & colors, icons, favicon generator and the color converter are now grouped under Branding in the menu.
See the tools →The PDF-to-Markdown tool gets a larger, clearer drop zone with an icon, headline and stronger visual feedback on drag-and-drop. On mobile, a big green button lets you pick files. Info cards and conversion mode are moved below the zone for a better overview.
PDF to Markdown →Enter a color in any format and get all the others instantly: HEX, RGB, HSL, HSB, CMYK, OKLCH, LAB and the nearest Pantone. See contrast against white and black, printability, tints and shades and matching colors — and download it all as PDF, Markdown or an image.
Open Color converter →Drop several PDFs and get clean Markdown back. We show how many AI tokens you save, detect headings, lists and tables, strip headers and page numbers, and run text recognition on scanned files. Copy for AI, download .md or grab everything as a ZIP.
PDF to Markdown →Drop several Word files at the same time: each file gets its own status row (waiting, working, done, failed), its own download and a ZIP for everything finished. Global drops now remember the tool you picked last time.
Convert files →Picking tools like Crop, Video, GIF, Audiogram, Markdown or Anonymise after a drop used to open an empty page. The file is now handed over automatically.
Dropping a file anywhere on the site now asks what you want to do, with a recommended option first. The converter also accepts more Word files and turns them into PDFs that keep headings, lists, tables and images.
Open the converter →Every compression figure we publish now points to a traceable evidence record with date and method, and the methodology page lists the sources openly.
See the methodology →The sitemap now only reports change dates for pages that actually changed, and robots.txt has a documented per-bot policy — AI search engines are welcome, pure scrapers that never cite sources are not.
The homepage heading now describes the service — compress images, video and PDFs — in all five languages.
The tools button in the header now has a thicker icon on the left and the label "Tools" is displayed in uppercase for a more prominent and clear CTA.
Removed the numbered "01 / Live" eyebrow above the compare section on the home page for a cleaner, more minimalist look.
The section near the bottom of the home page has a new heading, a short intro and a plain explanation under each point, making it easier to understand what makes Krymp different.
The rotating tagline in the centre of the main header is now hidden on mobile screens so the menu doesn't become too wide or overflow.
The format river on the homepage now uses category-based colors and pastel tones so image, video, audio and document formats are easier to tell apart at a glance.
Homepage →The final section on the homepage has been broadened. It now speaks to all file types and tools — with quick links to image, video, PDF, AI, text and security.
Homepage →We removed the old privacy section from the homepage to make the content easier to scan.
The mobile menu now closes as soon as you pick a tool or link — and when you tap outside it.
The mega menu is now perfectly centred on the page, and the header animates out to the same width when you open Tools — and back again when you close it.
All tools →The main header's Tools button is now white with dark green text, a hamburger icon and a pulsing green ring that draws the eye. The rotating tagline in the centre is also bolder.
All tools →The main header's Tools button now has a dropdown icon, stronger contrast, a more visible shadow, and a light border so it stands out and is easier to read.
The rotating tagline in the main header is now displayed in uppercase for a clearer, more graphic look.
The main header is now significantly narrower and more centered, taking up less space and staying out of the page content.
The header is wider with all dividers removed. The Tools button is now green and easy to spot, the mega menu animates open, and the centre message rotates through short benefits.
Removed the section showing newsrooms, agencies and public-sector teams from the homepage to keep the content more focused and relevant.
The menu is now minimal: logo, the line "Free tools for your files", Tools and language. Images, Video, Social, About and Settings live inside the tools menu.
All tools →The tools menu was rebuilt: a category rail on the left with tools and short explanations on the right on desktop, plus a simple drill-in list on mobile. Search now understands everyday words like "email", "instagram" and "too big".
All tools →The workflow section (04 / Arbetsflöde) has been removed from the home page to make the landing page even more focused and minimal.
The "Helt gratis" section has been removed from the home page to create an even cleaner and more focused landing page.
The numbers section (03 / Siffror) has been removed from the home page to create a cleaner and more focused landing page.
The tools section further down the home page has been removed to make the landing page cleaner and more focused on the most important content.
The headline on the home page has been updated to "Fix the file." to clarify that Krymp's core tools are free.
Pages are now served from CDN cache instead of being re-rendered on every visit, and the home page's structured data shrank from 69 kB to 5 kB. First load is noticeably faster, especially on mobile.
The final wave of Swedish long-tail pages is live: images for Vinted (800×800, no HEIC), PDFs for Swedish courts (10 MB on Mina sidor), pre-OneDrive compression (5 GB free / 1 TB M365), Etsy product images (2000×2000 square), PDFs for rental agreements & landlords (email attachment under 10 MB) and pre-Facebook-Messenger compression (images <1 MB, video <25 MB — no extra graininess). Each ships its own SEO metadata, HowTo/FAQ JSON-LD and concrete limits.
Open the Vinted guide →Three more dedicated landing pages are live: Snapchat images (1080×1920 9:16 without autozoom), PDFs for contracts & e-signing (DocuSign, Scrive, Oneflow, Adobe Sign — signature fields preserved) and pre-Google Drive compression (more room on the free 15 GB shared with Gmail & Photos). Each ships its own SEO metadata, HowTo/FAQ JSON-LD and concrete limits.
Open the contracts guide →Three more dedicated landing pages are live: TikTok profile images (1080×1080, sharp even at 48×48), PDFs for Swedish bank submissions (10 MB caps at SEB/Swedbank/Handelsbanken/Nordea) and pre-iCloud compression (faster sync, more room on the free 5 GB plan). Each ships its own SEO metadata, HowTo/FAQ JSON-LD and concrete limits.
Open the bank guide →Three more dedicated landing pages are live: images for Google Business Profile (720×720 min., under 5 MB), PDFs for housing-agency 5 MB caps (income statements, references) and pre-Dropbox compression (faster sync, more room on the 2 GB plan). Each ships its own SEO metadata, HowTo/FAQ JSON-LD and concrete limits.
Open the Google Business guide →Three more dedicated landing pages are live: images for Blocket ads (under 500 kB for fast load), PDFs for Skatteverket's e-service (10 MB per attachment) and pre-WeTransfer compression (faster uploads even under the 2 GB cap). Each ships its own SEO metadata, HowTo/FAQ JSON-LD and concrete limits for the target service.
Open the Skatteverket guide →Three more dedicated landing pages are live: YouTube thumbnails (1280×720 under 2 MB), PDFs for Arbetsförmedlingen's e-service (10 MB) and files for Outlook attachments (20 MB per email). Each ships its own SEO metadata, HowTo/FAQ JSON-LD and exact limits for the target service.
Open the YouTube guide →Three more dedicated landing pages are live: images for LinkedIn Ads (1200×627 / 1080×1080, under 5 MB), PDFs for Migrationsverket's e-service (5 MB cap) and files for Notion (5 MB on the free plan). Each ships its own SEO metadata, HowTo/FAQ JSON-LD and exact limits for the target service.
Open the LinkedIn guide →Three more dedicated landing pages: compressing PDFs for CSN (Swedish student aid), files for Discord (free 10 MB / Nitro 500 MB) and 2160×1080 event covers for Eventbrite. Each ships its own SEO metadata, HowTo/FAQ JSON-LD and concrete limits for the target service.
Open the Discord guide →Three more dedicated landing pages are live: compressing PDFs for Försäkringskassan (Swedish social insurance), files for Slack threads, and photos for Booking.com listings. Each ships its own SEO metadata, HowTo/FAQ JSON-LD and concrete facts about the target service's limits and recommendations.
Open the FK guide →Dedicated landing pages for compressing Airbnb listing photos, passport/driving-licence photos and files under Microsoft Teams' upload limit are now live. Each page ships its own SEO metadata, HowTo/FAQ JSON-LD and clear facts about the target service's limits — no need to hunt them down.
Open the Airbnb guide →Seven side-by-side comparisons between Krymp and the biggest global tools (iLovePDF, Smallpdf, TinyPNG, iLoveIMG, Squoosh, Adobe Express, CloudConvert) are now available in English at /en/compare/*. Each page ships its own title, description, JSON-LD (Article, Breadcrumbs, FAQ, Speakable) and hreflang linking back to the Swedish version — so AI search and Google can surface the right language for each visitor.
See comparison →A new side-by-side comparison of Krymp and CloudConvert covering pricing, quotas, privacy and when to pick which — especially if you only need to compress or convert common everyday formats.
Read the comparison →The adjust panel now starts with an Effect strength slider that mixes the whole remix against the original — 0% leaves the original untouched, 100% gives you the full remix, and everything in between is a smooth blend. You no longer need to dial each macro down one by one to soften the look. Works the same on mobile and desktop, and is preserved in shared recipe links.
Open Remix →The "More adjustments" panel now has eight extra macros: Sharpness, Glow, Vignette, Glitch, Warmth, Fade, Distortion and Kaleidoscope. All are centered at 50% = no change, so you can safely dial up or down without breaking the preset's baseline. They work across every preset on both mobile and desktop.
Open Remix →Quick presets now include Social media (1080 px, perfect for feeds) and Print & photo (original size, top quality). Max width also has 1080 px and 480 px built in, plus a Custom field where you can type any width between 64 and 8192 px. The presets work identically on mobile and desktop.
Compress image →The inline slider and fullscreen view now have clear aria-labels, aria-valuetext read aloud as a sentence by screen readers, and Page Up/Page Down for larger jumps. The focus outline is stronger (4 px ring with background contrast), only shows on keyboard navigation and works consistently on mobile and desktop. The fullscreen view moves focus to the close button on open and back to the eye button on close.
Compress image →The eye button on each finished image now opens a fullscreen view where you can switch between Compare mode (drag the divider) and Zoom mode (pinch/scroll to zoom, drag to pan). Perfect for inspecting details, edges and compression artifacts before you download.
Compress image →The embedded before/after slider now shows the compressed file size and savings percentage in a compact header row above the slider — clearer and faster to read, especially on mobile where space is limited.
Compress image →On touch devices each inline slider now plays a short ~1.6 s sweep the first time it scrolls into view — so you instantly see the difference between original and compressed without having to drag. Respects prefers-reduced-motion and stops as soon as you touch the handle.
Compress image →Each finished image in the result list now has its own compact before/after slider embedded below the row — drag to instantly see the difference without opening the modal. It uses Pointer Events (single code path for mouse, touch and pen), a 44×44 px grip target, full keyboard control (arrows, Shift for larger steps, Home/End) and correct aria-slider attributes for screen readers.
Compress image →Seven new SEO/AEO/GEO landing pages funnel search traffic into /remix: /postereditor, /affischgenerator, /citatkort, /event-poster, /instagram-story-mall and /typografi-poster in Swedish, plus /en/poster-maker as the English twin. Each page ships a unique title/description, breadcrumb + HowTo + FAQ + SoftwareApplication JSON-LD, its own keyword set, hreflang and sitemap registration — all rendered SSR-friendly and linked to the Remix editor.
Poster maker →The media dropzone (used by /some and other tools) now has a dedicated touch variant: no drag-and-drop on iOS/Safari — where drag events never fire from fingers — but a large, tactile picker button that opens the system image/video picker plus an optional camera button. A targeted iOS Safari QA pass (390×844, hover:none / pointer:coarse) confirms that /bild, /some, /remix and /vattenstampel scroll cleanly without wedging and that touch mode renders correctly.
Failed files in a batch now have a clear "Retry" button right next to the error message — for images and video alike. You can rerun single files without disturbing the rest of the batch and without losing any of the completed results.
Compress image →The results list now shows a clear row at the top with the total saved size and the average compression percentage across the whole batch, so you immediately see how much you gained across all files combined.
Compress image →Duplicate export CTAs are gone — still exports (PNG/JPG/WebP) are handled solely by the global Export button in the header (and the mobile bar), while animated outputs (WebM/GIF/video) live behind a dedicated Animate popover. The format picker in the side panel now shows channel context (Instagram, Story, YouTube …) under each ratio so it's easier to pick the right surface.
Open Remix →The preview now stays pinned to the top while scrolling through the controls on mobile, so the impact of every change is always visible. The starter dialog shows the current size value and HEX code next to the slider and color swatch with bigger touch targets. The cookie banner returns to its default bottom-left position on /remix now that the mobile tab bar is gone.
Open Remix →The toolbar below the preview is simplified: Shuffle and Hold-for-original stay inline, while Change image, Start over, Batch, Share recipe and Copy as PNG are grouped in a "More" menu. Fewer buttons in the default view makes the primary action obvious.
Open Remix →The editor is now a simple single-column layout: a big preview on top with an always-on before/after slider (drag the handle), and all controls directly below in four clear tabs — Style, Adjust, Text & layers and Export. The old long accordion side panel and the separate mobile nav are gone. Advanced blocks live in collapsible "More controls", "Generative", "Performance" and "Templates" panels so the default view stays minimal.
Open Remix →The "Image" card on the landing now opens the file picker directly instead of just scrolling to the dropzone, matching the other starter cards. A4 and poster formats show a readable ratio label (A format, 2:3) instead of raw GCD numbers. The starter dialog now has a sticky footer so Create/Cancel are always visible. The editor side panel is grouped into four sections (Look & feel, Content, Fine-tuning, Layout & output) to make the long list navigable.
Open Remix →Format buttons in "Create with text" and "Image + text" used to render raw i18n keys. They now show the channel (Instagram, Story/Reels/TikTok, YouTube, print) on top with ratio and pixel size as a small helper line, making it clear what each preset is for.
Open Remix →Uploaded JPEG images are now normalized into a plain canvas before they are sent to WebGL. This prevents EXIF orientation and WebGL's Y flip from clashing, which could still make some phone photos appear upside down after the first fix. The same loader is used in the main flow, Image + text and batch export.
Open Remix →Dragging Intensity, Color or Grain in the Adjust panel felt like the effect vanished the moment you released the slider. The live preview was switching from draft resolution (640 px) to preview resolution (1600 px) on release, which made the image look much sharper and the effect feel gone. Still images now always render at preview quality, so what you see while dragging is exactly what you see after release. Video sources still pulse the draft tier for smooth live playback.
Open Remix →/remix now uses the same global Export button in the header as the rest of Krymp (same placement, popover and format picker as /moodboard). The popover shows PNG/JPG/WebP with the source resolution as an estimate and triggers the same export pipeline. Video sources export as WebM. The embedded export bar with WebM/GIF settings (duration, fps, quality) stays for advanced flows.
Open Remix →The /remix side panel had fourteen sections stacked in one long scroll (Style, Effect stack, Variations, Background, Text, Templates, Layers, Generative, Fine-tune, Format, Frame & overlay, Performance …) — finding the right control was hard. Grouped every section into an accordion with only Style expanded by default, with an icon and clear label per section. Works the same on mobile and desktop.
Open Remix →The WebGL renderer uploaded the source image without a Y flip, which put the top of the image at the bottom of the canvas. Enabled UNPACK_FLIP_Y_WEBGL on upload so preview and export now show the image right-side up.
Open Remix →Removed the dropdown caret from the Tools pill in the main nav since it doesn't indicate a working menu on narrow screens. Also moved the 'Upload video' button inside the Remix dropzone so it aligns with the image upload flow.
Open Remix →Replaced 'per-layer' with 'per layer' in the /remix hero paragraph and How it works steps, plus the related SEO description, FAQ and the /guider/skapa-poster-online guide. `hyphens: manual` prevents auto-hyphenation, but a real hyphen in 'per-layer' is still a line-break opportunity — removing the hyphen keeps the phrase together at 1024 px and 1280 px.
Open Remix →The starter dialog's image picker (Image + text) now blocks images below 64 px and shows a clear toast — previously a 4×4 PNG slipped through and made it look like the tool had frozen on a solid preset background. The editor toolbar collapses 'Change image', 'Start over', 'Batch', 'Share recipe' and 'Copy PNG' to icon-only below 1280 px so the row no longer wraps to four lines on narrower desktops — labels return at xl+. A small status chip now shows in the header while a video/WebM/GIF export is running (or fails) so users still see progress after closing the export popover.
Open Remix →The /remix hero heading and intro paragraph now use `hyphens: manual` so the fallback font no longer breaks 'poster editor' and 'per-layer' across lines before the webfont loads. The language switcher in PillNav (desktop + mobile pill) and the drawer menu now get distinct aria-labels — 'Switch language (desktop)', 'Switch language (mobil)' and 'Switch language (meny)' — so screen readers and tab order no longer see duplicate buttons.
Open Remix →LandingPage was moved to its own module so TanStack Router can lift it out of the main bundle — it no longer ships with every route. The recurring PillNav hydration warnings are silenced by marking its decorative spans as suppressHydrationWarning.
Starter cards (Image / Text / Blank / Image + text) now render during SSR — no empty boxes for the first second. Background preset thumbnails lazy-mount via IntersectionObserver so mobile no longer builds 50+ canvases up front. Mobile toolbar gets short text labels + aria-labels, the ⌘K shortcut is hidden on touch, tiny images show a clear error, and the cookie banner no longer covers toasts.
Open Remix →The history FBO powering motion trails and slit-scan now has a 16-frame cap, an 8 MB memory budget and an auto-tune that halves both frame count and resolution on mobile viewports (<768 px). When the budget is exceeded the ring shrinks until it fits, keeping the preview responsive while you drag sliders.
Open Remix →Star field, ASCII matrix and Leather emboss were rewritten for lower CPU pressure — small stars skip the halo gradient, the ASCII grid is cached per size, and the emboss hoists color parsing out of the pixel loop. GIF export now has its own status row inside the export popover with a frame counter, progress bar and error message. The new background types are also renamed so the kind and preset labels match on sv/en.
Open Remix →Remix gets a Ladybug-inspired effect pack: CMYK halftone with separated per-plate dot angles, Riso duo with registration offset, sketch cross-hatch, Kuwahara oil painting, stained-glass Voronoi, posterize, zoom blur, directional motion blur, chromatic tunnel and swimming pool caustics. Each ships with a ready-made preset across the Print, Editorial, Digital and Generative categories.
Open Remix →Compare mode in Remix gets a clearer handle, percent readout, double-click to snap back to 50 %, and keyboard control (←/→ fine-tune, Shift = 5 %, 0/5/9 for 0/50/100). The slide now works when the source is a video too — the original plays alongside the effect so you can drag back and forth without pausing.
Open Remix →Beyond WebM and MP4, Remix can now record the recipe as a palletized GIF that loops seamlessly. Great for social avatars, Slack emojis and shares where MP4 doesn't autoplay. Encoding happens fully in the browser via gifenc, with automatic downsampling to keep file size manageable.
Open Remix →The background panel gets three new kinds that shine as starting points from a blank canvas: Star glow with accent color and adjustable halo, ASCII mosaic with tunable cell size, and Emboss with depth-simulated relief. Each ships with presets in the strip and custom controls in the Custom tab.
Open Remix →Remix video export (both animated stack and video-to-video) now downloads as .mp4 in browsers that can record H.264 directly (Safari 14.1+, Chrome 126+) — otherwise we fall back to .webm as before. Two new generative effects have also landed: ASCII mosaic which turns the source into a text grid, and Emboss for a tactile relief. Both ship with ready-made presets in the Generative category.
Open Remix →The Remix macro panel now shows the three most important sliders large and grouped together, with the rest tucked behind a 'More adjustments' toggle so you can focus on what actually changes the image the most. A new 'Tasteful shuffle' button jitters macros within an editorial safe zone (35–75%) so the result always feels balanced. Presets can declare which sliders to feature via a new optional `featuredMacros` field.
Open Remix →Reeded glass, Pinch & swirl, Polar warp and Glitch are new effects in Remix — classics that shine on posters, LP covers and social graphics. Each effect ships with a ready-made preset in the Generative category and can be fine-tuned with sliders in the effect panel just like our existing distortions.
Open Remix →On Remix and other editor views (Moodboard, Storyboard, Mediakit, Fokus, PDF-to-Markdown) the cookie banner now sits in the bottom-right corner so buttons like 'Download PNG' and the mobile tab bar are never covered. The starter cards on the Remix landing also got dedicated aria-labels so screen readers announce title + description clearly instead of merging them.
The starter cards (Blank canvas / Text / Image + text) on the Remix landing now render client-side so an SSR/hydration mismatch can no longer swallow the click and block the dialog. The image flow also gets an 'Add text' button directly in the stack editor so the text, material, shape, icon, template and layer panels are reachable when starting from an uploaded JPG/PNG. On mobile we auto-switch to the 'Adjust' tab once a source is loaded so the panels don't hide behind the canvas tab.
Open Remix →Krymp Remix has been added to the tools menu (desktop + mobile), dropdown list, homepage popular tools and new tools list so it can be found from everywhere we list tools.
Open Remix →Dropping an image now creates a Remix document automatically, so shapes, icons and templates can be added directly from the stack editor. We also fixed a hydration mismatch in the performance picker, rewrote the landing hero and how-it-works to reflect the actual tool (poster editor, ~60 fonts, 20 materials, templates, SVG export), and removed a duplicated FAQ section.
Open Remix →A deep dive into building a poster with Krymp Remix — from template or blank canvas, through typography, shapes, icons and the material panel (chrome, gold, glass), to PNG or SVG export. Published in both Swedish and English with HowTo, FAQ and QAPage schema plus hreflang.
Read the guide →Metadata for /remix and /en/remix now mirrors the full editor — ~60 fonts, 20 materials, 200+ icons, shapes, templates, per-layer effects and SVG export — with refreshed titles, descriptions, HowTo steps and five FAQ entries in both languages. Three new AEO answers (SV+EN) in /fragor and /en/questions cover 'create a poster online for free in the browser', 'chrome/gold/glass text material' and 'export SVG from an online poster editor' and flow into llms.txt, llms-full.txt, llms-faq.json, aeo/index.json and ai-sitemap.xml automatically.
Open Krymp Remix →Remix turns into a real poster editor. Nine shape types — rectangle, rounded, circle, ellipse, polygon, star, line, arrow and blob — drop straight into the canvas. A curated icon library with 200+ searchable, categorized Lucide icons plus the option to upload your own SVG (auto-sanitized) covers most graphic needs. Twelve starter templates (quote card, event poster, product hero, social thumb, story, banner and more) load from the stack editor in a click. A new layer list surfaces the whole document with select/duplicate/reorder/delete, per-layer pixel effects (blur, contrast, saturation, hue, grayscale) and arrow-key nudging for the selected layer (Shift = coarse, Alt = faster). A bonus export ships the document as pure SVG when every layer is vector.
Open Krymp Remix →Text and graphic layers can now carry a real surface feel — matte, glossy, metal, chrome, gold, glass, acrylic, paper, fabric, velvet, wood, stone, porcelain, plastic, rubber, wax, carbon, neon, foil and holographic. The Material Engine applies lighting and material passes clipped to the layer alpha: tint, Lambert-inspired rim light, specular band, reflections for chrome/glass and procedural textures for paper/fabric/wood/carbon. A dedicated per-layer material panel in the stack editor delivers 12 curated presets, roughness/reflectivity/bumpiness and up to three light sources (azimuth, elevation, intensity, color) so you can dial in the scene exactly.
Open Krymp Remix →Text layers are now first-class in Remix. A curated Google Fonts catalog with ~60 families (sans, serif, display, slab, mono, handwriting) loads lazily and rasterizes directly in the right family. The typography panel in the stack editor covers fill (solid or gradient), stroke, shadow, glow, italic/uppercase and 12 typography presets — plus advanced effects: extruded shadow, double exposure, knockout and text-as-mask over the layer behind. Font search, category filters and per-family weight pickers make it fast to land the right voice without leaving the canvas.
Open Krymp Remix →Starting from Text or Blank canvas now lets you pick a real background right away — solid, linear/radial/conic gradient, mesh, aurora, noise, grain, paper, dots, grid or stripes — with thumbnails and a curated preset strip. You can fine-tune the background after starting via a dedicated panel in the stack editor (colors, angles, intensity, size), and the canvas re-rasterizes on the spot so you see the result before layering on effects.
Open Krymp Remix →When you open a layer in the stack editor you now get two small side-by-side thumbnails: the image Without the layer and With the layer — reflecting its current blend mode, opacity and parameter overrides. You can see each layer's contribution before committing, and quickly compare multiply against screen or nudge opacity and see the difference right there instead of scrolling up to the main canvas.
Open Krymp Remix →First visit on low-end devices (≤4 GB RAM, ≤4 cores, data saver on or slow mobile connection, plus small touch screens) now defaults to performance mode — high FPS out of the box, no dial to tweak. The video remix loop has a target FPS cap (30 in performance, 45 in balanced, 60 in quality) and pauses entirely when the tab is hidden, so battery and heat stay in check even with heavy generative effects like liquidDisplace and filmBurn.
Open Krymp Remix →Remix no longer needs an uploaded image to get started. Four entry points on the landing: Image, Text, Blank canvas, and Image + text. The text and blank flows come with a size picker (square, story, poster, A4 and more), font, size and color — baked into a canvas that opens right in the editor where the full effects library, export and recipe sharing all work as usual.
Open Krymp Remix →On mobile, a compact canvas preview now sticks to the top while you are on the adjust or styles tab — you can see the image change in real time without the panel covering it. Controls in the generative panel now wrap: label and number on their own row, slider full-width with a 44 px touch target, and the collapse/reset buttons are bumped to the same hit area.
Open Krymp Remix →Every effect layer now has a blend mode (normal, multiply, screen, overlay) and an opacity slider — same tools as Photoshop, right inside the remix. Two new generative effects: Liquid Displace (soft fluid distortion) and Film Burn (warm analog light leak), with matching Liquid Lens and Film Burn presets. On mobile, the adjust tab now mirrors desktop — stack editor, generative panel and performance mode — and touch targets on the performance picker are bumped to 44 px.
