Image tools

Image Compression has the feature most compression tools are missing: a target file size. Instead of nudging a quality slider and re-exporting until the number is under the limit, you say '200KB' and it runs a binary search — probing quality levels between 0.30 and 0.95, then stepping the dimensions down if quality alone cannot get there, and accepting as soon as it is within a sensible margin of the budget. It is a small thing that removes an entire category of tedium, and it exists because government portals, job applications and university systems still impose byte limits with no guidance on how to meet them.

Two consequences worth knowing. Target mode is inherently lossy, so a PNG in target mode is converted to JPEG — a lossless format cannot be tuned to an arbitrary byte budget. And there are also auto, aggressive and manual-quality modes for when you do not have a specific number to hit and just want the file smaller.

Image Resizer covers the dimension side: exact pixels, a percentage, a longest-side constraint, or named presets, with crop and watermark tabs on the same loaded image so you are not passing a file between three tools. Crop Image is the focused version, with a draggable crop rectangle and aspect-ratio locks for the ratios you actually need.

Format Converter outputs PNG, JPEG, WebP and AVIF. Three of those the canvas can encode directly; AVIF it cannot, at all, in any current browser. So AVIF encoding is done by a WebAssembly encoder that loads on demand — which is why an AVIF export takes noticeably longer than a WebP one, and why AVIF is worth it anyway when the file-size difference matters more than the seconds.

On the input side, the awkward format is HEIC — the default on iPhones, and rejected by a great many upload forms. HEIC files are decoded through a WebAssembly decoder before the conversion runs, so 'photo from my phone that this website will not accept' has a two-step answer that does not involve emailing yourself the photo to force a conversion. The tool also verifies that the encoder actually produced the format you asked for, rather than silently handing back a PNG with a WebP file extension, which is a real failure mode elsewhere.

SVG to PNG rasterises vector artwork at a chosen multiplier, which is how you get a 2x or 3x asset out of an icon. It needs the SVG to declare intrinsic dimensions, and external references inside the SVG will not resolve — both are consequences of rendering through the browser's image pipeline rather than a full SVG renderer. SVG Tools is a different thing entirely: a visual editor for tweaking shapes and colours in an SVG directly.

EXIF Remover and EXIF Viewer handle the invisible payload — camera model, timestamps, and frequently GPS coordinates precise enough to identify a home address. The viewer shows you what is there; the remover strips it by redrawing the pixels, which means a JPEG is re-encoded rather than surgically edited.

Photo Censor covers the visible payload: draw regions and apply blur, pixelation or a solid fill. Because the canvas is flattened on export, the censored pixels are actually gone rather than obscured by a layer that a determined recipient could remove. Watermark Image does the opposite — adds text or an image mark with opacity, rotation, positioning and a tiling mode.

Color Correction adjusts brightness, contrast, saturation, temperature, vibrance and exposure through per-pixel arithmetic rather than CSS filters, specifically so the export matches the preview. That distinction sounds pedantic until you have used a tool where it did not, and exported something that looked nothing like what was on screen.

Favicon Generator renders your source at 16, 32, 48, 180, 192 and 512 pixels, hand-writes a genuine multi-image .ico file, and bundles the whole set with the HTML link tags and a web manifest into a zip — because the annoying part of favicons was never the resizing, it was remembering the six sizes and the markup.

Screenshot Beautifier pads a screenshot onto a gradient or solid background with rounded corners and a drop shadow, for changelogs and release posts. Phone Mockups composites a screenshot into a device frame, and does it by scanning the frame image's own pixels to find the screen area and its corner radius rather than trusting a hardcoded rectangle — which is why the screen lands correctly aligned rather than a few pixels off.

Photo Collage arranges several images in a grid with configurable gaps and rounded corners. Meme Generator does the Impact-with-a-black-outline thing with draggable text boxes and word wrap. Image Color Picker samples a pixel to HEX, RGB and HSL, keeps a history of recent picks, and extracts a dominant-colour palette from the whole image. And ASCII Art Generator maps luminance to characters across four ramps, exporting both text and a rendered image — which has no practical use whatsoever and is one of the most-used tools on the site.

The image tools that use machine learning — background removal, object cutout, upscaling, captioning, depth estimation — are gathered separately in on-device AI, because they behave differently: they download a model on first use and are much slower than the canvas operations here. For turning images into documents, see the PDF collection.