Open Krymp Remix →The batch view now runs a real queue: each image shows waiting/rendering/done/failed live, you can cancel mid-run without losing the already-finished files, and the ZIP always contains whatever has been rendered. Video export got a quality picker (low ~2 Mbit/s, medium ~6 Mbit/s, high ~12 Mbit/s) with a live toast showing percent + estimated time left. A new "Copy as PNG" button puts the current image straight on the clipboard so you can paste into mail, chat or Figma without a download detour.
Open Krymp Remix →New picker in the editor to balance sharpness against framerate: Performance downscales internal rendering (~65 %) for smoother FPS on weak GPUs/mobile, Balanced matches the quality tier's pixel cap 1:1 (default) and Quality supersamples (~140 %) for smoother edges (SSAA) on stronger devices. The setting affects only the live preview — export always renders at full source resolution — and is remembered between visits.
Open Krymp Remix →The renderer now isolates every effect layer: if a shader can't compile or fails mid-run, that pass is skipped, the user sees a clear toast ("Effect X doesn't work on your GPU") and the rest of the recipe keeps running — no black canvas or frozen video export. We also listen for webglcontextlost/restored: the preview shows an overlay when the context dies, repaints itself automatically once the browser restores the GPU context, and video recording is aborted with a message instead of producing an empty file. A one-shot verification runs on preset change so incompatible effects are flagged before the user hits download.
Open Krymp Remix →A new panel in the editor gathers every generative layer in the active preset (aurora, plasma, marble, ice, flow fields, kaleidoscope, color grade and more) and exposes the most common parameters with both a slider and a numeric input. Changes propagate through the pipeline in real time, are stored as per-layer overrides in the recipe and can be reset per effect. Classic effects keep going through the macro sliders so the panel stays focused.
Open Krymp Remix →The preset browser now has a search field that matches label, description, category and effect names — a search for "Generative" or "aurora" jumps straight to the right presets. Under the category chips there is a new smart-filter row with the most-used effect families (Bloom, Grain, Kaleidoscope, Color grade, …) and hit counts. New compare mode: mark up to four presets and open a side-by-side dialog with thumbnails, description, layer stack and a "Pick" button for a quick decision.
Open Krymp Remix →New preview panel in the editor: a hero tile with a split slider (original ↔ selected build step) and a horizontal row of thumbnails that add one effect layer at a time (Original → +Effect 1 → … → Full). Tap a tile to focus it and drag in the hero to compare against the original — before you hit Download.
Open Krymp Remix →Every effect in the layer stack can now be expanded and fine-tuned directly in the UI. Sliders for the most relevant parameters — blur radius, edge strength, wave displacement (amplitude, frequency, phase), color grade (temperature, saturation, contrast) and intensity across every generative effect — sit on top of the macro-modulated baseline. Overrides travel with the recipe link and can be reset per layer.
Open Krymp Remix →Massive expansion of the effect library. Refraction & light: Crystal Refraction, Glass Distortion, Prism Refraction, Aurora Flow, Aurora Dust, Plasma, Energy Field, Light Weaving, Signal Bloom, Fractal Bloom. Material & surface: Liquid Metal, Ceramic Glaze, Marble, Silk Surface, Fabric Noise, Shadow Fabric, Ice Cracks, Arctic Frost, Topographic Portrait, Living Gradient. Motion & particles: Neural Mesh, Ribbon Flow, Quantum Particles, Magnetic Dust, Flow Fields, Gravity Lines, Morph Field, Cloud Diffusion, Data Storm, Polar Wind. New Generative category in the preset browser with ten showcase presets — Crystal Veil, Plasma Poster, Aurora Editorial, Liquid Chrome, Ice Shatter, Topographic Editorial, Flow Mesh, Silk Noir, Data Storm and Marble Editorial.
Open Krymp Remix →New layers in the recipe: gaussian blur, unsharp sharpen, Sobel edge detect, radial chromatic aberration, sinusoidal wave warp, kaleidoscope and full color grading (temperature, tint, saturation, brightness, contrast). Six new presets — Tilt Shift Dream, Ink Outline, Anamorphic Glow, Liquid Ripple, Kaleido Bloom and Crisp Editorial — show how they combine.
Open Krymp Remix →Upload an MP4, WebM or MOV and apply preset, macros and effect layers live to the playback — everything renders in real time via WebGL2 in your browser. A new export button captures up to 15 seconds of the styled result as WebM. No video leaves your device.
Open Krymp Remix →Drop in multiple images and export a whole campaign as a ZIP with the same recipe. Five export formats with smart-crop (Original, 1:1, 4:5, 9:16, 16:9) for Instagram, Story/Reels, LinkedIn and YouTube. New overlay layer with headline text (3 fonts, 9-grid position, color, shadow), uploaded logo (4 corners, size and opacity) and 4 frame styles including polaroid.
Open Krymp Remix →Share your exact look with a link (preset + macros + layer order encoded in the URL, the image is never included). The receiver drops in their own image and the recipe applies instantly. A new animated export in the format picker records 2 s of WebM with subtle grain and shift motion.
Open Krymp Remix →Five new effects (Halftone, Bloom, Dither, Scanlines, Vignette) and six new presets (Halftone Press, Neon Bloom, Gameboy Dither, CRT Broadcast, Dream Haze, Posterize Heavy). The export button now has a format picker for PNG, JPG and WebP.
Open Krymp Remix →/remix now feels like a real editor: reorderable effect layers, six live variations to pick from, ⌘K palette for fast navigation, drag-slider before/after compare, and an upgraded grain engine with sharpness, color tint and shadow bias.
Open Krymp Remix →The preset catalog now has 12 styles (Duotone Signal, Riso Red, Newsprint, Pixel Plant, Chromatic, Monochrome Noir, Editorial Sepia, VHS Glitch, Posterize Lite, Print Cyan, Tokyo Neon, Fog Mint) with live thumbnails rendered on your own image. Five simple sliders (Intensity, Color, Contrast, Grain, Detail) reshape the entire style without exposing technical parameters. New split view on desktop, sticky bottom tabs on mobile, undo/redo (Cmd/Ctrl+Z).
Open Krymp Remix →The render engine now runs a dynamic effect stack via ping-pong passes. Two new effects (pixelate and RGB shift) plus a three-tier quality system: draft (~640 px) while interacting, preview (~1600 px) idle, and full source resolution on export. Tier switches are instant so you always see the sharpest image when you're not dragging.
Open Krymp Remix →Krymp Remix turns an ordinary image into a finished creative visual using WebGL2 effects on your device. Phase 1 ships with the "Duotone Signal" preset (duotone + grain + soft threshold), hold-to-compare and PNG export. More styles, macro sliders and variations are coming next.
Open Krymp Remix →When your images are done you now see a drag slider comparing the original and shrunk version right in the results — no modal click needed. Below it an auto-generated share card (−74% etc.) can be copied, downloaded, or sent through your device's share sheet.
Open image compression →The 404 page (and the root error boundary) now render in EN/DE/FR/ES instead of Swedish. The Brand kit AI (both QuickStart and Assist buttons) sends a matching system prompt for all five languages so the model replies in the user's language. The color chip on the home page is also localized.
PDF tool cards, the compare examples (MP4/JPG/PNG) and the "Open tool" CTA on the home page are now translated. Footer partner cards (Hasty, Fenny, Ellary) show French, German and Spanish taglines instead of falling back to Swedish/English.
About 6 200 UI strings translated to Spanish: nav, search, tool cards, benefits, tool pages and SEO copy. Remaining identical strings are intentional (brands, file formats, URL paths).
The Updated badge on tool pages, the footer's language label and partner taglines used to show Swedish on /de, /fr and /es. All three are now localized, or fall back to English instead of Swedish.
Nearly 6 300 UI strings translated to French: nav, search, tool cards, benefits, tool pages and SEO copy. Remaining strings that match English are intentional: brands, file formats, URL paths and cognates (Images, Original, PDF, etc.).
Every translatable UI string is now available in German. The remaining strings that match English are intentional: brand names (Krymp, Norrhavet Group), file formats (PDF, JPG, MP4, WebP …), URL paths, and German loanwords (Tools, Sitemap, Cookies, Feedback, Status, etc.).
Tool tiles and CTAs on /en, /de, /fr and /es now open the localized tool routes instead of the Swedish ones. A saved language choice also survives visits to Swedish-slug URLs — we redirect once to the matching localized twin.
Pages under /en, /de, /fr and /es now render in the right language on the very first server response — no Swedish flash before the nav, footer and landing pages catch up. The locale is derived from the URL before React renders and every language bundle is registered up front.
Previously /en, /de, /fr and /es rendered only a plain sitemap-style list or a stripped-down landing. They now mount the same full hero + compare + tool grid + manifesto + FAQ as the Swedish home — translated through the existing landing.* keys in each locale file. Each route also forces the i18n language to match the URL prefix so shared links land in the right language immediately. hreflang and JSON-LD (SoftwareApplication + HowTo + FAQ + Breadcrumb) are generated per locale.
PSI flagged missing Cache-Control on JS, CSS, fonts, videos and OG images (~3.7 MiB potential savings per repeat visit). We now set long-lived headers directly in the worker: hashed /assets/* and /fonts/*.woff2 are cached immutable for a year, /landing/*, /og/* and wasm modules get 30 days. Removed an unused Supabase preconnect on first paint so the browser skips a wasted connection.
/pdf-till-png ships with /de/pdf-in-png, /fr/pdf-en-png and /es/pdf-a-png. Render each PDF page as lossless PNG locally via WebAssembly — chosen DPI, preserved transparency, per-locale FAQ and full hreflang.
Open PDF to PNG →/pdf-konverterare ships with /de/pdf-konverter, /fr/convertisseur-pdf and /es/convertidor-pdf. Universal PDF ↔ JPG/PNG/Word/text conversion locally via WebAssembly — smart target formats per source, per-locale FAQ and full hreflang.
Open PDF converter →/word-till-pdf ships with /de/word-in-pdf, /fr/word-en-pdf and /es/word-a-pdf. Convert .docx locally via WebAssembly — layout, fonts, tables and images preserved, per-locale FAQ and full hreflang.
Open Word to PDF →/kopiera-text-fran-pdf ships with /de/text-aus-pdf-kopieren, /fr/copier-texte-pdf and /es/copiar-texto-pdf. One click copies PDF text straight to the clipboard — ready for Notion, Word or ChatGPT, per-locale FAQ and full hreflang.
Open text copy →/png-till-pdf ships with /de/png-in-pdf, /fr/png-en-pdf and /es/png-a-pdf. Combine multiple PNGs into a sharp PDF locally via WebAssembly — pick page size, margins and orientation, per-locale FAQ and full hreflang.
Open PNG to PDF →/pdf-till-text ships with /de/pdf-in-text, /fr/pdf-en-texte and /es/pdf-a-texto. Extract plain text from PDFs locally via WebAssembly — copy directly or download as .txt, per-locale FAQ and full hreflang.
Open PDF to text →/ta-bort-sidor-fran-pdf ships with /de/pdf-seiten-loeschen, /fr/supprimer-pages-pdf and /es/eliminar-paginas-pdf. Remove single pages or ranges locally via WebAssembly — remaining pages stay bit-exact, per-locale FAQ and full hreflang.
Open PDF page removal →/rotera-pdf ships with /de/pdf-drehen, /fr/faire-pivoter-pdf and /es/rotar-pdf. Lossless 90°/180°/270° rotation per page or full document — all local via WebAssembly, per-locale FAQ and full hreflang.
Open PDF rotation →/tecken ships with /de/zeichenzaehler, /fr/compteur-caracteres and /es/contador-caracteres. Live counting with platform presets (Tweet, meta description, Instagram, LinkedIn) and emoji-safe grapheme counting — per-locale FAQ and full hreflang.
Open character counter →/logo-grid ships with /de/logo-grid, /fr/grille-logos and /es/cuadricula-logos. Balanced client and partner logos with automatic size normalization — export to PNG/SVG/JPG, per-locale FAQ and full hreflang.
Open logo grid →/metadata ships with /de/bild-metadaten, /fr/metadonnees-image and /es/metadatos-imagen. Show and strip EXIF, GPS, IPTC and XMP from images — local in the browser, per-locale FAQ and full hreflang.
Open metadata →/signatur ships with /de/e-mail-signatur, /fr/signature-email and /es/firma-email. Outlook- and Gmail-compatible HTML signature with logo, colors and social icons — per-locale FAQ, breadcrumbs and full hreflang.
Open signature →/brandkit ships with /de/farben-extrahieren, /fr/extraire-couleurs and /es/extraer-colores. K-means palette with HEX/RGB/HSL, export to CSS variables, Tailwind, JSON and ASE — per-locale FAQ and full hreflang.
Open color extractor →/bilddelare ships with /de/bild-splitten, /fr/diviser-image and /es/dividir-imagen. Split images into grids for Instagram carousels, 3×3 grids or printable posters — per-locale FAQ, breadcrumbs and full hreflang.
Open image splitter →/audiogram ships with /de/audiogramm, /fr/audiogramme and /es/audiograma. Podcast clips to video with animated waveform and auto-captions (Whisper) — per-locale FAQ, breadcrumbs and full hreflang.
Open audiogram →/storyboard ships with /de/storyboard-erstellen, /fr/creer-storyboard and /es/crear-storyboard. Plan video scenes with image, camera motion and dialogue — per-locale FAQ, breadcrumbs and full hreflang.
Open storyboard →/maska-dokument ships with /de/dokument-maskieren, /fr/masquer-document and /es/enmascarar-documento. Real PDF redaction (text is removed, not just covered), metadata strip, per-locale FAQ and full hreflang.
Open mask document →/maska-bild ships with /de/bild-maskieren, /fr/masquer-image and /es/enmascarar-imagen. Pixelate, blur or cover faces and license plates — auto-detection, EXIF strip, per-locale FAQ and full hreflang.
Open mask image →/mediakit ships with /de/mediakit-erstellen, /fr/creer-mediakit and /es/crear-mediakit. Ready-made templates for creators, press and podcasts — case studies, per-channel numbers, per-locale FAQ and full hreflang.
Open media kit →/some ships with /de/social-media-formate, /fr/formats-reseaux-sociaux and /es/formatos-redes-sociales. One image or video → every channel (Instagram, TikTok, Facebook, LinkedIn, YouTube, Pinterest) with smart-crop, per-locale FAQ and full hreflang.
Open social formats →/komprimera-filer ships with /de/dateien-komprimieren, /fr/compresser-fichiers and /es/comprimir-archivos. Image, PDF, video and audio in the same session — local WASM compression, per-locale FAQ and full hreflang.
Open compressor →/video-thumbnail-generator ships with /de/video-thumbnail-generator, /fr/generateur-miniature-video and /es/generador-miniaturas-video. Frame picker, YouTube/Instagram/TikTok format presets, JPG/PNG export, per-locale FAQ and full hreflang.
Open thumbnail generator →/markdown ships with /de/markdown-editor, /fr/editeur-markdown and /es/editor-markdown. Live preview, export to .md/HTML/PDF, per-locale FAQ and full hreflang.
Open Markdown editor →/karta ships with /de/karten-poster, /fr/affiche-carte and /es/poster-mapa. Template-driven map posters with title, coordinates and GPX route — 300 DPI PDF, locally in the browser, per-locale FAQ and full hreflang.
Open map poster →/moodboard now has /de/moodboard-erstellen, /fr/creer-moodboard and /es/crear-moodboard. Auto-layout, extracted colour palette and PNG/PDF export — locally in the browser, per-locale FAQ and full hreflang.
Open moodboard →/ikoner ships with /de/icons-generator, /fr/generateur-icones and /es/generador-iconos. A full iOS/Android/PWA icon pack from a single source image — locally in the browser, per-locale FAQ and full hreflang.
Open icons generator →/klipp-pdf gains three new twins: /de/pdf-schneiden, /fr/decouper-pdf and /es/recortar-pdf. Cut away unwanted pages and keep only what matters — locally in the browser, per-locale FAQ and full hreflang.
Cut PDF →/ordna-om-pdf-sidor now has /de/pdf-seiten-neu-ordnen, /fr/reorganiser-pages-pdf and /es/reordenar-paginas-pdf. Drag-and-drop reorder, delete and rotate pages — all in the browser, per-locale FAQ and full hreflang.
Reorder PDF pages →/extrahera-sidor-fran-pdf now ships with /de/pdf-seiten-extrahieren, /fr/extraire-pages-pdf and /es/extraer-paginas-pdf. Pull any pages out of a PDF — locally in the browser, per-locale FAQ and full hreflang.
Extract PDF pages →/extrahera-text-fran-pdf gained three new twin pages: /de/text-aus-pdf-extrahieren, /fr/extraire-texte-pdf and /es/extraer-texto-pdf. Each explains how to copy text out of a PDF locally in the browser — UTF-8 safe, dedicated FAQ, full hreflang between locales.
Extract PDF text →Completed the /upscale cluster with French (/fr/agrandir-image) and Spanish (/es/ampliar-imagen) twin pages — German (/de/bild-vergroessern) already existed. WebGPU-based 2×/4× upscaling is now described in five languages with dedicated FAQ and full hreflang.
Open upscaler →Added localized twin pages for /beskriv: /de/bild-beschreiben, /fr/decrire-image and /es/describir-imagen. Vision AI for alt text, captions and SEO is now explained in the user's language with dedicated FAQ, HowTo steps and complete hreflang to the SV/EN parent.
Open image description →Added localized twin pages for /transkribera: /de/audio-transkribieren, /fr/transcrire-audio and /es/transcribir-audio. Whisper-based speech-to-text across 50+ languages is now described in the user's own language, with dedicated FAQ, HowTo steps and complete hreflang to the SV/EN parent.
Open transcription tool →Added localized twin pages for /vattenstampel: /de/wasserzeichen-hinzufuegen, /fr/ajouter-filigrane and /es/anadir-marca-agua. Each has its own FAQ, HowTo steps, related links and complete sv/en/de/fr/es hreflang pointing to the correct SV/EN parent.
Open watermark tool →The DE/FR/ES variants for resize/upscale/watermark/shrink previously pointed their hreflang at generic /bild and /pdf, which collided with the full twin clusters there. The 8 pages (de/bild-vergroessern, de/bild-verkleinern, de/pdf-verkleinern, de/pdf-wasserzeichen, fr/redimensionner-image, fr/reduire-pdf, es/redimensionar-imagen, es/reducir-pdf) are now bound to /upscale, /beskar-bild, /krymp-pdf and /vattenstampel-pdf with correct sv/en/de/fr/es hreflang.
The route files existed and rendered correctly, but only 11 of 62 locale tool pages were in the ROUTES catalog — the remaining 51 weren't discoverable via sitemap.xml. All 62 are now registered so Google can reach the full DE/FR/ES fleet directly via sitemap, complete with sv/en/de/fr/es hreflang on every row.
DE/FR/ES tool pages were listed in sitemap.xml but lacked the reverse link back to their SV/EN original — hreflang="sv" incorrectly pointed at the locale URL itself. The sitemap generator now resolves the SV parent via the translations map and emits correct sv/en/de/fr/es alternates on every row.
English tool pages (/en/*) now read the translations table and emit hreflang alternates to every sibling locale. Final piece of the reciprocal hreflang cluster — Google sees the same translation graph no matter which language the crawler lands on first.
Locale hubs /de, /fr and /es were acting as layouts and injecting their own canonical on top of the leaf pages — producing duplicate canonical tags on every tool page. The hubs are now index leaves, so each tool page has a single canonical. At the same time the hreflang cluster was completed: every DE/FR/ES leaf now reciprocally points at every sibling-language URL (sv/en/de/fr/es), letting Google cluster the full translation graph correctly.
Three new hub pages list every local tool in each language (22 DE, 20 FR, 20 ES) grouped into PDF, Images, Video/Audio and Other. Gives users switching language a clear entry point and gives crawlers a single deep link exposing the entire cluster — helping Google find and rank the full multilingual footprint faster.
Open German hub →Switching language from the menu now takes you straight to the equivalent page in that language — e.g. /pdf → /de/pdf-komprimieren, /fr/compresser-pdf or /es/comprimir-pdf — instead of just swapping the language on the same URL. Falls back to the language home when no translated twin exists, so you never land on a 404.
Krymp completes French + Spanish rollout: 13 new FR pages (reduire-pdf, jpg-en-pdf, diviser-pdf, signer-pdf, compresser-image/jpg/png/mp4, heic-en-jpg, redimensionner-image, flouter-visage, creer-gif, generateur-favicon) and 13 Spanish mirrors (reducir-pdf, jpg-a-pdf, dividir-pdf, firmar-pdf, comprimir-imagen/jpg/png/mp4, heic-a-jpg, redimensionar-imagen, pixelar-cara, crear-gif, generador-favicon). Now 20/20 FR + 20/20 ES tools live. Total multilingual footprint: 62 local pages (22 DE + 20 FR + 20 ES), ~2.4M monthly searches covered.
Open Diviser PDF →Krymp opens the Spanish-speaking market (Spain + LATAM): /es/generador-qr (246k monthly searches), /es/unir-pdf (246k), /es/comprimir-pdf (165k), /es/pdf-a-jpg (74k), /es/quitar-fondo (40.5k), /es/comprimir-video (27.1k) and /es/comprimir-audio (quick win). Full Spanish copy, HowTo + FAQ JSON-LD, hreflang for es/es-ES/es-MX/es-AR/es-CO/es-CL/es-419 ↔ sv/en. Total Spanish coverage ~798k monthly searches.
Open Unir PDF →Krymp opens the French-speaking market: /fr/fusionner-pdf (301k monthly searches), /fr/generateur-qr-code (201k), /fr/compresser-pdf (135k), /fr/pdf-en-jpg (60.5k), /fr/supprimer-arriere-plan (49.5k), /fr/compresser-video (18.1k) and /fr/compresser-audio (quick win). Full French copy, HowTo + FAQ JSON-LD, hreflang for fr/fr-FR/fr-BE/fr-CA/fr-CH ↔ sv/en. Total French coverage ~776k monthly searches.
Open Fusionner PDF →Krymp completes the German rollout: /de/qr-code-generator (165k monthly searches), /de/pdf-in-jpg and /de/jpg-in-pdf (2×18.1k), /de/pdf-verkleinern (22k), /de/gif-erstellen (9.9k), /de/mp4-komprimieren, /de/png-komprimieren, /de/bild-verkleinern, /de/bild-vergroessern, /de/pdf-wasserzeichen and /de/gesicht-verpixeln. The full DE_SLUGS map (22 pages, ~830k monthly searches total) now covers Krymp's core tools in German with local copy, HowTo/FAQ JSON-LD and full hreflang.
Open QR-Code Generator →Krymp doubles the German surface: /de/pdf-zusammenfuegen (301k monthly searches), /de/hintergrund-entfernen (60.5k), /de/video-komprimieren (14.8k), /de/pdf-teilen, /de/heic-in-jpg, /de/jpg-komprimieren, /de/pdf-unterschreiben and /de/favicon-generator. Each has full German copy, HowTo + FAQ JSON-LD, hreflang across sv/en/de and localized CTAs. 11 German pages live total.
Open PDF zusammenfügen →Krymp ships the first German pages: /de/pdf-komprimieren (135k monthly searches), /de/bild-komprimieren and /de/audio-komprimieren (low competition, quick win). Each has full German copy, HowTo + FAQ JSON-LD, hreflang across sv/en/de and localized CTAs. The full DE slug map (grounded in Semrush DE data) is ready for the next wave.
Open PDF komprimieren →Krymp now supports German, French and Spanish in addition to Swedish and English. The language switcher shows Deutsch/Français/Español, the hreflang generator and sitemap emit alternates for all five languages, and lazy-loading of translation bundles works across the board. Actual translations ship next — German first.
All four tools now render visible FAQ + HowTo sections below the tool, with JSON-LD (FAQPage + HowTo) mirroring the visible text — Perplexity, ChatGPT-search and Google SGE can lift answers directly from the page. FAQPage on /karta was reintroduced now that visible content backs the schema (previously a policy risk).
See map page →BingSiteAuth.xml is served from the root so Bing Webmaster Tools can verify krymp.io. sitemap.xml (429 URLs) was resubmitted to Google Search Console via API — most tool pages are still 'Discovered/Unknown', so trigger manual Request Indexing on the top 20 in the GSC UI for fastest pickup.
The 6 highest-traffic device & app guides now have English mirrors under /en/on/*: bild-outlook, bild-gmail, bild-android, pdf-outlook, video-imessage and video-whatsapp-desktop. Hreflang cluster between sv/en, dedicated JSON-LD (HowTo, FAQ, SoftwareApplication, Dataset) and sitemap coverage.
See all English device guides →Bild, Transkribera, Ikoner, Audiogram and Logo Grid now render .quick-answer on the intro plus a semantic <time> badge with the latest update date. Gives AI engines (ChatGPT/Perplexity/Gemini) and Google a clear freshness signal in the initial HTML — matches existing coverage on ToolIntro/UseCaseLanding pages.
Extends the device & app catalog with three high-volume long-tail queries: compress image for Teams (<1 MB for inline preview), PDF for Messenger (25 MB limit) and video for WhatsApp Desktop (2 GB but auto-compression over 16 MB). All with measurable platform limits in JSON-LD Dataset.
See all device guides →The freshness signal (semantic <time dateTime>) now renders directly under .quick-answer on /vad-ar/*, /jamfor/* and English landing pages — matching the pattern already used on tools, guides and /pa/*. AI crawlers (ChatGPT Search, Perplexity, Google AI Overviews) get the same date in initial HTML as in JSON-LD dateModified.
New programmatic-SEO wave targeting "compress X on iPhone/Android/iPad/Mac/Windows" and "compress X in Outlook/Gmail/Word/PowerPoint/Teams/iMessage". Each page names concrete platform limits (Outlook 20 MB, Gmail 25 MB, iMessage 100 MB, Teams 250 GB) — ideal citation fodder for AI Overviews and Perplexity. Index at /pa lists the full catalogue and the sitemap now includes them with hreflang.
Open index →The video sitemap (/video-sitemap.xml) now reads from src/lib/seo/video-demos.ts — the same catalogue the new <VideoObjectSchema> uses on tool pages. The sitemap now covers English routes (/en/compress-video, /en/guides/compress-video-for-email etc.) and emits contentUrl/duration/publication_date as soon as a demo clip lands in /public/demos/. HowToStep gained an optional image field, emitted as an ImageObject in JSON-LD and rendered next to each step in HowToSection — a Google requirement for HowTo rich-result eligibility.
All changes →New /press (SV+EN) with one-line pitch, quick facts, ready-to-use quotes and downloadable logos. Open /api/public/compress-stats (CORS-free JSON, CC-BY 4.0) for dev communities. oEmbed endpoint for rich previews on Reddit/Slack/Notion. New CitationBlock on /benchmark and /vad-ar/* with copy-paste templates for Wikipedia, APA, BibTeX and Chicago.
Open press kit →New .quick-answer blocks and a visible "Updated" badge on tool and guide pages help AI engines (ChatGPT Search, Perplexity, Google AI Overviews) surface fresh, fact-dense answers with clear attribution — not just indexable text. Six core AEO pages gained comparative facts with external sources (Netflix, Google web.dev, HTTP Archive, WhatsApp, Apple, caniuse).
Read the guides →The Krymp-till catalog now covers 35 more sizes across image, JPG, PNG, WebP, HEIC, PDF, video and audio (including "compress pdf to 4 mb", "cv under 1 mb", "passport photo under 200 kb", "video under 5 mb for mms"). Each page deep-links to the right tool with a preselected mode.
See all size pages →The mobile hamburger button now shows a visible "Tools" label next to the icon, making it immediately clear that the Tools menu and About page live behind it instead of a context-less icon.
The homepage hero search field now keeps the search icon inside the field layout instead of layering it over the input. This prevents the icon from briefly appearing and then disappearing on mobile.
Go to homepage →The homepage now carries a clearer, more action-oriented headline: "Fix the file. Keep working." — with a green accent on the second line to make it easy to scan and remember.
Go to homepage →The /krymp, /krymp-pdf and compress-files hubs now show a clear synonym note tying "krymp", "komprimera", "shrink" and "compress" together — so both searchers and AI engines land on the right tool. The hubs also link out to authoritative references (MDN, ISO, IETF, Mozilla) to strengthen credibility and topical relevance.
Go to compress files →On mobile, the hero section now only shows the two favorite tools, with a clear "Show more" button to expand the rest. This makes the homepage feel lighter and easier to scan on a small screen.
Go to homepage →A new page at /en/compress-files (and /komprimera-filer in Swedish) explains that 'krymp' means 'compress' and gathers every file-type tool in one place. HowTo, FAQ, comparison table and structured data (Speakable + FAQPage) help Google and AI Overviews rank us for 'compress files', 'shrink files' and related queries.
Open the hub →The H1 now states what the service actually does — 'Shrink images, video and PDFs' — so visitors and search engines understand Krymp at a glance.
The SoMe tool at /anpassa is back in sitemap.xml so Google and other crawlers can discover the page.
Open Anpassa →The mobile/tablet menu and top nav now have larger buttons, more padding and better spacing between links so they're easier to tap accurately.
The homepage quick search previously only looked through the popular tools and capped the results at 12 cards. It now searches the full tool catalog and shows every matching tool when you type.
A dedicated /en/convert-webp-to-jpg landing page with HowTo, FAQ and hreflang to the Swedish /konvertera/webp-till-jpg. Captures English search traffic for "webp to jpg" and points people to the local, in-browser converter.
WebP → JPG (EN) →The homepage H1 now describes the product directly: "Krymp — free in-browser file compression. Compress images, video, PDF and more." Better for humans and search engines alike. The language switcher on English landing pages now reads "Visa på svenska (Swedish)".
Go to homepage →The social share card at /api/public/social-cards/v1 can now be fetched as a true PNG via /api/public/social-cards/v1.png — same envelope, same window (window, dimension), rasterised server-side. Handy when platforms (Slack, Discord, email) don't render SVG.
The English PDF hub now reports page views to its own locale bucket instead of being counted as the Swedish /pdf view.
We QA'd and promoted 45 /en/ routes from noindex to ready — including /en/anonymize, /en/remove-background, /en/upscale-image, /en/transcribe, /en/qr, /en/favicon, the full PDF tool suite and 19 English guides. These pages now emit hreflang and can rank in English search.
The SoftwareApplication schema now points to the correct applicationCategory per tool (VideoApplication for video, BusinessApplication for PDF, DesignApplication for image, etc.) and exposes the screenshot field. Better signal to Google, ChatGPT and Perplexity about what each page actually is.
Decorative hero leaf illustrations are now deprioritised (fetchpriority=low) so the headline and quick-launch get bandwidth first. The below-the-fold section uses content-visibility: auto to skip rendering until the section scrolls into view — better INP and lower CPU load on mobile.
Press kit, copy-paste badges for linking to Krymp, open benchmark data (CC-BY 4.0) and machine-readable catalogs (llms.txt, ai-sitemap.xml, JSON catalogs) all on one page — makes it easier to cite Krymp in articles, comparisons and AI answers.
For developers & press →New /benchmark page compares Krymp with TinyPNG, iLoveIMG, Squoosh, Adobe Express and Compressor.io on the same 6 test files (photo JPG, PNG, WebP, PDF, video, HEIC). Reduction, SSIM and limits per row. Raw data as JSON and CSV under CC-BY 4.0 — free to cite.
See the benchmark →New use-case pages with the right presets per platform: images for Shopify & WordPress (skip the plugins), video for Instagram Reels & TikTok (correct spec on first try), PDF for the Swedish Tax Agency & job applications. Each with HowTo + FAQ schema for AI answers and Google.
Images for Shopify →New section catches 'compress X to Y kb/mb' searches: images ≤ 50 kB to 2 MB, PDFs ≤ 100 kB to 25 MB, video ≤ 8 MB for Discord, 16 MB for WhatsApp, 25 MB for Gmail. Every page deep-links to the right Krymp tool with the quality preset already selected.
See all size guides →Honest comparisons with tables and a verdict — when Krymp wins (batch, privacy, HEIC/AVIF) and when the others fit better (Squoosh for pixel-level comparison, Adobe for the design flow). Good citation material for AI engines.
Krymp vs Squoosh →New number-heavy answers for AI engines and Google: image quality vs file size (JPG Q levels with SSIM), how large a PDF you can email (Gmail/Outlook/iCloud limits), WebP vs JPG vs AVIF (encoding time + compression gain) and what actually happens during compression (lossless vs lossy). All measured on Krymp's own benchmarks.
Filled the last gaps in siteConfig: legalName ('Krymp c/o Norrhavet Group AB') and a public contact email. Google Knowledge Graph and LLMs can now tie the brand to a verified legal entity — a stronger foundation for the Knowledge Panel and clearer citations in AI answers.
Image, convert, resize and colors-PDF dropzones now show the small privacy badge right at upload so you can see files never leave your device. The batch progress bar got a per-file stepper that clearly shows waiting/processing/done on both desktop and mobile, and ZIP download works for every image compression tool — not just GIF.
New section on the security page states plainly that Krymp isn't certified — and concretely maps how the local architecture reduces risk on several controls (HIPAA §164.312, SOC 2 CC6/CC7, ISO 27001 A.8.2/A.13.2) and clarifies shared responsibility between Krymp and your organisation.
Read more at /sakerhet →After a successful run, tools can now show a subtle NextStepsPanel with 3 relevant Krymp tools for the same file type — e.g. image done → 'Remove background', 'Crop for social', 'Auto alt text'. Cross-tool discovery without being pushy.
A new hook (useDiscoveryNudge) fires ONE toast per tool after the 2nd successful run, with hard caps: max 1 per session and max 3 ever. Clicking 'Don't show' turns them off permanently.
New TryDemoFile component: a small button in the dropzone empty-state that fetches a public sample file and runs it through the tool so you see the result immediately. Lowers the first-time barrier.
The '🔒 Files never leave your device' pill is now clickable and takes you straight to /sakerhet where the full promise is explained — including the new HIPAA/SOC 2/ISO 27001 mapping.
See the security page →The '🔒 Files never leave your device' pill is now baked directly into the main dropzones (image+video tools and PDF tools) so the promise is visible the exact moment you drop a file.
A small new pill '🔒 Files never leave your device' can appear next to dropzones so you see the promise the moment you're about to drop a file — not only in the trust ribbon below the hero.
We now preload the hero font at the HTTP level so text appears earlier on first visit, and verified that every heavy library (PDF, DOCX, XLSX, image compression, video) only loads when you drop a file — not when you open the tool.
The trust ribbon under each hero shows two chips on mobile instead of stacking four, and the file lists for PDF, audio and GIF use smaller thumbnails so filenames and sizes fit without stretching. All four trust chips still appear on tablet and desktop.
The goal ring and Pomodoro pill are now compact — text no longer wraps, skip/reset move into the settings popover on small screens and the formatting toolbar takes less vertical space. More room for writing.
Focus writing →Previously "1080p" was always treated as height, so a portrait 2160×3840 clip shrank to ~607×1080 instead of 1080×1920. The limit now applies to the short side, so vertical clips keep their aspect.
Compress video →Cache headers are tuned: HTML always revalidates (so new deploys appear instantly) while hashed JS/CSS, fonts, OG images and landing media cache for a year so repeat visits start almost network-free. A new internal perf-trace script measures LCP, FCP, CLS, TTFB and long tasks per route between deploys.
The homepage hero illustrations now ship as responsive AVIF/WebP (~95% smaller transfer), Google Analytics loads on first interaction, and the hero animations run entirely on the GPU. PageSpeed LCP and TBT drop without any visual change.
Home →Ten new Q&A entries (Swedish and English) covering our latest tools, published at /fragor/<slug> and /en/questions/<slug> with QAPage schema, hreflang and sources. /ikoner also gains FAQ + BreadcrumbList schema and geo meta (SE, sv_SE) for better coverage in AI search, Google Overview and local results.
Q&A →The tools mega menu no longer scrolls internally — it widens so every tool fits at once.
'Fix the file.' is now shown in white while 'Keep working.' is highlighted in Krymp green (#94CE78), making the message clearer and more contrast-rich.
Home →The last word of the hero headline ('jobba.') is now also shown in Krymp green (#94CE78), together with the lead words on each line.
Home →The first word of each line in the home page hero headline is now shown in Krymp green (#94CE78), giving a clearer visual hierarchy against the dark background.
Home →The first word of the hero subtitle ('Free') is now shown in Krymp green (#94CE78) while the rest of the text is white.
Home →The main headline 'Fix the file. Keep working.' in the hero section is now shown in white for better contrast against the deep forest tone.
Home →Calmer hero with a deep forest tone, botanical illustrations in the bottom corners, a new headline ("Fix the file. Keep working.") and a glowing green outline around the search box.
Home →The contact section and site index on the About page were moved into i18n so English users see translated headings, body copy and link labels.
About & contact →'02 / Verktyg' is now shown as '02 / Tools' when English is selected. '01 / Live' has also been moved to the translation files.
'Visit hasty.se / fenny.se / ellary.se' in the footer's 'From our friends' section now follows the language selection.
Heading, category tabs and all shortcut links in the 'Popular shortcuts' section on the home page now appear in the correct language.
'100% in the browser', the saved ticker and the trust row (All local, No accounts, 0 KB uploaded) on the home page now appear in the correct language.
'Try it now', 'Drop zone' and 'Saved' in the 'Three motions. Done.' section on the home page now appear in the correct language.
The 'Free. And it's staying that way.' section on the home page is now fully translated to English.
The audience section on the home page is now fully translated to English.
The numbers section on the home page is now fully translated to English and only shows claims we can back up: tool count comes from the real nav list, average saving is clearly labelled as a benchmark, and the 'Right now' chip shows 'No files shrunk yet' instead of a made-up number when you have no local history.
Krymp can now send a browser notification when a longer job (e.g. video export or batch compression) finishes — but only if you've switched tabs. Enable it under Settings, or directly when a long job starts.
Open settings →The search box in the hero section now has higher contrast, a clearer border, a larger search icon, and a stronger shadow so it stands out better against the background image.
Go to homepage →The local Download entire list button is removed; exporting all icons is now handled via the global export button.
Open Krymp Icons →Animated hero showcase, ⌘K command palette, drag-to-select marquee, favorites & recents, larger calmer grid, hover stroke-draw effect, glass panels, gradient preview, group cards with icon previews, global size/stroke slider, export preview with folder structure, and an undo toast.
Open Krymp Icons →Click an icon to select it, click more to select several, click again to deselect. The hover checkbox is gone, the gradient preview renders live across the grid, and a 1-2-3 header (color → groups → select & download) makes the flow foolproof.
Open Krymp Icons →No presets — pick any color or gradient (from/to/angle) that applies to every icon. Build named color groups (e.g. White, Black, Sunset) and export a ZIP where each group becomes its own folder with the full selection.
Open Krymp Icons →Pick a color profile or gradient that applies to every icon at once. Multi-select icons and download them as a ZIP bundle — or grab the full filtered library in one click. All processed locally in the browser.
Open Krymp Icons →Search 1,900+ open Lucide icons and tune color, size, stroke, rotation and background right in the browser. Export SVG or PNG, or copy a ready-to-use JSX/Vue component. Everything runs locally — nothing is uploaded.
Open Krymp Icons →Step 3 now shows each format as a clear checkbox row. It's immediately obvious which formats are included in the batch and which format is active right now — especially clear during video export where formats are encoded one at a time.
Open SoMe →We swapped the frozen '0 of 5 formats done' view for a single status panel that shows what's happening right now, which format is encoding, and how much time is left. The bar now factors in live video progress so it keeps moving even while exporting a large 4K clip.
Open SoMe →We now pick batch size from the pre-downscaled frame instead of the source's 4K resolution. Result: seven formats on a 4K MOV run in ~2–3 ffmpeg passes instead of 7, so processing is ~2× faster and no longer looks stuck.
Open SoMe →We removed /logomotion and /en/logomotion from the site while we rethink the approach. Existing links now return 404. None of your locally saved projects ever left your device.
Three quiet upgrades to the video pipeline: batch size now adapts to source resolution (more formats per pass at 720p/1080p), sources much larger than the biggest output are pre-scaled once before the split so every branch processes fewer pixels, and a new "Faster video export" toggle switches to the ultrafast encoder for ~30 % faster runs at a small quality trade-off. The ETA calibrates automatically against the active mode.
Open Social →Every video in the Social flow now shows an estimated time remaining right after upload — and a combined "Done in X min" line at the top of the batch. The estimate calibrates locally on your device so it gets sharper the more files you run. Works the same on mobile and desktop.
Open Social →The JSON panel on /beskriv now lets you switch between pretty-printed and compact JSON. Both variants can be copied or downloaded.
Open Describe →The JSON tab on /beskriv now has live validation, syntax highlighting and word-wrap. Your selected tab (text or JSON) is remembered per image, and there's a clear 'Copy prompt' button in the text tab too.
Open Describe →Results on /beskriv now have a tab that lets you switch between ready-to-use text prompts and a structured JSON description.
Open Describe →On /beskriv there's now a JSON panel below each completed analysis — copy or download a structured .json with the description, prompts and full image analysis to plug into other tools or AI workflows.
Open Describe →The benefits row under the hero now uses h2 instead of h3 so the heading order follows h1 → h2 → h3 — better for screen readers and search engines.
Go to homepage →The search field now has more room for text on small screens so the placeholder shows without clipped characters, and the hero headline renders with real spaces between words both visually and for screen readers.
Go to homepage →We adjusted the homepage search so it runs full-width on small screens, has more consistent padding around text and icons, and no longer gets clipped edges when focused.
Go to homepage →On some Swedish pages (e.g. /transkribera) the FAQPage JSON-LD could ship in English when the i18n detector set English before SSR rendered the page. Language is now derived directly from the URL — /en/* is English, everything else is Swedish — so the schema always matches what AI and search engines see.
The pages /karta, /en/map, /krymp-index and /en/index-report shipped FAQPage schema with questions not visible on the page. Google requires FAQPage to mirror visible content — otherwise it risks manual actions. We've removed the schema from those pages. /en/audiogram now mirrors the same FAQ list shown in the app, so schema and content always match.
Open Krymp Index →The home headline animates word by word, but the words had no space characters between them — screen readers and AI crawlers read it as “Krympvägentillresultatetgratis”. There is now a real space between each word so the headline reads as “Krymp vägen till resultatet gratis”.
Open home →Several pages set the hreflang attribute with the wrong casing in React, which produced console warnings without affecting SEO. The casing is now fixed across all routes, so the logs are clean.
Pages like /bild-for-blocket, /bild-for-hemnet, /bild-for-instagram, /komprimera-for-email, /komprimera-for-whatsapp, /smallpdf-gratis, /tinypng-alternativ, /iloveimg-svenska, /fragor, /format-test-2026 and /en/questions were missing from the SEO registry and therefore absent from sitemap.xml. They're now included, sv↔en hreflang is consistent, and a new guardrail test catches the same regression if it reappears.
Open /fragor →The English /en/pdf-to-markdown page was built but missing from the SEO registry, which made hreflang point at an unregistered URL. It's now included in the sitemap and the sv↔en hreflang pair is consistent.
Open /en/pdf-to-markdown →We've added a typed foundation for SEO, AEO and generative discovery: a central SiteConfig, 10 Zod-validated registries (routes, redirects, sources, evidence, offers, locations, cases, briefs, keywords, flags), policy-guarded schema helpers that refuse to emit LocalBusiness or unsourced reviews, and 27 new guardrail tests. Full documentation under docs/seo/. No visible change for you — the goal is to prevent future changes from introducing broken canonicals, schemas or leaked noindex pages.
Your short links on /krymp are now shown, copied and shared as krymp.io/k/<code> instead of krmp.se/<code>. Existing krmp.se links keep working and 301-redirect to the new form, so nothing breaks.
Open Krymp link →We reserved /krymp on the short-link host, so krmp.se/krymp now 301-redirects to krymp.io/krymp instead of trying to resolve 'krymp' as a short code. Canonical and hreflang already pointed at krymp.io.
Open Krymp link →If something fails on /verktyg you now see a clear error banner with a Reload button instead of a blank screen. Client and server errors tagged [route-tree], [verktyg], [client] or [boundary] are automatically reported to /api/public/error-report (with rate-limit and dedupe). E2E tests for both domains and a dev cache header on routeTree.gen.ts prevent regressions.
Open tools catalog →Added a /verktyg smoke test, a developer-only console warning that immediately points at a stale routeTree.gen.ts, and clearer server-side logging ([route-tree]) when Vite transform errors surface. The short-link redirect logic is now unit-tested too.
Open tools catalog →Short-link handling now happens at the server host guard and /verktyg is reserved, so the tools catalog cannot be caught by the short-link route. Direct loads and internal navigation now land on the correct page.
Open tools catalog →Drop a PDF — we extract the text in your browser, infer headings and lists from the layout, and open the result in the Markdown Studio so you can polish, add frontmatter and export to .md, HTML, PDF or DOCX. Nothing leaves your device. For scanned PDFs we point you to /en/transcribe.
Open PDF → Markdown →/verktyg now has a sharper hero, short descriptions per area, premium icons and two new sections: 'Recently used' (updates automatically as you open tools) and 'New tools'. Search now shows match counts and highlights matching text in the list.
Open tools catalog →Clicking 'See all tools' now opens /verktyg — the entire toolbox in one list, grouped by area and searchable. The list and count update automatically when new tools are added. The homepage quick-start also has a new 'Popular' filter (Image, Video, SoMe, PDF, QR, Moodboard, Convert, Video Thumbnail, Markdown).
Open all tools →Right below the hero there's now a quick-start area. A large search field (⌘K on desktop) finds tools by name or function, filter chips let you switch between Image, PDF, Video & audio, Text, AI, Privacy and Other, and eight popular tools are surfaced as clear cards. Empty state offers a 'Show all tools' shortcut so you're never stuck. On mobile, chips scroll horizontally and the grid is two columns.
Open the home page →Three new control surfaces in the image tool: 1) A global error banner shows a clear reason and error code when the engine crashes (e.g. stale chunk after a deploy) and a 'Try again' button that performs a hard restart or page reload when needed. 2) A 'Cancel' button appears the moment a batch starts processing so you can stop right away and clear per-file status. 3) The queue is persisted locally (files, settings, status) so an automatic or manual reload no longer loses work — a 'Resume' button picks up where it left off. Per-file diagnostics (code + step) are visible in admin → System.
Open image compressor →Two final stability bugs are fixed: 1) After a new deploy the video and GIF tools could get stuck because the underlying engine error was translated before the queue could recognize it — the dead worker was never reset and every subsequent job failed until you manually reloaded the page. We now match on error type instead of on the translated string and silently reload when it's a stale-chunk error. 2) Image and video previews used a pattern that left old object URLs alive when you re-encoded the same file — over time that could eat several GB of memory. Each URL is now cleaned up as soon as its blob is replaced.
Three follow-up issues are fixed: 1) The image/PDF/watermark/background-removal queues got a per-file mutex so rapid retries or setting changes can't start two parallel runs on the same file — which is why occasional files used to get stuck on "processing" forever. 2) The format tool (multi-aspect) now has the same queue guard as the video tool so switching presets mid-run doesn't spawn double loops. 3) The PNG palette reduction at "High" is rewritten (sampling, typed arrays, quickselect, 5-bit LUT remap) — a 4K PNG that used to freeze the browser thread for 5–15 seconds now finishes in under a second.
Three root causes of the tab freezing are fixed: 1) Video, GIF, audio and format tools shared one ffmpeg engine with no queueing, so two parallel jobs could deadlock forever — heavy jobs now serialize through a shared queue. 2) The source video stayed pinned in browser memory across the entire multi-format export on /some, silently aborting large files — the buffer is now released immediately after the file is written to the engine. 3) Background removal on /bakgrund ran its AI model on the main thread and froze the tab for 5–30 s per image — it now runs in a Web Worker and the tab stays fully responsive.
Large videos (typically over ~250 MB) used to look like they were processing but never crossed 0 % — ffmpeg.wasm hits a silent memory abort when 16 outputs are encoded in parallel from one source. Oversized files are now rejected upfront with a clear toast that points to /video for pre-compression. We also encode at most two formats per ffmpeg pass (down from four) and reuse the dimension probe from upload, making multi-format export both more stable and faster.
Open SoMe →The /brandkit hero is now centered with the same title and subtitle sizing as the image, PDF and moodboard tools. TrustRibbon appears directly below the headline, consistent with every other tool. The secondary toolbar (Quick start, Share, Versions) has moved below the ribbon so the brand kit flow gets a clean, consistent top section.
Open brand kit →The /brandkit hero is now a single compact row with a much shorter subline. The 01–07 section nav is now plain chips without borders or numbers — only the active step gets a pill, completed steps get a small brand check, and a slim dot progress bar lives below. Quick start, Share style guide and Versions are now grouped behind a clean ”…” menu in the top right, giving the brand kit flow more room on both desktop and mobile.
Open brand kit →The /brandkit tool is now the Brand kit builder, with a headline that reflects what it actually does: logo, palette, typography, imagery, graphics, voice and messaging in one place. The section nav (01 Logo, 02 Colors, 03 Typography …) finally shows the right labels in Swedish and English, hides its scrollbar and has soft edge fades. Quick start, Share style guide and Versions are now grouped into a compact secondary toolbar so they no longer steal space from the brand kit flow itself.
Open brand kit →A new script (bun run seo:intent) clusters every public Swedish route by search intent (compress PDF, merge PDF, mask, video, ...) and writes audit/intent-map.json with a recommended primary per cluster. buildSeoTags now consults a central intent-canonical table (src/lib/seo/intent-canonical.ts) so variant routes can canonical-point at their primary while staying in the navigation. The table ships empty — each cluster is enabled manually after review, and the 301 list will follow.
View changelog →English pages that aren't fully translated are now noindex by default via a central manifest (src/lib/i18n/route-manifest.ts) — we never point Google at half-finished copy. Four top routes are already promoted: /en/pdf, /en/shrink-pdf, /en/compress-image, /en/compress-video. Three new CI scripts ship alongside: locale coverage (bun scripts/check-locale-coverage.ts), a tech-SEO crawler that validates canonical, hreflang and JSON-LD against the live URL, and a content-QA gate that fails the build if a route hard-codes a privacy claim. Finally a small event endpoint (/api/public/events) accepts tool_start/tool_complete/repeat_use without logging IP, cookies or filenames.
View changelog →The public /api/public/trends/v1/summary endpoint now has a formal Zod schema as the source of truth. Every response is validated before it ships, small buckets (below 5 events) are stripped in a second defensive pass, and responses include an X-Contract-Version header so integrators can track format changes. The /api page surfaces the contract version, dataset license and threshold — the same constants the code uses.
Read the API docs →The /trender, /var-paverkan and /status pages are now server-rendered with real numbers on first paint — no more 'Loading…' flash or empty skeleton before the data lands. The trends aggregation lives in a shared server helper so both /trender and the public /api/public/trends/v1/summary endpoint always count the same way.
See trends →The tools section on the home page now has the same search field and category chips (Images, PDF, Video & audio, Text, AI…) as the mobile menu — type or pick a category to filter the grid instantly. The desktop mega menu keeps the chips visible while you search, so free text and category can be combined. The footer index also gets a clearer 'Changelog' label instead of 'What's new'.
Try the filter →We removed the shortcut to /admin/login from the footer on every public page — admins reach the sign-in directly via URL. We also fixed the CO₂ box in the Krymp Index that previously showed 0 kg whenever savings were below half a kilo: small values now render in grams and mid-range values with one decimal. In the background we added the foundation for a central tool capability registry (LOCAL/HYBRID/CLOUD) and a claim registry, so privacy and statistics statements will come from a single source of truth instead of being hardcoded in every template.
Read more →Bundle analysis showed that the changelog database (~460 kB) and several large SEO data files (definitions, comparisons, use-cases, AEO answers) were ending up in the main chunk every page loaded — even though only a handful of routes use them. We moved the changelog page component into a separate module that the router code-splits, and converted top-level imports in the sitemap, llms.txt, RSS/Atom/JSON feed and AI-sitemap routes into dynamic imports inside the handler body. Result: the main chunk is ~0.7 MB smaller and Time-to-Interactive improves measurably, especially on mobile and 4G.
Audiogram, Map, Map themes and shared brand-kits used to sit in the main bundle because their components were exported from the route files, which disabled TanStack's automatic code-splitting. We moved those components into their own modules so the home page and other tools no longer load that code unnecessarily — noticeably faster first paint for everyone except visitors landing directly on those tools.
We found a barrel-import (import * as) of the entire Lucide icon package in the brand-kit tool that dragged ~540 kB extra JS into the global icons chunk every page loaded. Switched to tree-shakable named imports — the vendor-icons chunk shrank from 591 kB to 75 kB. We also stopped the landing page from auto-preloading three heavy routes; now it only preloads /bild, and skips that entirely when Save-Data is on or the network reports 2G. Measurable Time-to-Interactive win on both mobile and desktop, especially on slow connections.
Previously 49 Swedish pages were flagged enPathPending (canonical EN slug mapped, but the page not built yet) — sitemap and hreflang skipped them. Now each one has a real English landing: all compression catalogues (/en/compress/image, jpg, png, pdf, mp4, mov), tools (character-counter, markdown, social-media, mask-document/lawyers), brand/about pages (about, methodology, editorial-policy, our-impact, feedback, trends, api, press, changelog, compare, what-is), the full guide series (/en/guides + 19 how-tos) and the legal pages (cookies, privacy, security, accessibility, integrity, sitemap). Each page ships its own <title>, meta description, canonical, hreflang sv/en/x-default, SoftwareApplication / HowTo / FAQ / Breadcrumb JSON-LD, and a link back to the Swedish original.
Browse the English pages →Three concrete fixes on /transkribera: 1) Words from ElevenLabs Scribe are now joined with spaces — previously the whole transcript came out as one long word ("BitcoinandtheentireCrypto…"). 2) SRT/VTT/TXT/Markdown/CSV exports now ship with a UTF-8 BOM and `charset=utf-8`, so Swedish characters no longer turn into mojibake ("Längd" → "Längd") in Notes, Word, Apple Files etc. 3) The trust ribbon under the hero now states the truth: the file is uploaded to the transcription AI and deleted right after — no more "Local in your browser".
Open Transcribe →The routes catalog (src/lib/seo/routes.ts) now declares an English twin for every public Swedish page — tools, guides, about pages and legal pages included. 49 routes previously lacked an EN slug, leaving the hreflang chain uneven. Routes where the English page isn't built yet are tagged `enPathPending` so the sitemap and hreflang never point at a 404 — when the EN page ships we just clear the flag.
On iOS Safari the file picker didn't open when tapping the upload zone on /transkribera — display:none file inputs don't always click through on iOS. The input is now visually hidden but kept in layout, so a tap opens Photos/Files as expected and you can pick audio or video straight from your phone.
Open Transcribe →New in Logo Grid: an Auto-position button that snaps symbol size to 1.0 / 1.25 / 1.5 / 2.0× cap height and gaps to quarter-cap steps — classic lockup practice without doing the math. And the export menu now offers 'ZIP (all variants)' which packs all six lockup variants (horizontal, symbol-right, vertical, centered, symbol-only, wordmark-only) on both light and dark backgrounds as SVG + PNG in one bundle, with a README documenting fonts and colors.
Open Logo Grid →Two bugs that hit the upscale tool are fixed. The local AI engine now automatically falls back to WASM when your browser lacks a working WebGPU backend (typical on older Safari/iOS) instead of failing. The cloud engine can also read AI images returned as a link — not just inline data — so more upscales complete without the 'Could not decode upscaled image' error.
Open Upscale →You can now upload your wordmark as a file (SVG/PNG/PDF) alongside the symbol — or keep typing it as text. The tagline shows in every layout, not just the vertical ones. A new variant matrix shows all six canonical lockups at once (horizontal, symbol-right, vertical, centered, symbol-only, wordmark-only) — click any cell to focus and fine-tune. The export menu now offers a full grid sheet as PNG or A3 PDF with all variants, gridlines and clear space drawn in — ready to hand off to print or share with your team.
Open Logo Grid →A new Auto-balance button analyzes the symbol's optical center of mass and tunes the nudge for you. The background strip now shows WCAG contrast as a colored dot on every variant so you can spot invisible symbols or text at a glance. Below the preview you'll find the legibility floor ("holds down to ~24 px") and a new fullscreen button (or the F key) maximizes the lockup for a clean final review.
Open Logo Grid →The symbol is now auto-tinted to match your wordmark (or brand/custom color) so it stays visible on light, dark and brand backgrounds. You can also drop in SVG, PNG, JPG, WebP, GIF, BMP, AVIF, HEIC, TIFF, PDF, AI and EPS — the first page is rendered automatically. A small background strip shows the lockup on all three backgrounds at once so missing contrast is impossible to miss.
Open Logo Grid →Logo Grid helps you build a balanced logo lockup without opening Figma. Drop in your symbol, type your wordmark and optional tagline, pick one of six lockup layouts, and tune with sliders in cap-height units — not pixels. It overlays baseline, x-height, cap-height and clear space, offers an optical nudge that lifts the symbol ~2% (almost always reads better), stress-tests the lockup at multiple sizes and backgrounds, and exports to SVG or PNG. Linked to the brand kit both ways.
Open Logo Grid →The install chip in the header now captures the browser install prompt the moment the page loads (it used to miss it and the button did nothing). On iPhone/iPad, Safari on Mac and Firefox we instead show a short platform guide explaining exactly how to add Krymp to the Home Screen or Dock. A clear notice also appears if you try to install from an embedded preview where the browser blocks the dialog.
The auto mode in /thumbnail-generator used to spin up a fresh <video> element per frame, which stalled the browser even on small batches. It now reuses a single decoder and shows a live counter (e.g. 42 / 200) plus a progress bar while frames are captured. The count is capped at 200 to keep the tab from running out of memory.
Open the tool →Increased opacity on secondary text in the savings badge and the bottom trust row so it meets WCAG contrast on both desktop and mobile.
Replaced 71 hardcoded Swedish strings with t() calls in ExportPanel, RestorePanel, BatchPanel, MatchList, Overview and AllowlistPanel. Added 63 new keys under anonymisera.ui.* across sv.json and en.json — toasts, labels, risk levels and disclaimers now follow the language switch.
View EN twin →Replaced ~125 hardcoded Swedish strings with t() calls across SignatureForm, SignaturePreview, SavedSignaturesList, SignatureScore, InstallGuide, QuickStart, InstallDropdown, AudioSettings, AudioFileList, BrandKitSections and the /brandkit export block. UI and toasts now follow the language switch on the /en twins without leaking Swedish.
View EN twin →Added English landing pages (/en/shrink, /en/colors, /en/transcribe, /en/watermark, /en/anonymize, /en/email-signature, /en/qr, /en/compress-audio, /en/gif, /en/describe-image, /en/favicon, /en/image-metadata) with bilateral hreflang back to the Swedish tools. Closes the biggest gap in the SEO audit and lets Google index the EN variant correctly.
View EN twin →Filled in English FAQ items for 11 tools (pdf, video, background, audio, describe, favicon, metadata, some, signature, charcount, anonymize) so en.json now matches sv.json. Produces richer FAQPage structured data on every EN twin.
Search engines now get real titles, descriptions and keywords for the brand kit tool instead of the generic Krymp fallback. Also added an AEO answer entry for 'How do I create a brand kit online for free?' that covers start paths, mockups, AI suggestions, templates, versions and every new export format (CSS, Tailwind, Figma Tokens, iOS, Android) along with the public styleguide at /brand-kit/{slug}.
Open brand kit →A new 'Versions' button on /brandkit lets you save named snapshots of your brand kit (palette, typography, voice, messaging) and restore them anytime — stored locally in the browser, no login required. The Quick Start dialog also gets a fourth 'Template' tab with twelve ready-made brand kits (tech startup, bakery, law firm, skate, health, restaurant, consultancy, fashion, construction, edtech, B2B SaaS, local craft) that fill in the entire kit in one click.
Open brand kit →The ZIP export now ships W3C Design Tokens (Figma Tokens Studio), iOS Colors.xcassets + Typography.swift and Android colors.xml + typography.xml in addition to the existing CSS and Tailwind tokens. A new 'Share style guide as a link' button on /brandkit generates a public, read-only style guide at /brand-kit/{slug} (sv + /en/brand-kit/{slug} with hreflang) you can send to a client, developer or designer without them needing an account.
Open brand kit →Each section gets a sparkles button in the header: Messaging generates taglines, short descriptions, value props, mottos and key phrases; Voice generates tone words, headline and body examples plus avoid-words that match the sliders; Imagery suggests descriptions and style tags. Each suggestion has a 'Use' button that fills the right field. Powered by Lovable AI in Swedish or English depending on the active language.
Open brand kit →See your palette, typography and logo applied to six real surfaces: website hero, business card, Instagram feed, mobile app, invoice and merch/T-shirt. Toggle light/dark and desktop/mobile, page between mockups and screenshot any mockup as a PNG to share on LinkedIn, in a proposal or with a client.
Open brand kit →A new 'Quick start' flow with three on-ramps: paste a URL and we extract title, description, colors, fonts and favicon from the site — upload a logo and we extract the palette automatically — or describe your brand in one sentence and let AI suggest name, tagline, palette, fonts and tone words. The dialog auto-opens the first time you visit the tool with an empty kit.
Open brand kit →The /brandkit export now leads with Brand kit as the primary format. A ZIP bundles the style guide PDF (extended with logo, voice and messaging pages), original logo files, moodboard images and tokens in CSS, Tailwind v4 and JSON. Logo previews now render correctly on every background and Graphic elements shows live icons from the chosen library.
Open brand kit →The colors tool has grown into a complete brand kit in seven sections. Upload logos with roles and rules, pick font weights and headline templates, build a moodboard, choose icon library and patterns, dial in tone of voice with sliders and words, and write tagline, value proposition and boilerplate. A sticky section nav follows you through the flow on both desktop and mobile.
Open brand kit →The colors tool (Brand kit & colors) has moved up to the primary tools list — both in the desktop dropdown and the mobile menu — so it sits side by side with Images and PDF.
Open brand kit →The toolbar above the editor (Versions, Send via email, Change template, Import from web, AI assistant, Reset) has been rebuilt — glass surface, larger touch targets, clearer current template, pulsing badge on Versions, AI assistant as a hero CTA with a shine sweep and Reset as a quiet destructive icon button.
Open mediakit →You can now pick "Style guide" directly in the global export button on /brandkit. The PDF layout has been reworked: proper section typography, consistent line height for the type specimen, info cards for the fonts, tidier contrast tiles and token blocks with backgrounds — no more overlapping text.
Open colors →The web import has friendlier waiting copy ("Still working on it" instead of "This site is slow"). "New suggestions" now always produces fresh harmonies — even when every role is covered — and you can switch between Auto, Harmony and Neutrals to see analogous, complement, triad or monochrome additions. The contrast section now shows recommended pairs for headline / body / CTA and an interactive WCAG popover. A new typography step lets you pick primary and secondary typefaces from Google Fonts with a live preview. And a new "Style guide (PDF)" button exports colors, typography, contrast and tokens in one document.
Open colors →New "Clear all" button in the media kit toolbar — wipes the whole kit (with a confirm) and lets you pick a template again. The live preview has also been rebuilt to mirror the exported PDF exactly: brand spine on the cover, eyebrow labels above section titles, accent rules and facts as cards. When the kit is empty you now get a clear prompt to start filling in instead of a blank A4 — add info first and the preview takes shape.
Open media kit →Starter templates no longer pop up as a dialog — they're embedded directly in the editor so you can start without dismissing anything. Importing from a website after picking a template now overwrites the template's text and URL with real info from the site. The PDF has a refreshed look too: calmer section headers with an eyebrow label and accent rule, a smarter cover with date/location/web meta strip, and facts as clean cards.
Open media kit →New AI button in the media kit editor. Enter your company name (plus optional industry, audience and extra context) and the assistant proposes a tagline, short and long description, a press-ready boilerplate and 3–5 journalist-friendly facts. You preview every suggestion and tick which ones to apply; facts are appended and text fields are replaced. Powered by Lovable AI with zero setup.
Try the AI assistant →Two new quick starts in the media kit tool. "Import from web" pulls company name, description, brand color (theme-color) and favicon from a URL and fills out the kit in seconds — just paste the address. We also shipped /en/mediakit/examples — a gallery with four ready examples (SaaS startup, e-commerce/D2C, consultancy and creator/podcast) you can open in the editor with one click. Everything runs locally in the browser; nothing is stored on the server.
See examples →Four new PDF themes in the media kit tool — Minimal, Editorial (serif headings, purple), Bold (orange accent) and Dark (dark background, cyan). A "Versions" dialog stores up to 10 named local snapshots you can restore in one click. The last PDF page now gets an automatic QR code to your company's website. A new "For developers" section generates a JSON-LD Organization block (schema.org) you can copy into your press room's <head> for better SEO. New "Send via email" button opens your default mail client with subject and boilerplate pre-filled.
Open media kit →Five new content modules in the media kit tool: a cover image as the hero on the first PDF page, press mentions with quote and link, a team section with portraits and bio, awards, and a timeline of milestones. Every section has reordering (up/down), empty states, and shows up both in the live preview and the exported PDF — press mentions as a quote list, team in a grid, awards as a list, and the timeline with a time axis.
Open media kit →Two new sections in the media kit tool: Brand (color palette with color picker, heading and body fonts, plus a boilerplate paragraph up to 600 chars) and Social channels (Instagram, LinkedIn, X, YouTube, TikTok, Spotify, Apple Podcasts and more). Everything shows up both in the live preview and the exported PDF — color swatches on a dedicated brand page, the boilerplate on the about page, and socials on the press contact page. The completion meter's "Boilerplate" and "Brand color" now jump to the new brand section.
Open media kit →The media kit tool now has four starter templates (SaaS, E-commerce/D2C, Consultancy/agency, Creator/podcast) with ready-made copy, key facts, brand colors and social handles — customize instead of starting from a blank form. A completion meter shows how close the kit is to done with jump-links to missing sections, and a "Switch template" button is always in the editor. The data model is extended in advance for upcoming stages (color palette, typography, team, awards, timeline, PDF themes) with a safe migrator for existing local kits.
Open media kit →The desktop Tools menu now has the same category chips as the mobile menu (All, Image, PDF, Video & audio, Text, AI, Security, Other) so you can quickly narrow down to the right kind of tool. Popular list refreshed: Moodboard, Markdown Studio and Video thumbnails are now in, and Shrink link has moved to All tools where it belongs.
Fixed that a new scene appeared to 'vanish' when a filter was active — filters now clear automatically on Add so the new scene is always visible, with a short toast explaining it. The list view now stretches to the full width of /storyboard so every column (scene, tag, description, type, camera, lens, movement, location, length, status) is visible on desktop without horizontal scroll. Color coding now always shows a visible dot even when location/day is empty (per-scene fallback) so it's clear that Location/Day actually works. Save sketch now shows a confirmation toast and the button no longer disables unnecessarily. The intro text above the tool has been removed for a calmer view.
Open Storyboard →PDF and ZIP exports from /storyboard are now unmistakably Krymp documents. Every page carries a KRYMP wordmark with the brand-green puck, the scope label, and a thin brand-coloured rule in the header. The footer shows krymp.io/storyboard, the project title and pagination. The cover gets a larger wordmark, scope title, project stats and a QR code that leads back to the tool. Shooting script pills and storyboard grid labels are brand green, the tech card has a thin brand top-line, and the call sheet has a small QR per shooting day — handy on set. PDF metadata (title, author, creator) is set properly so Finder/Adobe/Drive show the right info. The ZIP now ships a markdown README.md and a krymp-link.url shortcut back to the tool. Filenames follow the krymp-{format}-{project}-{date}.{ext} pattern so multiple projects sort cleanly in Downloads.
Open Storyboard →/storyboard got a big flow polish. Keyboard shortcuts (N new scene, D duplicate, Enter to open, ⇧↑/↓ to move the scene, / focuses search, ? shows help) let you work entirely mouse-free on desktop. A new search + filter bar above the list: filter by status (Planned/Shot/Approved), location or day, or search freely across title, description and location. The runtime summary now shows both total runtime and a per-shooting-day breakdown — clicking a day chip filters the list. Color-coding by location or day gives visual grouping in both grid and list. Empty projects get three clear CTAs (import images, use a template, blank scene) and the project header has a subtle 'Saved ✓ 2:32 PM' indicator.
Open Storyboard →Big flow polish on /storyboard. Arrows are gone — sort scenes via drag and drop in both the grid and the table, works equally well on desktop and mobile. Paste several images with Cmd/Ctrl+V or drop a batch on the page — each image becomes its own scene. Scene and shot numbers are filled in automatically when you add a scene, and a new 'Renumber' button gives the whole list a clean start. Every PDF export (shooting script, storyboard, shot list, call sheet) now opens a preview with the first scenes before download — so you see the layout before blindly exporting a whole film.
Open Storyboard →Big update to /storyboard. Boards and shot list are now merged — a scene has image, description AND technical shot fields in one row. New export menu: Shooting script PDF (one scene per page, image left / notes right), Call sheet (grouped by day and location), the Storyboard grid and Shot list table as before, plus a ZIP with all four PDFs + voiceovers + project.json. Mobile on-set mode with swipe (left = shot, right = skip), sticky progress and 'up next' list. Four project templates (music video, 15s ad, 3-act short, interview + B-roll), per-scene voiceover recording, live runtime counter in the header, local snapshots and a QR code to krymp.io/storyboard on the cover page. Still runs entirely in your browser.
Open Storyboard →Krymp Storyboard & shot list is a free tool for filmmakers. Add frames, drop in images or sketch with the pen, write scene, shot number and camera notes — and build a shot list with camera, lens, movement, audio, location, duration and status. Both modules can be toggled independently. Export a combined PDF (cover + storyboard + table shot list), separate PDFs, the shot list as CSV or a ZIP with everything. Runs locally in your browser with autosave to IndexedDB — no uploads, no account.
Open Storyboard →Krymp Media kit gathers logos, press images, products/services, company description, key facts and a press contact — and exports everything as a polished PDF and a ready-to-send ZIP with originals, credits, boilerplate and vCard. Drag-and-drop logos (with auto-role detection: primary, dark, light, monochrome, favicon), live PDF preview while you type, and a single export button with PDF / ZIP / both. The whole tool runs locally in your browser with autosave in IndexedDB — no uploads, no account.
Open media kit →PDF export failed with "unsupported color function 'lab'" because the old renderer can't read modern CSS colors like oklch and lab. We switched to an updated renderer that supports them, so MD, TXT, PDF and DOCX download instantly again.
Open Fokus →The page now shows title, lead and trust ribbon immediately while the heavy editor stack (Tiptap + markdown) streams in. PDF, DOCX, MD and TXT export engines are only fetched when you actually export, making /fokus feel noticeably faster to open.
Open Fokus →The benefits bar under the hero is rebuilt from scratch. The cards are gone — instead six evenly distributed items show icon, title and a short caption separated by hairline dividers. The result feels calmer, more balanced and equally polished on mobile, tablet and desktop.
Go to home →The writing surface is rebuilt as a true WYSIWYG editor: headings grow, bold looks bold and lists get bullets live as you type. The character you type always sits under the line above, so the caret can no longer drift when you format. Normal and fullscreen modes now share the exact same typography and behaviour.
Open Fokus →The cards under the hero now have uniform height and a fixed grid — 2 columns on mobile, 3 on tablet and 6 on desktop — so titles no longer wrap into different line counts and the row feels balanced on every device.
Open homepage →Headings, bold, italic, lists and quotes now render live in the editor as you format, the scrollbar is hidden so the full white area is writable, and when the Pomodoro timer hits zero you get a prominent notice with a two-note chime — the timer now also keeps ticking even if you're just thinking and not typing.
Open Fokus →Changing focus minutes, break lengths or sprints-until-long-break on /fokus now updates the countdown in the pill immediately — previously it stayed on 25:00 until you reset.
Open Fokus →You can now edit the document title in fullscreen mode on /fokus — the field sits discreetly bottom-right and fades with the rest of the UI. The trust row pills ("Local in your browser" etc.) are now hidden in zen so nothing distracts from writing.
Open Fokus →The writing surface on /fokus now uses the full card width with a taller minimum height, so you get more room to write right away. Goal and Pomodoro pills now share the same height for a cleaner row.
Open Fokus →Fullscreen on /fokus now feels like a dedicated writing surface. Site header, footer and other chrome are hidden in zen, the editor uses a soft serif with comfortable line length, a thin progress line at the bottom tracks Pomodoro or goal progress, and controls auto-fade while you type. A new zen settings popover (cogwheel, top-right) lets you switch Serif/Mono, enable an edge-dim mask and control auto-fade. On mobile you can swipe down on the exit pill to leave focus mode.
Open Fokus →Fullscreen on /fokus now has a large, always-visible "Exit focus mode" button in the top-right with an Esc hint, making it easy to close on both desktop and mobile. Goal and Pomodoro pills moved to the bottom-left so they no longer collide with the close button.
Open Fokus →The /fokus writing tool gets a polish pass: clearer lead, calmer mobile header (title + autosave + zen on one row, goal and Pomodoro pills in a horizontal scroll so nothing clips at 360px) and a more descriptive "About Fokus" copy. The tool now appears everywhere in navigation — desktop tools menu, mobile menu and search — and has its own FAQ answer entry for AI answers.
Open Fokus →Write distraction-free in your browser with a Pomodoro timer, custom goal (words, characters, minutes or pages), fullscreen mode and export to Markdown, plain text, PDF or DOCX. Everything is saved locally — your text never leaves your device.
Open Fokus →Krymp moves to krymp.io as the canonical domain. All canonical URLs, sitemap, robots, OG images and hreflang now point there. krymp.hasty.se keeps working via 301 redirect so existing links and bookmarks stay intact.
The hero now features sharper copy: the H1 reads "Shrink the path to results free", the subtitle focuses on smart, anonymous and free tools, and the CTA button says "Open the toolbox".
Go to the start page →We now surface Krymp's promises directly below the hero — free, no accounts, runs locally, files stay with you, no ads, no limits. A horizontal list with clear icons, scroll-snap on mobile and a six-column grid on desktop.
Go to the start page →The live preview stalled or jumped backwards while dragging the scrubber, especially on mobile. We now coalesce seeks via requestAnimationFrame, use fastSeek where available and ignore the video's timeupdate event while you drag — so the picture tracks your finger in real time. The frame-by-frame arrows now measure the video's real FPS and step exactly one frame at a time.
Open the tool →Captured stills were sometimes soft or showed the wrong frame after a seek. We now wait for the actual frame to be decoded (via requestVideoFrameCallback with a double-rAF fallback) before grabbing it, render in high-quality mode and bump the JPG default to 95%. Result: crisp, correct stills even mid-clip.
Open the tool →A new slider below the video lets you scrub precisely and step frame-by-frame with the arrows — works just as well on mobile as on desktop. Auto mode no longer has a hard cap: the frame-count slider grows dynamically so you can drag it as high as you want, and the input accepts any number.
Open the tool →When your board has video slots, you can now export it directly as MP4 (h264, faststart) — great for iMessage, Instagram and Reels that don't always play WebM. The conversion runs locally via ffmpeg.wasm. New layouts: Filmstrip with 2, 3, 4 or 5 landscape sections stacked vertically — pick the count that matches how many clips you want to show.
Open Moodboard →Moodboard has supported both stills and video clips for a while, but it wasn't visible in the hero, dropzone or English landing page. Now it's spelled out: drop JPG/PNG/WebP or MP4/MOV/WebM, clips auto-play muted in a loop, scrub the poster used for static exports, and export the whole board as PNG/JPG/PDF or an animated WebM loop. HowTo, FAQ and meta descriptions updated in sv + en.
Open Moodboard →Auto-mode now has its own heading and badge so it's obvious. You can pick exactly how many frames you want (type any number — the 30-cap is gone). The tool is now also in the tools menu and on /om so you can find it without guessing the URL.
Open Thumbnail Generator →The Video Thumbnail Generator has been rebuilt to match the rest of Krymp's tools: Steg 1/2/3 rhythm, the shared dropzone, the global export button with mobile support, sticky export bar, FAQ + HowTo, and full Swedish/English translations. The English route now runs the same tool as the Swedish one — no longer a separate landing page.
Open Thumbnail Generator →Extract still frames from a video clip at full resolution. Auto-generate multiple frames or capture your own, then export to YouTube, Instagram, TikTok, LinkedIn, Facebook, X and Pinterest sizes — or a custom size. Multiple frames and formats are bundled as a .zip automatically. Send any frame straight to our other tools (compress, remove background, crop, upscale, anonymize, moodboard). Everything runs locally in your browser.
Open Thumbnail Generator →Drop MP4/MOV/WebM clips into slots next to images. Clips play live in the editor; trim in/out, pick a poster frame for still exports, and save the whole moodboard as an animated WebM loop. Everything runs locally in your browser.
Open moodboard →You can now upload videos of any size to /video. We removed the hard 500 MB limit — very large files just show a warning so you know it may take longer. All processing still happens locally in your browser.
Open video compression →We now run automated regression tests for image alt text, sitemap/noindex consistency and JSON-LD structures on every pull request. Broken metadata is caught before it reaches production.
Generic "Read more" links are now descriptive — "Read the full FAQ", "Learn about our privacy and security". Image previews in file cards now use descriptive alt text instead of the raw filename — better for screen readers and SEO.
/var-paverkan now shows PDF pages, words and media seconds we have processed, the most common input formats and the share of runs that happened fully in your browser. Tools like Compress image, PDF, Video and Convert now log formats and volume anonymously so the numbers fill in automatically.
See our impact →Krymp Index now shows top input formats, top conversions, weekday rhythm, share of fully local runs and total volume in pages, words and media seconds. All anonymous, aggregated and quotable — powered by the same anonymous event model as before.
Open Krymp Index →Suggestions in 'Add' now update live when you change a base color — the popover no longer closes when you tweak nearby colors. Plus a new accessibility preview where you pick text and background from the palette and see WCAG contrast in real time with sample text.
Open Colors →Each role in the brand palette on /brandkit now gets its own full-width row. When you add a color, we suggest colors that match the one you're working with (analogous, complementary, triadic, monochrome) — or pick your own. A WCAG contrast matrix across all base colors is back at the bottom for quick accessibility checks.
Open Colors →On /brandkit you can now build a flexible brand palette with any number of primary, secondary, accent and neutral colors. Each color keeps its own 50–900 scale, and AI/preset can either add to the palette or replace it.
Open Colors →When tapping Index in the footer, the page now automatically scrolls so the full index panel is visible. The scroll now follows the open animation instantly — without delay.
/var-paverkan crashed during render due to a missing QueryClient. The page now fetches the numbers directly and shows the totals as expected.
Open Our impact →Trend data in the press room now shows aggregated, anonymous figures since launch instead of just the last 30 days.
Open Press room →The press room now shows live anonymous 30-day usage figures, the correct tool count (auto-updated) and downloadable Hasty logos.
Open Press room →Full one-hour clips can now be transcribed without hitting a "too many transcriptions" error. We raised the per-IP cap so every chunk from a 1-hour clip fits.
Open Audiogram →You can now upload audio up to 1 hour (was 5 minutes). Great for full podcast episodes and interviews — still entirely in your browser.
Open Audiogram →Picking a preset (Instagram 3×3, Story, panorama, 2×2, 4×4) now scales each tile to the correct channel target: 1080×1080 px for the IG feed and 1080×1920 px for Stories. The tile-size readout shows the export size instead of the source pixels.
Open Image splitter →Removed the "Step 1" label, turned "Copy link" into an icon button, and moved Undo/Redo right under the map preview for easier access.
Open Map →Settings on /karta are now grouped into five tabs (Style, Place, Text, Details, More) and the sidebar is sticky so the map stays visible while you tweak.
Open Map →You can now toggle roads, water, buildings and parks individually on /karta — not just labels. The terrain theme stays raster and shows all layers.
Open Map →Krymp Index now offers ready-made 1200×630 press images per dimension: data saved, time freed, CO₂ avoided, total compressions and local-only data. Values refresh automatically and the images can be hotlinked from /og/krymp-index.svg?dim=…&lang=sv|en. A new "Press images per dimension" block on Krymp Index previews each image and copies the URL with one click.
Open Krymp Index →Published Krymp Index milestones (e.g. "first petabyte saved", "millionth protected file") are now available as an open RSS feed at /krymp-index.rss. The feed is auto-discoverable from both /krymp-index and /en/index-report via <link rel="alternate">, so Feedly, journalists' readers and AI answer engines can subscribe directly. 30 min cache, open CORS.
Open Krymp Index →Krymp Index now shows a separate "Files that never got uploaded" figure summing bytes processed entirely in the browser, a "Krymp by industry" chart splitting runs and saved data across marketing, design, office, media, dev/web and personal, plus two expandable blocks — "How we calculate" and "Limitations & caveats" — documenting formulas, time estimates, anonymity thresholds and uncertainties right on the page. The industry split is driven by a new field in the KPI catalogue and is also exposed in the data layer.
Open Krymp Index →Krymp Index now ships a press-room kit: a live SVG badge at /embed/krymp-index/score.svg (light/dark, sv/en, open CORS) that any site can embed with an <img> tag, a large 1200×630 share image at /og/krymp-index.svg that is now wired as og:image and Twitter image on both /krymp-index and /en/index-report, a new CO₂ panel that converts saved bandwidth to kg CO₂e (0.06 kg/GB, Carbon Trust 2021), a "This week's movers" list showing percentage change per dimension over the last 7 days vs. the previous 7, and extended JSON-LD with Schema.org Observations per dimension so AI answers and data tools can cite the numbers directly.
Open Krymp Index →Krymp Index now has a complete English counterpart at /en/index-report with its own JSON-LD (Dataset + FAQ), hreflang sv↔en and x-default. Every KPI card shows a 90-day sparkline directly under the number. Quotable blocks got stable anchor IDs (#q-data-30d, #q-time-30d, …) plus a separate "copy permalink" button so journalists can deep-link individual numbers. Mobile gets a sticky compact bar with the friction score once the hero scrolls out of view, and copy confirmation now uses a toast instead of silent feedback. The loader pulls 90 days of daily aggregates in a single call.
Open Krymp Index →New page /krymp-index sums up Sweden's file friction across five dimensions: data saved, time freed, protected files, shareable artifacts and creative works. Every tool is mapped in a KPI table so audiogram, map, transcription and favicon count too — not only byte-savers. A 0–100 friction score shows the trend vs. the previous month. The full index ships as machine-readable JSON and CSV under CC-BY 4.0 at /data/krymp-index.json and /data/krymp-index.csv. Quotable sentences and ready-made article angles can be copied straight from the page.
Open Krymp Index →The page is no longer just a landing page — you can now drop one or many images and resize them right in the browser. Pick longest side (1080 / 1920 / 2560 px presets), percent (25 / 50 / 75 %) or exact pixel dimensions with aspect-ratio lock. Output as JPG, PNG or WebP with a quality slider. Batches download as a .zip. Everything runs locally via the Canvas API — nothing is uploaded. New English counterpart /en/resize-image with hreflang, HowTo + FAQ JSON-LD.
Open the tool →The entire AEO answer library is now bilingual. New routes /en/questions and /en/questions/<slug> render English answers with QAPage JSON-LD (inLanguage: en), bilateral hreflang to /fragor and /fragor/<slug>, plus their own sitemap entries. A new /en/llms-full.txt mirrors /llms-full.txt in English for Perplexity, ChatGPT Search and Claude. ai.txt now lists both LLM feeds and declares Languages: sv-SE, en.
Open English Q&A →Shared chrome that appears on every tool — the feedback prompt, share modal, lifetime savings chip and preset menu — now uses i18n with both Swedish and English copy. Toasts, aria labels, share text, mailto subjects and confirmation buttons now follow the language toggle.
Every Swedish page that has an English counterpart now links bilaterally via hreflang in <head> and sitemap.xml — Google clusters sv/sv-SE/en/en-US/x-default as a multilingual group. Covers 37 PDF tools, /bild, /pdf, /video, /bakgrund, /upscale, /moodboard, the audiogram suite, both /maska-bild and /maska-dokument families, the full map suite and /bilddelare.
79 missing keys in en.json are now translated: every SEO body for compress/image/video/watermark/convert, every keywords meta, FAQ for image and transcribe, plus HowTo structured data for 8 tools (moodboard, image, transcribe, signature, character count, audio, markdown, anonymizer). English users no longer see Swedish fallback text on /en/ routes.
We swapped half-transparent text tones (text-foreground/50, text-muted-foreground/60–80) for full design tokens across navigation, search lists, paste hints, comparisons and cards. Body text now hits the WCAG AA contrast target (4.5:1) in both light and dark mode.
After splitting you can now write a caption + hashtags once and ship a captions.txt inside the zip with one block per post in posting order. Quick-buttons open Buffer, Later or Meta Business Suite in a new tab and copy the caption to your clipboard at the same time. A new poster PDF (A3/A4) combines every tile on one page with optional crop marks — great for review prints and sticker walls.
Open the image splitter →Big lift for Instagram creators: a new carousel mode splits the image into 1080×1350 slides with configurable overlap so the swipe reads as one. Auto-focus uses native face detection (Chrome/Edge) or saliency to keep the subject inside a tile instead of on a seam. Safe-zone overlays show where the profile UI clips the top and bottom of each tile. A new gutter (white/black/transparent) gives the border-grid look, and an IG profile mockup shows what your feed looks like before you download.
Open the image splitter →New brand mode on /brandkit. Build a palette from scratch via AI prompt, industry preset or a base color that auto-generates secondary, accent and neutral with 50–900 scales. Pull colors from multiple sites, images and PDFs, and drag the colors onto the right role (primary/secondary/accent/neutral). With light/dark preview, contrast check and the same exports as before.
Open colors →Faster fine-tuning in the preview: use the arrow keys to nudge the crop area (hold Shift for bigger steps), + and − to zoom in/out, and 0 to reset. Pinch-zoom and drag on mobile are now rAF-batched for smoother updates, and the crop area can never be dragged past the image edges.
Open the image splitter →You can now zoom and pan inside the preview to pick exactly which part of the image gets split — drag with your finger, pinch, or hold Ctrl while scrolling. New nudge buttons (← ↑ ↓ →) make it easy to fine-tune the crop area, especially on mobile. You can also adjust grid line thickness (1–6 px) and pick white, black or brand color so the lines stay readable on any image.
Open the image splitter →The image splitter now updates tiles live as you tweak rows, columns or format — no extra click needed. Instagram mode numbers tiles in posting order and names files post-01.jpg, post-02.jpg and so on. Images that don't divide evenly are auto-trimmed, and a new lock-to-square toggle yields perfectly square Instagram tiles even from non-square sources.
Open the image splitter →New tool that splits an image into tiles — perfect for Instagram grids (3×3), panorama posters or print. Pick rows and columns, see the grid overlay live and download every tile as a .zip. Everything runs in your browser, no upload.
Open the image splitter →The language switcher now shows the active flag and language code directly in the button. The dropdown shows both the native name and translation, the active language is marked with a clear "Active" pill, and hover state can no longer be confused with the selected language.
The mega menu is now wider and calmer. The left column shows the 8 most-used tools as icon cards, and the right column lists every other tool in two columns — easy to scan without categories or images.
The mega menu now shows every category at once — Image, PDF, Video, Text, AI, Privacy and Other — with a quick jump-to row on top. No more category tabs swapping by accident when the cursor passes over them.
The map is no longer a tinted web map. We fetch vector data and draw every road width, fill and label ourselves — so Editorial, Mono, Blueprint and the other styles look like designed posters instead of filtered web maps. Editorial is the new default: cream paper, thin dark roads, lots of air.
Open map →The map tool's style picker has been rebuilt with clearer style families and better preview cards. The default now opens in a clean premium style, while mono, blueprint, editorial and relief/topography feel more like distinct map worlds straight in the preview.
Open map →The tools menu has been rebuilt to scale as we add more tools: a search field at the top (⌘K), categories on the left (Image, Video & audio, PDF, Text, AI, Security, Other) and a denser grid on the right. At the bottom you'll find Popular and your recently used tools.
You can now zoom the map straight from the preview — pinch on mobile, scroll with your mouse on desktop, or tap the new +/− buttons in the top-right. The current zoom level is shown under the buttons.
Open map →Opening the mobile menu now pins the page behind in place — no more jumps or rubber-banding when you scroll inside the menu or tap the dim backdrop. Your scroll position is restored when the menu closes.
The frame (thin, polaroid, paper) is now drawn in both preview and export — what you see is what you download. Custom palette colors now tint the whole map, not just text and markers. Layer toggles that need the upcoming vector map (roads, water, buildings, parks) are clearly marked so you don't have to guess — labels can still be toggled off.
Open map →New use-case pages with complete HowTo, FAQ and tips: /en/running-map (GPX from Strava/Garmin), /en/hiking-map (long trails, Komoot, AllTrails) and /en/poster-map (A3/A4 print). All hreflang-linked between Swedish and English.
See poster guide →You can now grab and drag any marker directly on the map preview — works smoothly with mouse on desktop and finger on mobile. Short clicks without movement keep the old behavior and recenter the map. No settings to read, just natural drag-and-drop.
Open map →Pin your favorite styles with the star on each theme chip — favorites get their own row at the top for quick switching. Power-user shortcuts (T = next theme, R = random, M = new marker, F = favorite, +/− = zoom, ⌘S = export, ⌘K = copy link, ⌘Z = undo). Press ? anywhere for a help overlay with every shortcut.
Open map →Export the map as SVG where title, subtitle and coordinates are real vector text. Open in Figma, Illustrator or Affinity Designer and tweak typography, color and layout without losing sharpness. The base map is embedded at full resolution (1×/2×/4×).
Open map →Upload your run, hike or cycle (.gpx from Strava/Garmin/Komoot or .geojson) and see it drawn on top of the map with custom color, width and style (solid/dashed). File stays on your device. Also toggle a compass rose and a meters/km scale bar for a wall-print-grade map.
Open map →Pick from 10 curated fonts (Inter, Bebas Neue, Playfair Display, IBM Plex Mono, Cormorant, Outfit, Space Grotesk, Archivo Black, DM Serif Display + System). Tweak weight, uppercase, letter-spacing and size live. Fonts load on demand and are baked into the export — so the file looks exactly like the preview.
Open map →Big flexibility update: custom palette (background, water, roads, labels) on top of the chosen theme, multiple markers with icon + label, four frame styles (thin, polaroid, paper), random style button, copy link to share your exact map, plus undo/redo (⌘Z) — all in the browser.
Open map →Adds Midnight, Coral, Neon and Patina (copper patina green) — 12 curated map themes total, inspired by Terraink. New /en/map-themes page lists them all with click-to-apply and clean SEO markup.
Browse all themes →Tre nya stilar (Papper, Mono, Terräng), tunad Blueprint, separat label-pass för skarp text, ny stilväljare med gradient-förhandsvisning, samt mobil-perf (DPR-cap, pool och cache-budget).
Öppna karta →Bytte basemap-tiles till Carto Positron/Dark/Voyager för en renare poster-känsla, med subtil vinjett och uppdaterad pin.
Öppna karta →The map preview now uses a stable tile-based render path instead of the setup that could stall on a blank canvas in some mobile environments. Stockholm and other places show up directly again.
Open Map maker →Map now uses OpenFreeMap's hosted base styles (Positron/Bright/Liberty) and recolors them per preset. The preview shows a real map instead of staying blank or falling back.
Open Map maker →Map now uses a more robust init flow, extra resize checks and a clear fallback if rendering fails. The preview no longer stays on "Loading map…" without reason.
Open Map maker →New in Map: a button to jump to your own location (geolocation), recent searches as chips under the search box, and a draggable marker — click or drag on the map to fine-tune the position without searching again.
Open Map maker →The Map preview could occasionally remain behind the loading overlay even after the style was ready — typically right after searching for an address. Switched the ready signal to MapLibre's styledata event with an extra resize, so the map shows as soon as the style is loaded.
Open Map maker →New landing pages showing exactly how to use Krymp Map for CV, wedding invitations, property listings and Instagram — with the right style, format, resolution and tips per use case. Swedish + English with hreflang, JSON-LD (HowTo + FAQ) and linked AEO answers.
Open Map maker →Type an address or city, pick from five styles (Noir, Pastel, Blueprint, Krymp, Warm), choose a format (Instagram post, story, A4, A3, circle and more) and export as PNG, JPG or print-PDF at up to 4×. The whole map renders and exports locally in your browser — only the place lookup goes via OpenStreetMap. Perfect for CVs, wedding invites, real-estate listings or just a beautiful city map for Instagram.
Open Map maker →The audiogram video watermark used to read "krmp.se" — it now says "Krymp". You can still turn it off in the FX panel if you want a completely clean video.
Open Audiogram →Audiogram (audio → shareable video with a waveform) is now surfaced in the top/mobile menu, the mega menu and the tools list on the home page — no more typing the URL directly.
Open Audiogram →Sign PDF, Extract text and PDF to images are now fully localized — buttons, toasts, page navigation and quality labels. Images→PDF lost its last Swedish strings (Fit, Keep proportions / Fill page) too.
Open Sign PDF →The watermark tool is now fully translated (position, color, opacity, rotation, toasts) and picks up files dropped on the /pdf hub. English users get the right language throughout.
Open Watermark →Estimates and privacy notes in the global export button are now translated across every PDF tool (compress, merge, split, reorder, images→PDF, PDF→images, sign, watermark, extract text). No Swedish leftovers on the English site.
Open PDF hub →The Reorder pages tool now picks up files dropped on the /pdf hub and has translated strings (select, rotate, remove, export estimate and privacy note) — no Swedish leftovers on the English site.
Open Reorder pages →The Split PDF tool now picks up files dropped on the /pdf hub automatically. Export-button estimates and privacy notes are translated to English instead of hard-coded Swedish.
Open Split PDF →The rotate tool now shows a thumbnail grid where you can turn individual pages, drag to reorder, or remove pages — on top of quick buttons that rotate the whole document. Still 100% local in the browser.
Open Rotate PDF →Turn every page 90°, 180° or 270° right inside /pdf. Local, free and no account — perfect when scanned landscape pages ended up in a portrait PDF.
Open Rotate PDF →The English version of the PDF hub is live at /en/pdf with proper canonical, hreflang (sv/en/x-default) and its own metadata. English users now land in their language instead of seeing the Swedish hub.
Open English PDF hub →The /pdf hub now shows a clear comparison vs iLovePDF and Smallpdf (local, free, no account, no uploads), each tool card links to its dedicated page (e.g. /krymp-pdf, /sla-ihop-pdf) and the pdf-lib bundle only loads on hover or once you open a tool — so landing feels faster.
Open PDF tools →Drop a PDF anywhere on /pdf and we pick the right tool: 1 PDF → Compress, several → Merge, images → Images to PDF. On mobile the cramped tab strip is replaced by a big select menu that's easier to tap.
Open PDF tools →Translated RecentFilesPanel (title, transparency panel, clear button, footer), PillNav settings button (aria/title) and the compare slider with result stats — all now go through i18n.
Translated more global UI bits: the savings toast ('You saved …'), the session pill in the header, the 'Install as app' chip (PWA) and the keyboard shortcuts panel (? modal) — all now go through i18n with proper plural and number formatting per language.
Translated CloudConsentDialog (the prompt shown before a tool sends data to an external AI/cloud service) — title, intro, field labels (Service, What gets sent, Why, Retention), the privacy link and the Cancel / Allow once / Allow for this session / Always allow buttons now go through i18n.
Translated the last bits of the upload flow — the paste hint (⌘V/Ctrl+V), global drop-overlay toasts (pasted / unsupported), the URL field example placeholder and the mobile 'Files' button. English visitors now get fully English uploads across every tool.
Translated the feedback dialog (title, rating labels, fields, buttons, toasts) and the generic tool-error fallback (title, body, Retry / Home / Report buttons) via new feedback.* and toolError.* keys. English visitors now see English copy when something crashes or when they send feedback.
Translated the global ModelLoadingBar (sticky bar for local AI model downloads) and FirstTimeLocalOverlay (first-time card on /describe, /upscale and others) — titles, descriptions, status (Error/Done/Advanced), 'Run in cloud instead', 'Hide' and 'End-to-end private' now go through i18n.
The footer index (statement, intro, section headings §01–§04 and all 16 links across Krymp/Platform/Legal/Internal) plus Close/Index toggle, trust links (About, Security, Privacy) and feedback button now go through i18n. English visitors see English copy across the whole footer.
The PDF hub's 9 tool cards (title, description, example, mobile short label), 'Open tool', 'All PDF tools', the privacy box and the dropzone label now go through i18n. English visitors on /pdf?tool=… see English copy across the navigation.
Open PDF →Swapped the last hardcoded Swedish strings in /konvertera (read/decode toasts, batch bundling, sample file, step headings, fine-tuning, target size, privacy note) and /vattenstampel (privacy note, guide link). English visitors now see English copy end-to-end.
All remaining Swedish strings in the /audiogram editor (FX chips, audio controls, format presets, trim, transcript editor, export status, language picker, share toasts, footer links, etc.) now go through i18n. /en/audiogram mounts the same editor with an English head, hreflang and forces UI language to English — AI flows (transcription, highlights, titles, translation) already pick up the language. Data catalogs (themes, templates, waveforms, export presets) use labelKey so chip/preset names follow the language too.
Open audiogram →Swapped the last Swedish strings in /beskriv (model loading, PDF export, privacy note), /upscale (export estimate, privacy, guide), /metadata (export status) and /brandkit (guide, 'All colors locked', eyedropper fallback). Everything now goes through i18n so English visitors see English copy end-to-end.
All hard-coded Swedish aria-labels in /audiogram (close drawer, Generate AI background, Export WebM/MP4/GIF/WAV, dropzone, demo clip) plus the 'WAV (audio only)' button are now in i18n. Screen readers on /en/audiogram finally read the right language.
Open audiogram →Swapped the last Swedish strings in /some: the ETA counter ('Calculating time…', '~X min left'), upscale toasts (failed images, cloud fallback), download cancelled, and the 1×/2×/4× labels plus heavy-reasons in the export popover. Everything now flows through i18n so English visitors see English copy end-to-end.
Open Social →Background, Metadata, Favicon, Video, QR and Colors still showed their privacy copy in Swedish inside the export button. The strings are now moved to shared i18n keys (shared.privacyNote*) so English visitors see English text in the export popover.
The full changelog is now translated — every one of the 395 entries has a title_en and body_en. On /en/changelog everything reads in English with no Swedish fallbacks, from today's release down to the very first entry.
Open changelog (English) →Removed the last hard-coded Swedish strings in /bild (export toast, cancelled download, AVIF warnings, progress and privacy copy). Everything now goes through i18n so English visitors no longer see a half-translated UI.
Open the image compressor →Changelog entries now have optional title_en and body_en. On /en the English version is shown automatically and link labels also follow the language. Over 110 of the most recent entries are translated (audiogram, mask image/document, moodboard, PDF, mobile menu, the /en rollout and more); the rest fall back to Swedish until they are translated.
Open the changelog →The landing page still claimed the editor was Swedish-only. Since the whole audiogram editor now follows the language switch, the note has been replaced with a message confirming that UI, captions and AI prompts all speak English when the header is set to en.
Open audiogram (EN) →The 'Use Audiogram for…' heading, the three cards (Podcast, Marketing, Education) and the translation language option 'Dutch' are now translated. On /en the cards point straight to /en/audiogram-podcast, /en/audiogram-marketing and /en/audiogram-education instead of the Swedish routes.
Open audiogram →'Transcribe now / again', the progress text, 'X words matched', 'Color captions per speaker' and the ElevenLabs consent note now show in English on /en and Swedish on /audiogram.
Open audiogram →The toast about alternative formats (1:1, 9:16, 16:9) and the 'Switch format' button now follow the language, as do the transcript editor's 'Split', 'Merge' and 'Remove' tooltips. The last Swedish UI strings in /audiogram are gone when /en is active.
Open audiogram →Suggest clips/title, Surprise me, caption position (Top/Middle/Bottom), smart highlight, auto-emoji, transcript editor, the drop-zone prompt and the feature cards under the uploader now follow the language switch — no Swedish UI fragment left in the editor when /en is active.
Open audiogram →Placeholders, caption-mode chips (Off/Manual/Auto), Translate transcript, logo and background buttons, AI background copy, helper text, the demo button and the uploader hint now follow the language in the header. No Swedish UI remains in the editor when /en is active.
Open audiogram →Copy link, Share on X, Share on LinkedIn and Copy transcript render and the tweet text is sent in English when the header is set to English, and in Swedish otherwise.
Open audiogram →New English landing pages for audiogram-podcast, -marketing, -education and the gallery (/en/audiogram-podcast, /en/audiogram-marketing, /en/audiogram-education, /en/audiogram/gallery) with their own head, canonical and hreflang. Auto-captions now follow the header language instead of always being Swedish, and consent + toasts are translated. The watermark now reads krmp.se.
Open audiogram →The FAQ section at the bottom of /audiogram now shows English questions and answers when the header is in English, and Swedish when it is in Swedish. Hreflang sv↔en + og:locale verified on both routes.
Open audiogram →On mobile the preview is now compact and sticky so you always see your audiogram, and Settings opens in an overlay drawer on top of the preview instead of disappearing below. Faster to tweak theme, captions and export without losing context.
Open audiogram →All themes, formats, waveforms, templates, export presets and caption styles now have i18n keys — the lists in the editor switch language instantly. AI-generated titles and highlights also follow the language you picked in the header, so English users no longer get Swedish suggestions.
Open audiogram →The audiogram editor now uses i18n keys for the header, onboarding, shortcuts, all control-panel headings, the export popover and the most common toast messages. Switching language in the header takes the whole tool with it — essential for international visitors on /en/audiogram.
Open audiogram →The audiogram tool now has an English landing page with its own title, description, FAQ, HowTo steps and hreflang links between /audiogram and /en/audiogram. International visitors find and understand the tool immediately, and search engines can index both languages correctly.
Open audiogram (EN) →The audiogram editor, gallery and use-case pages now have rich structured data (SoftwareApplication, HowTo, BreadcrumbList, ItemList, extended FAQPage), a slot in the sitemap, llms.txt, llms-tools.json, the llms-faq corpus and a dedicated AEO answer at /aeo/audiogram.md. Audiogram is now a top-level topic in ai.txt so ChatGPT, Perplexity and Google AI Overviews can cite the tool more easily.
Open audiogram →The preview now follows along as you scroll through the settings on mobile, so you always see how the audiogram changes. The theme picker became a horizontal snap row with larger tiles — quicker to swipe between looks on a small screen.
Open audiogram →Audiogram use-case pages and guides now show a sticky 'Open the tool' button at the bottom on mobile so the CTA is always a thumb-tap away. The audiogram editor got larger tap targets on chip buttons, a new Settings shortcut in the global Export bar on mobile and a square mini-demo that gives the waveform more room to breathe.
Open audiogram →The audiogram tool now uses the same global Export button as the other Krymp tools. Pick a format (MP4, WebM, GIF, WAV or all) directly in the popover, see progress in the trigger and skip scrolling the side panel — works the same way on mobile and desktop.
Open audiogram →The use-case pages for podcast, marketing and education now show a short animated audiogram preview with a waveform and word-by-word caption just above the closing CTA. You see straight away how the result can feel before you open the tool.
See example →We cleaned up accessibility in the audiogram editor: control groups are read out with their heading, choice chips behave like radio buttons, export buttons have descriptive labels and export progress is announced live to screen readers. Clearer focus ring during keyboard navigation.
Open audiogram →After your first audiogram export in a session we now suggest a complementary format — e.g. 1:1 for LinkedIn when you already exported 9:16 for Reels. One click switches format so you can quickly render both and cover more channels.
Open audiogram →New waveform 'Aurora 3D' stacks three Aurora layers at different depths that move at different speeds and tilt gently with the bass. Gives a cinematic 3D feel without costing performance — pick it in the waveform selector in the editor.
Open audiogram →A new 'Generate AI background' button in the audiogram editor creates an abstract, subject-free background image from your title and the theme's color palette. No text or objects — just soft cinematic gradients that let the headline and waveform take centre stage.
Open audiogram →There is now a 'Try with a demo clip' button in the empty state of the audiogram editor. We generate a 14-second audio clip locally in the browser so you can click around the editor, try templates and see the export directly — without hunting for your own audio file.
Open audiogram →New minimal embed player at /embed/audiogram — embed audiograms as an iframe without leaving the page. Takes audio URL, title and color via query parameters, completely without login or server storage.
Open audiogram →New gallery page with six curated audiogram styles — from Aurora and Podcast Mono to Karaoke Pop. Click to open a template directly in the editor and adapt it to your voice.
Browse the gallery →The 'Karaoke' caption preset shows one word at a time with a big pop-in and accent glow — perfect for punchy quotes. After exporting on mobile you can share the mp4 directly to Instagram, TikTok or your messaging app via the system share menu.
Open audiogram →New panel that lets AI pick the three best 30-second segments from a long episode, suggest three punchy titles and randomise the entire design within your brand kit. Finds time-coded words from the auto-captions and scores pauses, speaking pace and keywords.
Open audiogram →Auto-captions now fetch speaker IDs from ElevenLabs and color captions so that an interview clearly shows who-says-what right on screen. Stable color per speaker across the whole clip.
Open audiogram →New export path via WebCodecs: H.264 video and AAC audio are encoded directly in the browser without ffmpeg.wasm. 5–10× faster in Chrome, Edge and Safari 17+. Firefox and older browsers automatically fall back to the standard path.
Open audiogram →Three new pages showing how Krymp Audiogram is used in practice — for podcast, marketing and education. With HowTo, FAQ and AnswerBlock schemas for Google and AI answers.
Audiograms for podcasts →Translate the transcribed text into 12 languages (English, Swedish, German, Spanish, French and more) — captions switch language in sync with the audio and the .srt/.vtt export follows.
Open audiogram →Two new frequent questions (GIF/WAV export and local saving) and FAQPage schema (JSON-LD) so answers can be shown directly in Google.
Open audiogram →Toggle the krmp.se watermark in the FX panel, copy the share link with UTM tags or share directly to X and LinkedIn. Once you've transcribed you can also copy the entire transcript with one click.
Open audiogram →Space, ←/→, M and E give a faster flow in the editor. New welcome banner explains the three steps — drop audio, tweak, export — and can be dismissed permanently.
Open audiogram →Quick start with four ready-made templates (Podcast, Reels/TikTok, Interview, Neon promo), save your own brand kits locally and restore the latest project with one click — everything is saved in the browser without login.
Open audiogram →New Audio panel: automatic normalization to -1 dB peak, fade in/out and manual gain — all applied at export time. We also show estimated BPM so you can pick a waveform and FX that match the tempo.
Open audiogram →Three new visualizations — Particles, Radial and Liquid blob — that react to bass/mid/treble. New FX panel with camera shake, beat flash and vignette gives a cinematic feel without extra settings.
Open audiogram →New export formats: animated GIF (for sharing where video does not work) and pure WAV audio. Format presets for Reels, Shorts, X, LinkedIn and the Instagram feed set the right format with one click. New batch button exports all formats in a queue.
Open audiogram →Captions stepped up to pro level: 5 style presets (TikTok, Descript, Neon, Editorial, Minimal), pop-in animation on the active word, automatic highlight of keywords, optional emoji injection, transcript editor and export to .srt/.vtt. Safe zones keep the text inside the Reels/Shorts UI.
Open audiogram →Three additions to the audiogram tool. Auto-captions are now drawn in karaoke style — the active word grows, changes color and glows in the theme accent so the viewer follows the speech. You can add your own background image that covers the whole surface with a dark overlay so text and waveform still read. And you can trim start and end with a dual handle — both preview and export follow the trim window.
Audiogram →Two additions to the audiogram tool. You can now pick caption mode: Off, Manual (paste text that is distributed evenly) or Auto (AI) — where we send the audio to ElevenLabs and get back words with timestamps so the text follows the speech exactly. Auto-captions require consent the first time and everything else still happens locally. We also added a nicer share image when you share /audiogram on social media.
Audiogram →Upload an audio clip, pick a format (1:1, 9:16, 16:9 or 4:5), theme and waveform — and you get an animated video you can share on social media. We built a signature waveform, Aurora, that reacts to bass, mid and treble separately and pulses with the music. You can add a title, subtitle, logo and captions, and export to WebM (fast) or MP4 (universal). Everything runs in your browser — the audio is never uploaded.
Audiogram →On Mask Image there is now an invisible interaction surface around the image — so corner handles, edges and the delete button that fall outside the image can be clicked and dragged just as if they were inside. You no longer need to drag the selection toward the center just to be able to resize it.
Maska bild →Previously the selection snapped back to the image edge as soon as you dragged across it — which made it tricky to cover something right at the edge, especially with soft edge or ellipse shape. Now you can drag, move and resize freely outside the image edge. The part of the effect that ends up outside is automatically ignored at export, but the shape and feather keep their real size so the transitions look clean all the way out.
Maska bild →On Mask Image the soft edge did not fade out the selection as it should — for rectangles the transition ended too sharply, and for ellipses the softness was clipped against the selection box edge. Now the shape is drawn on a larger internal surface so the entire blur tail fits, then clipped back. Result: a smooth, soft transition all the way around — whether you picked rectangle or ellipse and however high you pull the slider.
Maska bild →On Mask Image and Mask Document (PDF) there is now a layer panel that lists all selections you have placed on the canvas. On desktop it shows as a side panel to the right of the image/page — on mobile it opens as a bottom sheet via the new Layers button. Tick multiple selections and press Delete to remove them at once, or use Select all / Deselect to clear quickly. Clicking a layer jumps to the right page and highlights the box.
Maska bild →On Mask Image it could happen that the selection box disappeared when you pressed the red × or deleted it, but the pixelated/blurred area remained in the image. Now the canvas is redrawn immediately with the new list of selections in the same step as the deletion — so the box and its mask always disappear together.
Maska bild →On Mask Image every tap on the canvas created an empty history entry — the Undo button became active without anything to roll back, so you pressed and nothing happened. Now history is pushed only when a selection has actually been created, moved or resized. There is also a clear red × above the selected mark (visible on both mobile and desktop), and Delete/Backspace removes the selected box on desktop. On Mask Document the delete button on each selection is now always visible — no longer hidden behind hover.
Maska bild →On Mask Image the Settings pill landed on top of the Export button because it was absolutely positioned. Now the secondary button and Export sit side by side in a shared flex flow with breathing room between them — the same clean bottom bar in every tool (Mask Image, Mask Document, DOCX mask and Moodboard).
When the mobile menu opens each tool row now slides in from the right in sequence with a soft delay. The active tool pulses with a subtle dot, the star button pops to life when you favorite, and the soft light orbs in the background drift slowly for a more lively feel.
The mobile menu is rebuilt to quickly find the right one among all tools. Search field at the top with live filter, category chips (Image, PDF, Video & audio, Text, AI, Security, Other), star tools as favorites so they land at the top, and the 'Recent' list shows the four most recently used tools. Larger tap targets and a unified design replace the previous tiny chip rows.
The Undo button and history on Mask Image could stop working after you uploaded a new image or switched active image in the list — the buttons appeared to do nothing. The reason was that history was written to the wrong document. Now every change is saved and undone against the right image, both for a single image and in batch mode.
Maska bild →Previously 'Layout', 'Settings' and 'Share' floated as separate pills on top of the Export bar and could cover images and canvas content. Now the secondary button (Layout on moodboard, Settings on Mask Image/Document) is integrated into the same bottom bar, to the left of Export. The Share button is removed — Export + the system share menu do the same job without an extra button.
Testa på moodboard →The floating plus button in mobile view sat on top of the images in the pool and blocked the tools for editing the image. The button is removed — the same 'Add' function is still available in the bottom bar.
Moodboard →Added 12 new high-intent questions about PDF on iPhone/Mac, sending large files via email, HEIC→JPG, video for WhatsApp, image size for Hemnet/Blocket/Tradera, security in online PDF, and more. The answers flow automatically into AI search (ChatGPT, Perplexity, Google AI Overviews) via FAQPage/QAPage structured data and llms-full.txt.
Se alla frågor →On moodboard, Mask Image and Mask Document the Layout/Settings button sat in the same row as the Export bar and could overlap on narrow screens. The button now sits clearly above the bar.
Moodboard →You can now choose between rectangle and ellipse/circle as the mask shape, and fade out the edge with a soft-edge slider so the mask blends into the image instead of leaving a hard boundary.
Maska bild →Settings are now reachable from the bottom bar on mobile in Mask Image and Mask Document — opens as an overlay drawer instead of being pushed below the canvas.
Maska bild →Six new SEO landing pages on /en for compress-image, compress-video, convert-heic-to-jpg, remove-background, upscale-image and create-moodboard — each with its own HowTo, FAQ and SoftwareApplication schema, hreflang and a direct link to the Swedish tool.
Compress image (English) →Added new entries in /jamfor and /vad-ar covering image, audio and document workflows — for richer SEO coverage and AI-search citations.
Se alla jämförelser →Three new long-form guides answering high-intent questions about emailing large files, optimizing images for WordPress and compressing video for WhatsApp.
Skicka stora filer via mail →After a new release the client could request stale chunks and throw 'Failed to fetch'. We now detect the mismatch and reload silently — users no longer see the cryptic error.
The lawyers page on /maska-dokument now has its own OG image and the correct hreflang link to the English version.
Maska dokument för advokater →/maska-bild has a new OG image so the tool stands out when shared on social media.
Maska bild →New /en/mask-document.lawyers and .hr pages with their own meta, FAQ and HowTo schemas, plus hreflang to the Swedish versions.
Mask documents for lawyers →/maska-dokument got a richer meta description, a dedicated OG image and an English counterpart at /en/mask-document.
Maska PDF →Auto-detect on Mask Document now flags personal names in addition to emails, phone numbers and personal IDs.
Maska dokument →Export now strips PDF metadata, the 'Mask all' button applies every suggested mask in one click, and the export pipeline was hardened so masks can never be undone in the output.
Maska dokument →Auto-suggestions detect emails, phone numbers and personal IDs; DOCX files are now supported alongside PDF; and new use-case pages target legal, HR and journalists.
Maska dokument →New tool that lets you redact PDF documents in your browser — draw boxes, apply pixelation or solid fill and export a clean PDF without uploads.
Maska PDF →Mask Image now registers with the global Export button in the header, with the same popover, progress and cancel behavior as the other tools.
Maska bild →The /pdf hub now lists popular searches so visitors quickly land on the right tool.
PDF-verktyget →38 new English PDF keyword landing pages went live — each with HowTo, FAQ and hreflang back to the Swedish counterpart.
38 new English PDF keyword landing pages went live — each with HowTo, FAQ and hreflang back to the Swedish counterpart.
Krymp PDF →Four new use-case pages tailored for journalists, lawyers, HR and healthcare on /maska-bild.
Maska bild för journalister →Eight new Swedish keyword landing pages for the PDF tools — better coverage on long-tail searches.
Slå ihop PDF →The strength slider on Mask Image now updates the preview live as you drag — no need to release to see the result.
Maska bild →Strength on Mask Image (Low/Medium/High) is now a slider instead of three chips — easier to dial in exactly the right amount of blur or pixelation.
Maska bild →'High' blur on Mask Image now applies a much stronger Gaussian blur so the masked area becomes truly unreadable.
Maska bild →Blur was leaking outside the selection in some browsers — now it is clipped exactly to the selected shape.
Maska bild →On very large images the blur was too small to see — the blur radius now scales with image size so it is always visible.
Maska bild →The action buttons on Mask Image now float at the bottom on mobile so they are always reachable with the thumb.
Maska bild →Fixed a bug where the mobile menu on Mask Image opened on first tap on the canvas.
Maska bild →Mask Image got a round of mobile polish — bigger tap targets, sticky bottom bar and smoother dragging.
Maska bild →You can now process multiple images at once in Mask Image and export them all in a single zip.
Maska bild →New tool that lets you blur or pixelate parts of an image — draw boxes over faces, plates or IDs and export a safe version, all locally in the browser.
Maska bild →New tool that adds a text or image watermark to PDFs — pick position, opacity and font, locally in your browser.
Vattenstämpel PDF →New tool that extracts the text from a PDF and lets you copy it or download it as a .txt file.
Extrahera text →You can now select multiple PDF pages with Shift/Cmd+click and rotate, delete or extract them in a single action.
Organisera sidor →Reorder PDF pages with drag and drop in the new page grid — much faster than the old up/down arrows.
Organisera sidor →The PDF tool got a major rewrite with a new page grid, multi-select, drag-and-drop reordering and faster rendering.
Öppna PDF-verktyget →Fixed a bug where pinch-zooming on the canvas could open the browser's context menu.
Moodboard →Moodboard guides now snap to other elements and the center of the canvas while dragging.
Moodboard →You can now drag the dividers in the moodboard to resize columns and rows freely.
Moodboard →Long-press an image to get context actions, share the board directly via the system share menu, and the moodboard now supports landscape on mobile.
Moodboard →The moodboard got a big mobile polish round — bigger tap targets, sticky toolbar and smoother dragging.
Moodboard →Tapping an empty slot in the moodboard now also lets you pick from images already in the pool — not just upload new ones.
Moodboard →You can now tap an empty slot in the moodboard to add an image — no need to drag from the pool.
Moodboard →The moodboard got bigger tap targets and a thumb-friendly layout on mobile.
Moodboard →Fixed a regression where the mobile menu was not rendered after a route change.
Export status text for the PDF watermark tool now follows the language switch.
Vattenstämpel →The /trender and /guider sections now follow the language switch.
The /fragor (Q&A) section now follows the language switch.
The /jamfor (compare) and /vad-ar (what is) sections now follow the language switch.
Settings and feedback dialogs now follow the language switch.
The /status, /impact and /changelog pages now follow the language switch.
The /security, /privacy, /api and /press pages now follow the language switch.
Editorial policy, methodology and accessibility pages now follow the language switch.
Shared SEO components (breadcrumbs, trust ribbon, related-tools) now follow the language switch.
The GIF tool now follows the language switch.
Öppna GIF →Several SEO landing pages now follow the language switch.
More tools now follow the language switch.
The /impact page got a new hero, a sparkline of bytes saved over time and direct sharing to social media.
Se uppdaterade Vår påverkan →New /impact page that shows how many bytes Krymp has helped users save — fully public, no account needed.
Öppna Vår påverkan →The XML sitemap now includes the privacy page so search engines can discover it directly.
Öppna integritet →Many links now have more descriptive labels for screen readers instead of generic 'click here' phrasing.
Cookiepolicy →The /changelog page now has chips to filter by type (new/improved/fixed/security/data) and by area (image, PDF, video, AI…).
Öppna changelog →A new FAQ entry covering how to count characters, words and readability for SEO and editorial work.
Läs svaret →More tools and guides now cross-link to related Krymp pages so visitors can find the right next step.
Uppskala bilder →More tools got dedicated OG share images and stronger internal links to related guides.
Bygg e-postsignatur →Added more frequent questions on four of the most-used tools to better serve AI search and Google snippets.
Komprimera PDF →12 guides now have English share text and OG metadata so they render correctly on international social media.
Alla guider →The /bakgrund (remove background) page got richer copy, FAQ and use-case examples.
Ta bort bakgrund →Five tools got a clearer step-by-step section and additional frequent questions.
Bygg e-postsignatur →Each compression tool now has its own OG share image instead of the generic Krymp default.
Komprimera bild →The /bild and /transkribera pages got richer copy, FAQ and use-case sections.
Öppna transkribering →Three guides got tighter SEO titles and meta descriptions for better click-through from search results.
Läs moodboard-guiden →New /redaktionell-policy and /metodik pages plus open benchmark data at /data/benchmarks.json for full transparency.
Läs metodiken →Tools, guides and Q&A now cross-link to each other based on topic — better navigation and stronger SEO.
Bläddra i frågor →New /aeo Markdown corpus exposes Krymp's answers in a format AI search engines can ingest directly.
Se /aeo/index.json →Three new top-level pages target the highest-intent Swedish searches: compress image, compress PDF and crop image.
Öppna /komprimera/bild →New /fragor section that collects in-depth answers to the most common questions about file compression.
Bläddra bland frågor & svar →List and index pages now expose ItemList and CollectionPage schema for richer Google and AI-search results.
ChatGPT, Perplexity and Google AI Overviews now find a dedicated answer library exposed via llms-full.txt and the /aeo corpus.
Cleaned up duplicate meta descriptions and tightened wording so each route shows a unique, high-quality snippet.
Added llms.txt, ai.txt and an AI-friendly sitemap so AI search engines can index Krymp's tools and answers.
The /some video tool now shows per-file progress, retries and errors more clearly in batch mode.
Öppna SoMe-verktyget →You can now cancel an in-progress video job on /some without losing the rest of the batch.
Öppna SoMe-verktyget →More tools got their own OG share images.
Testa Ta bort bakgrund →More content pages got their own OG share images.
Se jämförelser →10 more glossary and comparison pages went live.
Se alla jämförelser →11 new pages covering glossary terms and direct tool comparisons.
Öppna ordlistan →The top logo's size is now reserved up front so the layout no longer jumps when the page first loads.
Cleaned up hreflang so pages without a real English version no longer claim an English mirror — search engines see the correct language map.
Glossary and comparison pages now show the author, publish date and last updated date for higher trust and better E-E-A-T signals.
Glossary and comparison pages now show breadcrumbs at the top so visitors and search engines see the page hierarchy.
Added RSS/Atom/JSON feeds for the changelog and Organization + WebSite JSON-LD on every page for richer SEO.
Added new AI-friendly catalogs (llms-tools.json, llms-faq.json) and a dedicated AI sitemap so AI search engines can ingest Krymp.
The /some video tool got a major performance boost — exports to all formats are several times faster thanks to a new encode pipeline.
Öppna Some →The markdown editor now has synced scroll between editor and preview, Mermaid diagram support, math (KaTeX) and a zen mode without chrome.
Öppna Markdown →The markdown editor gained image insertion, table builder, checklists and search & replace.
Öppna Markdown →The transcription tool now uses the global Export button in the header — same popover, progress and cancel behavior as the rest of Krymp.
Öppna Transkribera →Längre ljud- och videofiler stötte tidigare i serverns tidsgräns mot transkriberings-AI:n. Nu styckas ljudet upp i 3-minuters bitar direkt i webbläsaren och skickas i tur och ordning — varje bit hinner alltid svara i tid, och tidsstämplar/talare slås ihop korrekt i slutet.
Öppna Transkribera →The moodboard heading is now clearer and matches the rest of Krymp's typographic scale.
Öppna Moodboard →You can now build a custom gradient as background on the moodboard — pick two or more colors and direction.
Öppna Moodboard →You can now pick font, weight and size for the moodboard title and subtitle.
Öppna Moodboard →Paste text or an image directly onto the markdown editor — it's inserted at the cursor.
Öppna Markdown →Each moodboard slot now has hover icons for quick actions, and Undo/Redo lives in the toolbar.
Öppna Moodboard →New color harmony strip suggests complementary colors, drag and drop got smoother spring animations.
Öppna Moodboard →You can now set per-side margins, see a title band in the preview and drag-pan the canvas freely.
Öppna Moodboard →Moodboard settings now open as a bottom sheet on mobile — easier to reach with the thumb.
Öppna Moodboard →Fixed a layout bug where the Export button overlapped the menu on tablet-sized screens.
Efter lokal ljudextraktion skickades ljudet base64-kodat genom en serverfunktion, vilket buffrade hela filen och slog i serverns tidsgräns mot ElevenLabs. Nu strömmas ljudet istället via multipart till samma route som molnfallbacken — snabbare, mindre overhead och inga fler timeouts på medellånga filer.
Till Transkribera →The mini visualization in the Workflow section on the home page now renders correctly.
Till startsidan →Moodboard got bento-grid layouts, in-place cropping, gradients and a transparent background option.
Öppna Moodboard →Moodboard got keyboard shortcuts, a drop-overlay when dragging files and a fix for an SSR hydration warning.
Öppna Moodboard →Moodboard got more layout presets, a title field, JPG export and smarter background color picking.
Öppna Moodboard →Moodboard got a centered hero and now uses the global Export button in the header.
Öppna Moodboard →New tool: drag images onto a canvas, pin notes, sample colors and export a polished moodboard as PNG, JPG or PDF — all locally in your browser.
Öppna Moodboard →Korten för "Släpp in filen" och "Ladda ner" visade tidigare innehållet halvt utanför rutan eller för dimmat när de inte var aktiva. Nu hålls fil-kortet och resultat-kortet alltid centrerade och fullt läsbara, så du ser vad varje steg gör även när auto-cyklingen är på ett annat steg.
Öppna startsidan →The before/after example buttons on the home page now have a clear focus ring and full keyboard control.
Öppna startsidan →Switching examples in the before/after module on the home page now uses a smooth crossfade.
Öppna startsidan →The /konvertera index page got a clearer layout with grouped format pairs and clearer headings.
Öppna konverteraren →I steg 1 låg "Drop zone"-etiketten ovanpå filkortet när det landade — den tonas nu ut mjukt så texten inte krockar. I steg 3 trängdes ikon, filnamn och −85 %-badgen ihop på smala skärmar; layouten är nu nowrap med rätt avstånd och badgen sitter snyggt ytterst till höger.
Öppna startsidan →The home page now lists updated target audiences — marketers, creators, small businesses and everyday users.
Öppna startsidan →The before/after comparison on the home page is clearer with bigger numbers and a cleaner layout.
Öppna startsidan →The hero video on the home page is now lazy-loaded — saves bandwidth on first paint.
Öppna startsidan →Static assets now have longer cache headers so repeat visits are much faster.
Cleaned up font loading — only the weights actually used are downloaded.
Heavy tools (ffmpeg, AI models, PDF engine) are now lazy-loaded so the first paint is much faster.
The hero image on the home page is now prioritized so the first paint feels instant.
Öppna startsidan →Re-encoded the before video on the home page — 90 % smaller without visible quality loss.
Öppna startsidan →Re-compressed images across the site for a faster first paint.
Moved the savings counter and recent files to a cleaner spot in the header.
Öppna inställningar →Added a fresh before/after clip on the home page that shows the savings live.
Öppna startsidan →The supported-formats section on the home page got a cleaner, more readable layout.
Öppna startsidan →The /brandkit color tool now uses the global Export button.
Öppna Färger →The markdown tool now uses the global Export button.
Öppna Markdown →The PDF watermark tool now uses the global Export button.
Öppna Vattenstämpel →The /ljud audio tool no longer reports false out-of-memory errors on long files.
Öppna Ljud →The PDF compression tool now uses the global Export button.
Öppna PDF →The metadata tool now uses the global Export button.
Öppna Metadata →The GIF tool now uses the global Export button.
Öppna GIF →Removed duplicate download buttons from several tools — the global Export button is the single source of truth.
QR code export is now consolidated into the global Export button.
Öppna QR-verktyget →The remove-background tool now exposes the global Export button.
Öppna Ta bort bakgrund →Files you process in one tool now carry over to the next tool — no re-upload needed.
Fixed a bug where clicking the Export button a second time did not reopen the popover on desktop.
You can now swipe files away in more file lists on mobile, across more tools and formats.
Testa på PDF →Settings popovers now open as bottom sheets on mobile — easier to reach and read.
The mobile header is cleaner — settings have moved into the menu instead of cluttering the top bar.
Öppna inställningar →The global Export button now sits at the bottom on mobile so it is always reachable with the thumb.
You can now drag HEIC files from iPhone straight into /bild — they are converted and compressed in the same step.
Öppna bildverktyget →Tightened SEO titles to match high-intent keywords, added longer copy and HowTo sections on many tools.
Testa PDF-verktyget →The global Export button now shows clearer per-format progress and a combined total.
Seven main tools got unique OG share images instead of the generic Krymp default.
Improved color contrast across the site and expanded the Swish QR guide with more examples and FAQ.
Läs om Swish-QR →You can now download all compressed PDFs as a single ZIP from the global Export button.
Öppna PDF →The AI upscaler now plugs into the global Export button — same popover, progress and cancel behavior.
Öppna Uppskala →Four more tools — GIF, audio, video and convert — now use the global Export button.
Öppna Video →Three more tools now use the global Export button.
Öppna Favicon →The remove-background tool's Export button has moved to the header for consistency with the other tools.
Öppna Bakgrund →Every image tool now shows a hint that you can paste an image with ⌘V.
Testa ⌘V →Heavy exports now show a confirmation step so you don't trigger them by accident.
Testa exporten →The Export popover now mirrors the same per-file status as the result list — no more divergence.
Öppna Krymp →If the actual result diverges from the settings in step 1, we now show a clear warning.
Öppna Krymp →Each image now previews exactly which output format it will become — before you upload.
Öppna Krymp →Press Cancel during an ongoing export, and both its progress and the button state will update instantly. The button changes to 'Cancelling…' and the progress bar clears. The abort signal follows the entire chain (upscaling, format conversion, and ZIP packing), so the job stops at the next checkpoint instead of continuing in the background. Applies to both SoMe and Krymp.
Testa exporten →The same export button found in the SoMe header is now also available on Image/Krymp. You'll see live progress as images are compressed, can change formats directly from the popover (which automatically re-runs compression), and download everything as a ZIP when done — without needing to scroll to the bottom button.
Öppna Krymp →Previously, AI upscaling ran once per crop, meaning an image with 25 channel formats resulted in 25 model calls. Now, the source image is upscaled ONCE (or skipped entirely if already large enough), then all channel formats are re-cropped at the target resolution using standard canvas. Format conversion (JPEG/PNG/WebP) also runs in parallel with up to 4 jobs. The result: same quality, a fraction of the time.
Öppna SoMe →Previously, it said '41 formats per image' even though images and videos generate different numbers of variants (25 image formats and 16 video formats, respectively). Now, the counter shows '25 per image · 16 per video' before you upload, and adjusts to your files: only images uploaded? Then it says '25 per image'. No more misunderstandings about what's actually produced.
Öppna SoMe →Cleaned up shortcuts to follow Gmail/Linear conventions. Image moved from g b to g i (Image), and new jumps added: g s (SoMe), g f (Favicon), and g c (Changelog). Also added , (comma) to open Settings directly — the Mac convention — without lifting hands from the keyboard.
Se alla genvägar →The savings chip ('X MB saved') and the floating 'Press ? for shortcuts' button in the bottom corner are gone. Instead, a neat cogwheel sits next to the export button at the top, at the same height. All content has moved under Settings: your savings (session + total) and the entire shortcut list are now gathered there.
Öppna inställningar →Removed the 'Pro' label from JPEG, PNG, and WebP — everything on Krymp is free. Select your format directly in the export panel, and all images will be re-encoded locally in the browser before packing. After running, we decode the first output and show a verification line ('✓ Verified: 2160×2160 · JPEG') so you can see in black and white that upscaling and format change actually occurred.
Öppna SoMe →The sticky export button on SoMe was previously in the lower-right corner, hidden behind the shortcut pill. Now it's integrated next to the main menu (by the PWA/savings chips) so it's always in the same place and doesn't conflict with anything else.
Öppna SoMe →Added the most common web resolutions under the Web channel: hero 1920×800 and 1440×900, mobile-hero 1080×1920, blog 1200×800, image card 800×600, product 1000×1000, square 1080×1080, thumbnails 300×300 and 400×400, background 1920×1200, and banner 1600×500 and 1920×600. Description and FAQ updated with correct info on what the service actually does (smart cropping with focal point — no safe zone overlays).
Öppna SoMe →Replaced the old download bar with a consolidated export panel inspired by shots.so. Choose format and quality (1×/2×/4×) in one place, see estimated time and size before running, and get live progress + Cancel directly in the same button. Heavy jobs are confirmed inline instead of in a separate popup.
Testa SoMe →Four new guides that show how to hit exactly 100 KB or 1 MB for JPG (MozJPEG) and PNG (oxipng + palette). Linked directly from /komprimera/jpg and /komprimera/png as well as from the guides page.
Se guiderna →Updated the title, description, and heading on /bild to lead with 'compress image' and also cover 'reduce image size' and 'downsize images' — to rank better on Google for what people actually search for.
Öppna →Trimmed page titles on several popular pages (Krymp image, Convert, guides, alternative pages, etc.) so they stay under Google's 60-character limit and aren't truncated in search results.
Tell Krymp how large the file can be — e.g., 100 KB, 1 MB, or 2 MB — and we'll automatically choose the quality for you. Works for both Krymp image and Convert (where you can convert + shrink in one step). Quick selectors for 100 KB, 500 KB, 1 MB, 2 MB, and 5 MB, plus a free text field for custom sizes.
Testa nu →New slider icon on each file in the results list: drag the quality slider and instantly see how the image is affected in a before/after view. Once you find the right level, click 'Apply this quality' to replace the result with full resolution at the chosen quality.
Testa nu →You can now paste an image or file directly from the clipboard into all our tools — perfect for screenshots (Cmd+Shift+4 on Mac, Win+Shift+S in Windows) or a copied file from Finder/Explorer. Unnamed screenshots automatically get a timestamped filename. On desktop, a small tip appears in the dropzone.
New discreet feedback button in the footer, a small emoji prompt after successful download (max 1 time/30 days), and a one-time question after the third download for returning users. The form has 4 emojis + a free text line + optional email, and responses go directly to the editor.
Läs mer →All buttons and links — including icon buttons for upload and download — now get clear visual tap feedback (slight scale + opacity) when pressed on mobile and tablet. At the same time, we've globally ensured a minimum 44×44 px hit area for interactive elements on touch devices, according to Apple's and Google's guidelines. The unsightly default-blue tap highlight has been removed.
The bottom sticky action bar (download CTA) now gets .pb-safe so it lands above the iPhone's home indicator and notch instead of being covered by it.
The table on format-test-2026 is now wrapped with .table-scroll so it can be scrolled horizontally without the category column being compressed. The first column (Category) is sticky on mobile so you always see which image type each row refers to even when scrolling horizontally among formats.
Se jämförelsen →På touch-enheter ersätts den streckade drag-och-släpp-ytan med en stor primär "Välj från galleri"-knapp och tydlig "flera bilder samtidigt"-hint. Sekundära knappar för Ta foto (bakkamera) och Filer ligger som 2-kolumns grid under. Galleri-pickern triggar nu accept="image/*" så Android öppnar systemets fotogalleri istället för filhanteraren. Desktop-läget med drag & drop är oförändrat.
Inputs now have a minimum 16px font-size on touch devices (no more iOS auto-zoom when focusing a field). Buttons and links get visible tap-feedback with a scale-animation on touchscreens. Display typography is clamped with `clamp()` below 640px to prevent large headings from overflowing. Sticky elements can now use `.pb-safe`/`.pt-safe` for iPhone notches. Scroll-padding ensures anchor links land below the sticky nav instead of behind it. Icon buttons are guaranteed a minimum 44x44px tap-target on mobile.
Global `overflow-x: clip` + `max-width: 100vw` on `html`/`body` prevents horizontal scroll previously caused by hero illustration overflow. Heavy hero animations (CTA rainbow-conic, breathing glow, gloss streak, aurora-drift, hero-float) are disabled below 640px for faster initial rendering and fewer dropped frames on mobile. React DOM warnings for `imagesrcset`/`imagesizes` are fixed (camelCase per React 19).
Three new entity-bound landing pages expand GEO coverage: image for Instagram (feed/story/reels), image for Facebook Marketplace (with GPS stripping), and product image for WooCommerce (WebP export, batch). The same JSON-LD package as other use-cases — HowTo, Dataset/variableMeasured, FAQPage, and `dateModified` — allows LLMs to directly cite measurable claims.
Bild för Instagram →New article at /format-test-2026 features an open table of relative file sizes for PNG, JPG q85, WebP q80, and AVIF q60 across five image categories — plus methodology, conclusions, and citation block. Powered by Dataset + TechArticle JSON-LD allowing Perplexity, ChatGPT Search, and AI Overviews to cite the figures directly.
Format-test 2026 →All use-case pages now include measurable claims (Dataset/variableMeasured), a fixed 'Last updated' date, and name the underlying codecs (MozJPEG, libwebp, FFmpeg-WASM, pdf.js, etc.) as HowTo.tool. `llms-full.txt` browses the same data, providing LLMs with citable facts without guessing.
Exempel: Hemnet →Use-case pages now display an 'In Numbers' block with measurable, citable claims (e.g., 'Avg. JPG compression: 78%', 'Data to server: 0 bytes'), also emitted as Dataset/variableMeasured in JSON-LD. Each page has a 'Last updated' date and `dateModified` in structured data for freshness. Four new use-case pages: image for LinkedIn, CV, Hemnet, and Tradera.
Bild för LinkedIn →Each guide now has a clear direct answer (H2 question + 40-60 word paragraph) that AI Overviews and Perplexity can extract directly. `/vad-ar/*` pages now feature a Wikipedia-style TL;DR block. Tinypng alternative, Smallpdf free, and iLoveIMG Swedish now have new direct comparison tables with concrete figures (price, file limits, data locality). `llms.txt` has enhanced key facts and `dateModified` per section. The footer always links to About, Security, and Data Privacy.
Guider →All guides on `/guider` now feature a visible FAQ section at the bottom with three concrete answers — and the same questions are published as FAQPage structured data, allowing Google and AI assistants to cite answers directly in search results.
Guider →Upon completion of a download on the SoMe page, a clear summary now appears: how many images were upscaled and how many potentially reverted to original. No guesswork — you immediately see what happened.
SoMe →The confirmation box on the SoMe page now shows a clear summary before heavy jobs start: approximate time (e.g., 1-3 minutes locally), number of images, and the size of the ZIP compared to 1x. Buttons are now labeled 'Cancel' and 'Continue and upscale,' and you're reminded that the process can always be interrupted.
SoMe →When upscaling images on the SoMe page, you now see a visible panel with both steps (X of Y), a progress bar, and estimated time remaining ticking in real-time. The 'Cancel' button is located within the panel and in the bottom bar — so you can stop the process instantly without searching.
SoMe →Before hitting 'Download' on the SoMe page, you now see an approximate time and the ZIP size for 2x and 4x — so you know what you're choosing. The confirmation dialog for heavy jobs shows the same estimate compared to 1x.
SoMe →The dropdown note for 2x/4x on the SoMe page is now even clearer: it explains that all upscaling occurs directly in the browser and that no images leave your device or are sent to a server.
SoMe →Each scale (1x/2x/4x) now shows a brief explanation and how many images it applies to, along with a truthful note that upscaling occurs locally in your browser. Heavy jobs (4x or more than 20 images) are confirmed in a dialog before starting, and during the process, you see a percentage + an estimated time remaining and can cancel at any time.
SoMe →2x and 4x in the SoMe download use the same local upscaling engine as `/upscale`. Images never leave your device — no cloud processing needed.
SoMe →The app now automatically adapts animations for users who prefer less motion in their operating system. All sweeping effects, pulsations, and stepwise transitions are toned down or turned off immediately, while all content remains fully visible and usable.
The mobile menu (hamburger) now has the same glass-pill effect as the main desktop menu. Tapping a link reveals a soft background expanding behind it, and the active page glows with a warm brand-glow for clearer navigation.
The download menu on SoMe now clearly states that 2x and 4x use Lovable Cloud for upscaling — images are encrypted, processed instantly, and deleted afterward. You should never have to guess where your files end up.
SoMe →Hover and active states in the main menu now feature softer, more lively animations. When you hover over a link, a glass-pill background emerges with a soft fade and scale, and a small brand-colored dot rises from below. The active page also has a soft radial glow around it for clearer navigation.
Previously, there were two download locations (top and fixed bottom button) plus a separate 1x/2x/4x selector. Now there's just one button: 'Download' at the bottom. Clicking it opens a menu where you choose 1x (ZIP), 2x (ZIP), or 4x (ZIP) — upscaling happens immediately upon selection. Faster and fewer clicks.
Öppna SoMe →Next to the ZIP button on /some, there's now a 1x/2x/4x selector. If you choose 2x or 4x, all finished channel-adapted images will be AI upscaled (same engine as /upscale, locally when possible — otherwise cloud) just before the zip is created. Filenames will get an @2x/@4x suffix so you immediately see which version you've downloaded. Videos are unaffected. If an individual image fails, the original size will be downloaded instead, ensuring the zip is always complete.
Öppna SoMe →Added three new channels to /some: Spotify (Canvas 9:16, playlist 3000x3000, podcast cover, and artist header), Apple Music (3000x3000 album cover, playlist, and 16:9 shelf banner), and Threads (1:1, 4:5, 1.91:1, and 9:16 story). For web, new additions include hero 16:9 (1920x1080) and 21:9 (2560x1080), blog featured 1200x675, thumbnail 800x800, card 4:5, mobile 4:3, Twitter card 1200x628, and app icon 512x512.
Öppna SoMe-formaten →On narrow screens, long prompt lines in the preview could extend beyond the viewport, intensity buttons could be too wide, and result card buttons might get cramped or disappear. Now, prompt text always wraps neatly, intensity controls use more compact text on phones, and all action buttons in result cards wrap and show icons with text-only on larger screens.
Öppna Beskriv →Styles (Photorealistic, Watercolor…), purposes, framings, lighting, moods, generator tips, and variant badges (Faithful, New angle…) are now fetched from translation files instead of being hardcoded in Swedish. If you switch your language to English, the entire /beskriv interface will consistently be in English, including the tooltips explaining each option.
Öppna Beskriv →In English, some labels on /beskriv were still displayed in Swedish — including the entire Midjourney fine-tuning section (Version, Stylize, Avoid (--no), Add --style raw), the quickstart box, the prompt language hint, and the idea mode's aspect ratio/example buttons. Added missing keys for both languages so the UI is consistent in Swedish and English.
Öppna Beskriv →The influence pills on /beskriv can now be focused with the keyboard, have a clear focus ring, are correctly read by screen readers (\"Influence: strong — text strongly controls\"), and the tooltip opens on hover, focus, and mobile tap. Slightly larger text on phones for easier reading, same compact size on desktop.
Öppna Beskriv →Once the local image AI finished loading, it entered a silent initialization phase (preparing the model in the graphics card) that could take several seconds without the bar moving — many thought the app had frozen. Now, the bar slowly creeps towards 99% during this time, and the text says \"Initializing model (may take a moment)…\" so you see that something is actually happening.
Öppna Beskriv →Directly below the Step 2 controls on /beskriv is now an always visible influence row with percentage, colored level badge (None/Low/Moderate/Strong/Extreme), and top-3 axes. It updates in real-time as you change style, purpose, framing, lighting, mood, or drag the intensity sliders — so you see how far from the original the generation will land without opening the prompt preview.
Öppna Beskriv →Hover over (or long-press) each small colored influence pill on /beskriv to see exactly what the level means — from \"No influence, AI decides itself\" to \"Extreme influence, complete change replacing the original medium\". The total meter in the prompt preview also has a tooltip explaining how the percentage is calculated (style and purpose weigh double).
Öppna Beskriv →Next to each option on /beskriv (Style, Purpose, Framing, Light, Mood, Midjourney fine-tuning) there's now a small colored pill — None, Low, Moderate, Strong, or Extreme influence — which updates instantly when you click or drag the intensity. At the bottom of the prompt preview, you'll see a total meter with a percentage and the top-3 axes that most drive the result, so you know if the generation will be close to the original or diverge significantly.
Öppna Beskriv →When you pressed Cancel (or changed your idea), the card disappeared immediately from the list, but the local idea AI still completed its strict second attempt in the background — wasting battery and risking old results overwriting new ones. Now, a cancel signal is sent all the way down to the model execution, and both the first and second attempts are stopped instantly, without any late updates reaching the UI.
Öppna Beskriv →When you run the same idea again with the same settings in local mode, we now skip the entire model execution and reuse the latest successful JSON response directly. The cache is saved in the browser (up to 50 latest) and persists after reloading, making retries, small adjustments, and toggling back and forth instantaneous instead of waiting for a new inference.
Öppna Beskriv →The local idea model (Qwen2.5) sometimes responded with ```json code blocks, smart quotes, single quotes, or a trailing comma — and previously, we lost the entire response and fell back to the template. Now, these common LLM habits are automatically cleaned before parsing, so more responses pass on the first attempt.
Öppna Beskriv →When the local idea model (Qwen2.5) couldn't run — the model failed to load, crashed, responded in the wrong format, or missed some roles — the same template result was previously shown without you knowing why it didn't feel like the AI you chose. Now, a clear yellow banner appears at the top of the result card with the reason (\"model could not be loaded\", \"wrong format\", etc.) and a \"Try again\" button that re-runs just that card directly — without you having to retype the idea.
Öppna Beskriv →Previously, only the cloud reacted to style and purpose intensity on /beskriv — the local engine (Florence-2 and Qwen idea) ignored the sliders, and results looked the same regardless of your choice. Now, the same recipe (pickIngredients) is woven into the local prompts, and the local idea AI receives the same guiding instructions as the cloud via buildStyleInstruction/buildPurposeInstruction. Subtle feels subtle, Extreme dominates — in both modes.
Öppna Beskriv →We've improved the onboarding experience for users interacting with local AI for the first time, making the process smoother and more intuitive.
Öppna Beskriv →Sticky-baren på /beskriv visar nu som standard bara progressbaren och en kort fas-text (t.ex. "Laddar modell…"). Device, bytes, nedladdad storlek och förfluten tid döljer sig bakom en "Avancerat ▼"-expand — så vanliga användare slipper överväldigas av WebGPU, 270 MB och timing-siffror, medan teknik-intresserade fortfarande kan kika på detaljerna.
Öppna Beskriv →Om den lokala AI-modellen på /beskriv inte kan laddas ner eller initieras visas nu en "Kör i molnet istället"-knapp direkt i sticky-baren. Ett klick växlar läge till moln och kör automatiskt om filerna som fastnat — du slipper läsa felmeddelandet och starta om manuellt.
Öppna Beskriv →Previously, there could be a brief delay before anything appeared on /beskriv when you dropped an image in local mode — the local AI engine couldn't report progress yet. Now, a pulsating \"Preparing analysis…\" is shown immediately in the sticky bar the moment you drop the file, so you always get visual confirmation that something is happening.
Öppna Beskriv →The large \"Local AI Model\" card has been removed from step 1 on /beskriv. Instead, a narrow sticky bar appears at the bottom of the screen when your local model is loading or warming up. It doesn't shift the layout, can be dismissed, and disappears on its own when the model is ready. Less clutter, same information.
Öppna Beskriv →When working in local mode on /beskriv and the model is loading for the first time, an explanatory line now appears above the progress bar: \"Loading your private image understanding model for the first time (~270 MB). Saved in browser – next time it will be lightning fast.\" This directly answers three common questions: Why is it loading? How long will it take? Will it always be this way?
Öppna Beskriv →Previously, the \"Local AI Model\" card appeared on /beskriv as soon as you selected Local in step 1 — even before you dropped any file. Now, it only appears when you have added at least one file, so it's only visible when the model is actually loading or running. Less visual clutter in empty state.
Öppna Beskriv →Skapa-knappen på /beskriv "Från idé" aktiveras nu först när du skrivit minst 8 tecken (tidigare 1). Om knappen är inaktiv visas en tooltip: "Skriv minst en mening om vad du vill se". Dessutom rensas emojis och skräp-symboler bort på klienten innan någon idé skickas till molnet — så bara meningsfull text går vidare till AI-motorn.
Öppna Beskriv →The in-browser AI that powers the "Local AI" mode on /beskriv "From Idea" is upgraded from SmolLM2-360M to Qwen2.5-0.5B-Instruct. It maintains the JSON format significantly better and runs fast on WebGPU. If the first response isn't valid JSON, we now make a second attempt with a lower temperature (more strict) before falling back to the template engine—fewer silent fallbacks. The status box now shows two clear steps: "Step 1/2 · Loading model" and "Step 2/2 · Writing prompts (attempt 1/2, creative → strict)" instead of a vague progress bar.
Öppna Beskriv →Flera UX-fixar på /beskriv när du jobbar från text: (1) chip-rad för bildförhållande direkt ovanför textarean — 1:1, 3:4, 4:3, 16:9, 9:16, 21:9 — så du själv styr aspect istället för att alltid landa på 1:1. (2) En kompakt "Vi tolkade din idé som: motiv = …, miljö = …, stämning = …"-panel visas efter generering, med en Justera-knapp som öppnar samma berikningsförslag som "Bygg ut min idé". (3) Exempel-chips skriver inte längre över din text — de fogas in vid markörpositionen. (4) Dina 5 senaste idéer ligger kvar i sessionen som klickbara chips för snabb återanvändning. (5) Under pågående generering syns en Avbryt-knapp, och om du börjar ändra texten avbryts det pågående jobbet automatiskt.
Öppna Beskriv →The local template engine on /beskriv "From Idea" now considers all your choices—style intensity 1–4, purpose intensity 1–4, crop, light, and mood—and weaves them into all four prompts, just like the cloud mode. At higher style intensities, the style's "avoid" list is also added to the creative prompt. Result: distinctly different outcomes when you adjust sliders, even when working completely offline without AI credits.
Öppna Beskriv →På /beskriv "Från idé" finns nu en "Bygg ut min idé"-knapp som låter Lovable AI föreslå tre berikade versioner av din text — välj en med ett klick så ersätter den textarean. Samtidigt analyseras vad idén SAKNAR (tid på dygnet, ljus, stämning, palett, miljö, inramning, stil) och visar 3 färdiga chips per saknad axel som du kan appenda direkt med ett klick — t.ex. "+ blå timme", "+ varm jord", "+ extrem närbild". Snabbaste sättet att gå från en luddig idé till en rik prompt.
Öppna Beskriv →The Midjourney fine-tune on /beskriv now includes v8.1 (latest) as an option, and Midjourney's default uses v8.1 instead of v6.1. v7, v6.1, and Niji 6 remain available for backward compatibility.
Öppna Beskriv →När du genererar prompts från en textidé på /beskriv kör vi nu först en synthesizer som expanderar din idé till samma strukturerade analys-JSON som bildflödet använder — motiv, miljö, ljusriktning + tid på dygnet, kameralins, palett med 5–8 hex-färger, material, stil-fingeravtryck. Sedan komponerar prompt-motorn mot den analysen i stället för att gissa fritt. Resultat: fyra mer distinkta, konkretare prompts som faktiskt skiljer sig åt per roll, en "Vi tolkade din idé som …"-panel under kortet, och Remix fungerar nu även från idé-prompts. Din originaltext är fortfarande sanningens källa — synthesizern får aldrig motsäga den.
Öppna Beskriv →Every control on /beskriv step 2 now has a small colored badge—None · Low · Moderate · Strong · Extreme—showing exactly how much that specific choice will shift the result. Style + intensity, Purpose, Framing, Light, Mood, and Midjourney-extras each get one. At the top of the prompt recipe, there's a "Total Output Impact" meter with a percentage and a top-3 list of the most influential factors. Updates live as you click.
Öppna Beskriv →When you change the target generator on /beskriv, Style, intensity, Framing, Light, Mood (and for Midjourney, --stylize and --v) now automatically adapt to what that model excels at—e.g., "Cinematic + cinematic light + epic + v6.1" for Midjourney, "Photoreal + studio light + close-up" for SD/Flux, "Photoreal + golden hour + calm" for Nano Banana. Active chips you've selected are never touched. An undo toast appears with each change, and you can click "Reset defaults" under the generator selector at any time.
Öppna Beskriv →On /beskriv step 2, there's now an expandable "More Control" section with three extra chip rows: Framing (extreme close-up → ultra wide), Light (golden/blue hour, harsh midday, neon night, chiaroscuro, etc.), and Mood (calm, ominous, epic, dreamy…). All are optional—auto is default. When you select a value, it overshadows the image analysis on that axis in ALL four prompts, so the same image with the same style/purpose yields noticeably different results when you, for instance, switch from auto to "blue hour + extreme close-up + ominous." Active selections appear live in the prompt recipe and the copy button.
Öppna Beskriv →On /beskriv, Purpose now defines four different roles per prompt, not just extra tone directives. Recreate yields faithful + detail zoom + material zoom + light zoom (all anchor: fidelity), Variation gives anchors + new angle + new light + new palette, Style Change leads each prompt with the style signature instead of the original medium, and Inspiration provides four conceptually free takes (loose interpretation, alt-concept, alt-medium, alt-era). The result cards show the actual role of each prompt. Previously, variations were often nearly identical regardless of purpose.
Öppna Beskriv →Style + intensity are now fully integrated with the AI on /beskriv. Strong and Extreme force the style's signature ingredients (e.g., anamorphic, halation, screen-print halftone) into ALL four prompts, add a "DO NOT" list per style, and resolve conflicts when Purpose=Recreate meets a powerful style (subject preserved, medium changed). A post-check mechanically ensures ingredients are included even if the model is careless. Previously, "Cinematic + Extreme" could return a plain DSLR image.
Öppna Beskriv →On /beskriv step 2, there's now an expandable "Prompt Recipe Preview" showing exactly which ingredients Style + Purpose and intensity add to the prompt. New ingredients flash in brand color as you adjust sliders, removed ones are listed separately, and Midjourney suffixes appear live. A button is available to copy the entire recipe to the clipboard.
Öppna Beskriv →Fixed several small bugs in the top menu: 1) "Tools" now shows a chevron directly from the first frame (previously, it appeared only after dropdown code loaded), 2) hover effect is subdued so it no longer looks like an active state (you said "About" looked selected just because the mouse was there), 3) "About" now highlights on sub-routes (/about/…), 4) the active dot is slightly clearer and correctly positioned even on Tools, 5) aria-current and focus-ring added for better keyboard/screen reader accessibility.
On /beskriv → From Idea, the "Local (AI on device)" engine is now the default instead of Cloud. Anonymity is important in Krymp—your idea text does not leave your device unless you choose Cloud yourself. Previous selections are saved and respected as usual.
Öppna Beskriv →Drag-between-tools now also works for Upscale, Background, and Watermark—when a result is ready, you can drag the entire row to a next-step-pill (e.g., "Krymp"), and we'll open the next tool with the file already queued. Previously, this only worked from Image.
Det lilla gröna pillet i headern som visar hur många MB du sparat denna session har fått ett kugghjul i kanten. Tidigare var det inte uppenbart att man kunde klicka — nu syns det direkt att det leder till inställningar (där du ser detaljerad statistik och kan rensa). Kugghjulet snurrar 45° vid hover så det känns levande.
On Image, you can now press "Compress again" on each result (or the entire batch) and choose between two modes: 1) Squeeze harder iteratively—uses the previous result as new input for even more compression, 2) Rerun from original—reloads the original file from scratch if you want to change level/format. Each run is saved as P1, P2, P3… in a pass history, allowing you to download any version or revert. We automatically roll back if an iterative pass compresses less than 2% to avoid unnecessary quality loss. Max 5 passes per file.
Pröva på Bild →Three improvements to Describe: 1) a quick start area at the top where you can drop an image directly without scrolling through Steps 1–2, 2) each generated prompt can now be edited inline (pencil icon → save with ⌘V) — copy and Open in… use your edited version, 3) when Midjourney is selected, you get extra controls in Step 2: version (v6.1 / v7 / Niji 6), --style raw, stylize-slider, and --no-field. Suffixes are appended directly to all MJ prompts without consuming new AI calls.
Öppna Beskriv →The active tab in the pill menu now has a small brand-colored dot beneath it for instant visibility—in addition to the white pill. Hover provides a soft background (not just text color), and transitions are consistent across Images, Video, SoMe, Tools, and About. A thin separator between Tools and About clarifies the grouping. The same dot also appears on mobile shortcuts.
New "Media Library" tab in the uploader: paste a page, a WordPress library (we read public REST endpoints), a whole list of links — or just a link to an image. We show a preview, you check which ones you want, and we retrieve four in parallel with progress per file. All anonymous: no accounts, no saved links, private Dropbox/Drive folders are intentionally not supported.
Pröva på Bild →On Image, Describe, Watermark, Background, Upscale, Metadata, and Colors, you can now select the "From URL" tab and paste a direct link to an image. We'll fetch it for you (max 20 MB) and queue it directly—no need to download and drag it in yourself.
Pröva på Bild →Once an image is compressed, you can now drag the result row directly to one of the "Next Steps" buttons below the list. The file automatically transfers to the next tool—no re-upload needed. Buttons light up and gently vibrate when you drag to indicate available drop zones.
Each prompt variant now has an "Open in…" button that takes you directly to Midjourney, ChatGPT, Gemini, or fal.ai with the prompt pre-filled where possible—no more copying and pasting. When the first image is ready, we also scroll down to the result automatically. If the local AI fails an image, you get a button to directly try the same image in the cloud instead. And a redundant footnote about language selection has been cleaned up.
Öppna Beskriv →The Tools menu now opens on the first click instead of requiring two. The mobile menu slides in smoothly even the first time you tap Menu. The Menu button toggles between a hamburger and an 'x' to clearly indicate if the sheet is open. The Tools link is now properly visible over the dark star-hero. Duplicate Images/Video/SoMe entries in the Tools dropdown have been removed.
On Background, Watermark, and Upscale, a draggable slider between original and result now appears directly on the result card — no need to click a "preview" button first. Drag the center to see exactly what the tool did. For Background, you also get a checkerboard pattern behind it for clear transparency.
Testa på Bakgrund →On mobile, a "📷 Take Photo" button now appears next to the regular upload on Image, Describe, and Metadata. Tap it → the camera opens → the picture goes directly into the queue and is processed. Skip the detour via the photo library.
Pröva på mobil →När en fil misslyckas visar vi nu ett begripligt felmeddelande istället för kryptisk teknisk text. På Bild-verktyget får du dessutom en "Försök med lägre kvalitet"-knapp som kör om filen ett snäpp hårdare komprimerat — perfekt när webbläsarens minne tog slut på stora filer.
Pröva på Bild →Power users can now navigate Krymp without lifting their hands from the keyboard. Press ? anywhere to open a list of all shortcuts. Gmail-style "g" followed by a letter takes you directly to the tools: g b → Image, g k → Shrink, g v → Video, g p → PDF.
On the Image tool, there's now a presets menu above the settings. Choose from four factory presets (Web Image, Email, Instagram, Original Quality) or save your own combinations for quick reuse. Everything is saved locally in your browser.
Testa presets →The homepage now features a "Recent Files" panel with your 5 most recently processed files. Click to jump directly back to the correct tool. Only file names and settings are saved — the files themselves are never stored — and everything resides locally in your browser.
Se startsidan →Under varje resultatkort finns nu en liten utfällbar panel som i klartext beskriver vad som faktiskt hände med din fil — skalades den ned, vilket format användes, togs metadata bort, gick något genom AI. Tanken är att du ska kunna förstå Krymp utan att veta vad EXIF eller H.265 betyder. Funkar på Bild, Krymp, Konvertera, PDF, Video, Bakgrund, Uppskala, Vattenstämpel, Favicon, Metadata, Ljud, GIF, Beskriv, Transkribera och Anonymisera.
The homepage now includes a small counter showing how many files you've compressed and how many MB or GB you've saved over time. The counter lives locally in your browser — nothing is sent to us. Once you pass 100 MB saved, a "Share with a colleague" button also appears, letting you easily recommend Krymp to others.
After you've compressed, converted, watermarked, or created something else, a "Next Step" row appears below the result. For example, click "Add Watermark" and the files will automatically transfer to the next tool — no new upload required. Now works on Image, Shrink, Convert, Video, GIF, Audio, Watermark, Upscale, Background, Favicon, Metadata, Describe, and Transcribe.
Krymp can now be installed as an app on phone and computer via a discreet "Install" button in the header (only shown after a few visits). Once you've saved some data, an easy way to refer a colleague also appears — via email, copied link, or your system's share function. No accounts, no subscriptions.
If a single tool misbehaves — for example, a corrupt file or an incomplete engine download — you now get a calm explanation with a "Try Again" button instead of a generic error page. Other parts of the site are unaffected, and no files leave your device.
Each tool now has a compact trust banner directly under the heading, indicating that files stay in your browser, no account is needed, and the service is completely free. The homepage also got a new section showcasing who actually uses Krymp — editorial offices, agencies, government bodies, small businesses, students, and photographers.
Läs om säkerhet →After you've compressed, converted, or processed a video, we now suggest natural next steps — e.g., add a watermark, remove the background, or create a GIF. The file automatically carries over, so you don't have to upload it again.
Prova bildverktyget →We've added eight new guides covering common practical questions: compressing an image to 100kb or 1MB, PDF under 2MB for tax authorities and banks, video under 25MB for email, free background removal, and iPhone-specific tips. All in Swedish, all local in the browser.
Läs guiderna →Video compression could get stuck at 5% and files at 2% because the engine's internal worker file couldn't find its helper files in the build. The worker is now self-sufficient, so the engine starts and processes videos as usual.
Komprimera video →The hero illustration is now served in multiple sizes (800/1280/1600 px) with lower AVIF quality, so mobiles and laptops load a much smaller image and the first impression of the homepage is noticeably faster.
The video engine was getting stuck at 5% and files wouldn’t process past 2% due to a version conflict in the ffmpeg engine. The versions are now locked, so the engine starts and compresses as usual.
Komprimera video →Links like krmp.se/abcd now redirect directly to the target URL. Previously, traffic on krmp.se took a detour via the main domain, causing short links to fail. Other traffic to krmp.se is forwarded to krymp.io, so nothing breaks.
Skapa kort länk →New short links now appear as krmp.se/abcd instead of krymp.io/k/abcd — four characters shorter and easier to read aloud. Existing /k/ links continue to function unchanged.
Skapa kort länk →Previously, /transcribe could hang for up to three minutes when the browser's audio engine didn't start. Now, the engine is checked in 5 seconds, small video and audio files are sent directly to the cloud without local extraction, and in case of an error, a 'Try again via cloud' button instantly re-runs the job.
Till Transkribera →Images in the drag-and-compare box are now served as AVIF and WebP instead of large JPEG/PNG. Saves about 4 MB of data traffic per visit and makes the section render noticeably faster on mobile.
Till startsidan →The status banner for heavy engines is now only mounted when something is actually loading, and the hero headline no longer flickers between language versions on page start. Overall, a faster and calmer first impression.
Till startsidan →The hero illustration now loads as AVIF/WebP with PNG fallback and is preloaded, significantly speeding up page rendering. Google Analytics also loads later and is skipped entirely in data saver or 2G mode.
Till startsidan →Increased contrast for muted text in the hero, file examples, and footer, making it easier to read — benefiting both eye comfort and accessibility.
Till startsidan →New intro highlights Krymp as a free Hasty tool for image, video, audio, text, QR, and more file formats — directly in your browser. The small trust bar under the button has been removed for a cleaner hero.
Till startsidan →Updated homepage texts to emphasize that the service is completely free and accessible directly in the browser with no hidden fees.
Se ändringarna →New section on the homepage explains that the service is completely free — and why it will remain so: powered by Hasty, belief in an open web, no ads or data collection.
Till startsidan →/guider now lists all five guides (HEIC conversion and PDF compression were hidden). Guide content matches current tools — preset names, model descriptions, and PDF levels (Light/Balance/Hard) are corrected. Each guide also has a back button to return directly to the tool you came from.
Öppna guider →A page size limit in the data layer meant /trender only showed about a third of events. We now paginate the entire window, so sessions, runs, top tools, and format savings match reality.
Öppna trender →Runs are now counted as files/jobs that have actually completed (matches admin), the misleading conversion percentage has been removed, and average savings per format now shows correct data for compression and conversion formats.
Öppna trender →AI assistants now receive complete descriptions of Image, GIF, Audio, Transcribe, Upscale, Watermark, Colors, Signature, Characters, Markdown, and Anonymize in llms.txt — and definition and comparison pages are also included in llms.json and llms-full.txt.
Muted micro-labels (index dots, 'network monitor', list separators in footer) are now darker/lighter to meet WCAG AA contrast against the background.
Trimmed titles and descriptions on the homepage, /konvertera, /transkribera, /beskriv, /bakgrund, /brandkit, and /vattenstampel so they aren't cut off in Google search results.
Consolidated page explaining why Krymp is anonymous and free — files stay in your browser, no account needed, and we only save anonymous usage statistics. Link available under the CTA on the homepage and in the footer.
Öppna Säkerhet & integritet →Directly under 'Start shrinking your files' now appears a small line: 100% anonymous · Completely free · No files leave your device. Discreet but where the eye lands after the CTA.
Öppna startsidan →New export format 'CSS roles' provides --color-primary, --color-secondary, --color-accent, --color-background, etc. — ready to paste into your project, for both light and dark themes. Copy or download directly from the export panel.
Öppna Färger →You can now name a palette and save it locally in your browser — load it back with a click, rename it, or delete it. No account needed, everything stays on your device.
Öppna Färger →Add your own colors via HEX, eyedropper, or an empty slot — even after analyzing a website. A new suggestion panel identifies missing roles (accent, secondary, light/dark neutral) and proposes specific colors that complement your existing ones. Click 'Add' or 'Replace' directly on the suggestion.
Öppna Färger →The line now shows protocol and path, e.g., https://www.hasty.se/, so you see the exact address we fetched colors from after any redirects.
Öppna Färger →When analyzing a website, it now shows which host we actually fetched from — e.g. 'hasty.se (apex)' or 'www.hasty.se (www)'. Also follows redirects when Firecrawl is used.
Öppna Färger →Website analysis now fetches the brand's actual primary, secondary, and accent colors via Firecrawl instead of guessing from CSS noise. During analysis, you'll see four steps tick off — 'contacting', 'reading stylesheets', 'identifying colors', 'classifying roles' — and a small label indicates which source responded.
Öppna Färger →Website analysis now only accepts a domain. We automatically add the protocol and try the www variant if the apex domain doesn't respond.
Öppna Färger →The form changes headings and fields based on selected category, bug mode has dedicated fields for 'what did you try to do/what happened/browser', email gets a live checkmark, drafts auto-save, ⌘/Ctrl+Enter sends, mobile CTA is sticky, and confirmation clearly shows next steps.
Öppna Feedback →You can now paste a URL to extract its colors, see the palette classified into primary/secondary/accent/neutral, drag colors into your preferred order — and download a multi-page PDF branded with Hasty and krymp.io.
Öppna Färger →The 'Start shrinking your files here...' button previously landed further down the section as content expanded after scrolling. Now, its position is corrected once the section has fully rendered, ensuring the '02 / Tools' heading always appears at the top.
Öppna startsidan →Added four new subtle file chips reflecting the toolbox: swish.svg (QR), intervju.m4a (transcription), palette.ase (colors), and porträtt@4x.png (AI upscaling).
Öppna startsidan →The 'Start shrinking your files here...' button now features a multi-colored outline (purple → pink → orange → yellow → cyan → indigo) that slowly animates around its edge. Subtle yet vibrant – the button itself remains still.
Öppna startsidan →Body text in muted colors now has higher contrast for better readability. Page titles and descriptions for the PDF, convert, background, upscale, and describe tools have been shortened to display cleanly in Google.
The mobile hero now has more space between the menu and heading, less dead space in the middle, and a softer fade at the bottom edge of the illustration. The previous side gradient is now only shown on wider screens and fades out much smoother, eliminating the harsh edge.
Öppna startsidan →The 'Start shrinking your files here...' button now features a rotating gradient outline, a softly pulsating glow, and a glint that slides across its surface — still calm, but impossible to miss. Clicking takes you directly to the Tools section. On mobile and tablet, the hero's height is condensed to eliminate large dead space between the button and illustration. Behind the title float four small file chips (photo.jpg, screenshot.png, clip.mp4, rapport.pdf) hinting at the product without stealing focus.
Öppna startsidan →The hero is now fine-tuned for mobile and tablet — the illustration scales proportionally so characters are always visible, and text and button get more breathing room. On wide screens, a soft gradient overlays the illustration's left and right edges, eliminating harsh lines, plus a gentle fade into the section below.
Öppna startsidan →The homepage hero features a new, wider illustration that sits as a band along the bottom edge — characters in the center are always visible without cropping. The 'Start shrinking your files here...' button is now a warm yellow with pulsating rings, set against a deeper green background for improved contrast and atmosphere.
Öppna startsidan →On laptops and desktops, the illustration now covers the entire hero area like a wrapper with a soft gradient for readability. On mobile, the illustration is removed for quicker user access. The 'Start shrinking your files' button is white on a dark background for maximum contrast and features pulsating rings that draw immediate attention.
Öppna startsidan →The homepage hero has been rebuilt as a 3-row grid: title, illustration, and CTA button each get their own row and can no longer push each other aside. Result: the 'Start shrinking your files' button is always visible, the illustration sits against the bottom edge without a white gap, and the layout scales beautifully from mobile to ultrawide.
Öppna startsidan →The illustration now covers the entire bottom of the screen without a gap to the next section, breathes softly with a floating animation, and has more breathing room against the intro text. The 'Start shrinking your files' scroll button is now in brand green with a glow pulse and scales nicely on all screen sizes.
Öppna startsidan →The main menu now switches to a light, glass-like variant when it's over the hero section and reverts to dark when you scroll further. The scroll button always remains at the bottom of the hero, regardless of screen size, and the illustration has moved a step closer to the intro text.
Öppna startsidan →The homepage has a new hero with an illustration, a dark background all the way up, and a tighter heading. The same fast, local toolbox – now with a bit more character.
Öppna startsidan →The glossary expands with JFIF, TIFF, RAW, MP4, H.264, and Whisper, and three new direct comparisons are published: JFIF vs JPG, TIFF vs PNG, and MP4 vs MOV. All with quick answers and DefinedTerm-structured data so answers can be directly cited by Google AI Overviews, ChatGPT, and Perplexity.
Öppna ordlistan →Two new deep dives: Compress PDF under 2 MB and Convert HEIC to JPG. Additionally, all guides now have HowTo-structured data, allowing steps to be displayed directly in Google search results and cited by AI responses in ChatGPT and Perplexity.
Läs PDF-guiden →Krymp now has six new glossary terms (WebP, AVIF, QR code, Swish-QR, OCR, PDF/A) and four direct image format comparisons: WebP vs JPG, WebP vs AVIF, PNG vs JPG, and HEIC vs JPG. All written with quick answers at the top so ChatGPT, Perplexity, and Google AI Overviews can cite directly.
Se jämförelserna →Krymp now features a glossary at /vad-ar with concise, citable explanations of concepts like image compression, HEIC, EXIF, and lossless vs lossy — written for both humans and AI search engines (ChatGPT, Perplexity, Gemini). Simultaneously, all our alternative comparisons are gathered on the new hub /jamfor.
Öppna ordlistan →Krymp is now verified in Google Search Console and sitemap.xml has been submitted. This provides a better overview of how pages perform in Google search and faster indexing of new tools and guides.
The sender address in /feedback was malformed and blocked by the email service with status 422. Now, both the editorial email to hej@hasty.se and your confirmation copy are sent as intended.
Öppna feedback →The homepage hero is compressed so headline, CTA, proof chips, and live bar are visible immediately without scrolling. Smaller headline clamp, tighter padding, and correct 100svh on mobile prevent the address bar from hiding content.
Siffersektionen på startsidan är ombyggd: gradient-italic på "riktiga bytes.", ett live-chip som tickar upp till veckans totala besparing (eller dina egna), och fyra stat-kort med varsin liten visualisering — progress-stapel från MB till KB, dot-grid över alla verktyg, en cirkulär dial för lokal bearbetning och en överstruken rad för "konton som krävs". Hover lyfter korten med brand-glöd.
You can now submit feedback, report bugs, or request new features directly via a form at /feedback. The email goes to the editorial team, and you receive a confirmation copy. Linked from the About page and footer.
Öppna feedback →The final "SHRINK NOW." section is rebuilt: step-by-step character animation with gradient-italic on "NOW", a spotlight following the mouse, kinetic outline-marquee in the background, a live savings chip at the top (shows your own saved MB/GB if you have data, otherwise weekly total), and a trust row at the bottom — All local · WebAssembly · No accounts · 0 KB uploaded.
The homepage manifest is rebuilt: the title gets a gradient accent on "clear", each statement has a highlighted key phrase with brand-tinted highlight, and to the right, a small live proof — Network: 0 KB ↑ with a pulsating green dot, a counter ticking up to 100/100 completed files, and a crossed-out "Log in" with a "No longer" badge. Numbers and icons lift on hover.
The "Three moves. Done." section is rebuilt into an auto-cycling mini-demo showing real-time conversion: a file slides into the drop zone, shrinks with animated progress and bytes counting down from MB to KB, and lands as a saved file with a −85% badge. Active step is highlighted with brand glow, a pulsating progress rail connects the steps, and hover/click pauses for reading. Respects reduce-motion.
The format band on the homepage is redesigned into three straight rows grouped by Image, Video & Audio, and Documents & Data — pauses on hover and respects reduce-motion, and looks consistent from mobile to ultrawide. SEO shortcuts (PDF under 2 MB, Compress JPG etc.) are moved to a subtly tabbed zone in the footer so they don't compete with main tools, but all links remain available for those seeking them.
Du kan nu generera de fyra prompt-utkasten direkt på din enhet i stället för att skicka idén till molnet. Välj mellan Cloud (AI) som tidigare, Lokalt (mall) — snabb mall-baserad generering utan nedladdning — eller Lokalt (AI på enheten) som kör en liten språkmodell (~380 MB) lokalt via WebGPU/WASM. Idén lämnar aldrig webbläsaren i de lokala lägena.
Öppna /beskriv →Previously, smart-drop always routed files to their 'correct' tool (e.g., an image → /bild) even if you were already on a different tool. Now, the global drop-overlay recognizes when you're on a page that accepts files, letting the active tool handle the drop — no need to restart in the wrong flow.
New endpoint /llms.json mirrors content from /llms-full.txt but in structured JSON — all public pages, keyword guides, quick facts, steps, and FAQ — so AI assistants and LLMs can more easily parse and cite Krymp.
Visa /llms.json →Conversion, compression, use-case, and alternatives pages now have a "Related Guides" section at the bottom, pulling adjacent pages from other groups based on format overlap (e.g., a JPG conversion links to compress-jpg and image-for-Blocket). Helps both visitors and search engines find the next relevant page.
The four prompts (faithful, mood, composition, creative) are now written with four different voices instead of all opening with the same template. Mood is led by light and feeling, composition by the camera, creative gets a freer start — and a diversity check ensures no two prompts start identically or feel like reformulations of each other.
Prova Beskriv →Once an image has been interpreted, you can now write a short instruction (or click a quick selection like Night scene, Winter, or As 3D-render) to generate four new prompts based on the same analysis. The original remains, and you can chain multiple twists.
Prova Beskriv →The AI now looks in more detail at your image: new analysis fields for light direction/quality, camera choice, materials and surfaces, fore-/mid-/background, post-processing (grain, color grade, LUT), and an extended color palette with roles. An extra detail pass reads small text, logos, and edges, and prompts are forced to quote the analysis instead of falling back on generic words. The model is upgraded to Gemini 3.1 Pro Preview for analysis.
Prova Beskriv →You no longer need to upload an image — briefly describe what kind of image you want, and the AI will generate the same four prompt variants (faithful, mood, composition, creative) in your chosen format (Midjourney, ChatGPT Image 2, SD/Flux, Nano Banana, or universal).
Prova Beskriv →AI assistants (ChatGPT, Perplexity, Claude, etc.) now receive a structured list of all new conversion, compression, use-case, and alternative pages — including intro, steps, and FAQ — so they can cite Krymp correctly.
New section at the bottom of the homepage lists the 18 most common conversions and use-cases — making them quicker for visitors to find and helping Google index them.
Till startsidan →Three new pages explain why Krymp is a great Swedish alternative — no upload, no monthly limits, and free batch processing. Good if you're looking for replacements for large image services.
TinyPNG-alternativ →Dedicated pages for concrete Swedish needs: PDF under 2 MB for authorities, images for Blocket, QR code for Swish, compression for WhatsApp and email, and anonymization of screenshots. Everything still happens locally in the browser.
PDF under 2 MB →New dedicated pages for "compress <format>" — each page explains the technique, shows typical savings, and links directly to the correct tool. Everything still happens locally in the browser.
Komprimera JPG →Krymp now has dedicated pages for the most common conversions — HEIC to JPG, PNG to JPG, MP4 to GIF, PDF to JPG, and eight more. Each page explains the format, shows quick facts, and links directly to the correct tool.
HEIC till JPG →Sitemaps now use actual modification dates from the changelog instead of today's date, hreflang points to the clean canonical URL without ?lang parameters, 404 pages are noindexed, and the color scheme follows the browser's UI light/dark mode.
Google Analytics now loads after the page is interactive to avoid blocking the first paint. BreadcrumbList displays three steps on tool and guide pages, and meta descriptions for several key pages have been rewritten with clearer keywords.
The MP4 tab in \"Drag. See the difference.\" now plays an actual Krymp-compressed screen recording side-by-side with the original — 29 MB → 1.1 MB, −96%.
Till startsidan →The PNG tab in \"Drag. See the difference.\" now shows an actual Krymp-compressed screenshot — 1.1 MB → 174 kB, −85%.
Till startsidan →The JPG tab in \"Drag. See the difference.\" now shows an actual Krymp-compressed image instead of a mocked SVG — 8.0 MB → 3.7 MB, −54%.
Till startsidan →Engine startup no longer aborts its wasm download on slow networks, skips double fetching for self-hosted core, and has a more realistic timeout that tries all CDN fallbacks before giving up.
Öppna video →Moved ffmpeg core/wasm out of /public and serve them via a dedicated static middleware with correct MIME types, so Vite stops blocking the worker's dynamic import of the engine.
Download the entire palette as a typographically strong A4 PDF with large color blocks, all formats (HEX, RGB, HSL, CMYK, PMS) and a link to krymp.io in the footer.
Öppna färger →Click the controls icon in a color swatch to see and copy HEX, RGB, HSL, CMYK, and the closest Pantone (PMS) — with an indication of how close the match is.
Öppna färger →Use-case presets (Photo, Illustration, Screenshot, Max everything), sample images to test with, history, and fine-tuning per file with sharpness, noise reduction, and optional output format (PNG/JPEG/WebP).
Öppna uppskalning →Sticky media player with karaoke highlighting, search box with results, editable segments, locally saved speaker names, summary style switching, and new export formats (Notion, Anki, Pretty SRT).
Öppna transkribering →Direct image manipulation, snap-to-grid or free mode, auto-contrast, macros like © {year}, custom saved templates, and per-file overrides for large batches. Plus shortcuts (arrows, R, [ / ]).
Öppna vattenstämpel →Previously, uploaded files would get stuck at 'Processing 2%'. Now, clear phases (fetching engine → processing) and a live progress bar are shown even when ffmpeg is streaming. The default mode has also changed to 'Extract Audio' to match most users' needs.
Öppna ljudverktyget →Added the icon badge above the title and removed the double intro text so the hero pattern is consistent across Krymp.
Step 2 now leads with quick selections (preset) before format choice, step 3 gained a cleaner group heading with a format popover, and the sticky download bar concretely shows saved space in MB and percent.
Öppna konvertera →Centralized metadata, HowTo + Speakable per tool, fact boxes (<dl>), HTML sitemap, IndexNow support, and dataset schema on /api.
New page with quick facts, quotable talking points, and contact information for journalists and editorial teams.
Öppna pressrum →Live status for the public API with latency, latest data generation, and rate limits.
Visa status →Documentation for /api/public/trends/v1/summary with example responses, cURL, and terms of use.
API-docs →Anonymous, aggregated figures on how Swedish users compress images, PDFs, and videos.
Visa trender →Vill du få notiser? Följ trender eller hör av dig på hej@hasty.se.