🔬🎓 Colorblind Vision Simulator

How the color shift works: Each pixel is analyzed in HSL color space. Reds (hues ~330°–30°) are rotated toward vermillion/orange (~18°); greens (hues ~75°–165°) are rotated toward blue-green/teal (~165°). All other hues (blues, purples, yellows, neutrals) are left untouched. Lightness and saturation are preserved, so all luminance contrast in your figure is maintained. The slider blends between original (0%) and fully shifted (100%); 50% is the default.

Recommended workflow: Start at 50% → click ▶ Apply → check the CVD preview box to verify distinguishability → adjust slider if needed → download. Output is always the same pixel dimensions as the input.

The single most effective strategy is to never rely on color alone as the only differentiating element. Combine color with shape, line style, pattern, or direct labels. Then choose a colorblind-safe palette.

Notes: Rotate/flip buttons apply immediately to the working image and can be undone one step at a time. All other adjustments (crop, resize, tone) are applied when you click ▶ Apply & Preview. The downloaded file is always the same as the preview.

Which format should I use? For most journal submissions, PNG is the best choice — it is lossless, widely accepted, and produces files well under typical per-figure size limits (e.g. 10 MB for Wiley/FEBS journals). TIFF is uncompressed and can be very large (>50 MB for high-resolution figures) — use it only if the journal specifically requires it. JPEG is fine for photographs but introduces compression artifacts unsuitable for graphs and line art. WebP is excellent for web use but not widely accepted by journal submission systems.

DPI and pixel count: DPI is metadata that tells a printer how large to render the image — it has no effect on pixel count or on-screen quality. A 4800×3600 px image tagged as 72 DPI is identical in quality to the same image tagged as 300 DPI; only the pixel dimensions determine sharpness. Most journals require “300 dpi at the intended print width” — for a 7.5″ wide figure that means at least 2250 px wide. At 4800 px wide your figure already exceeds 600 dpi at full page width, so no resizing is needed — just set the DPI tag and export.

TIFF output uses uncompressed RGB/RGBA with correct XResolution/YResolution tags, compatible with ImageJ, Fiji, Photoshop, and journal submission systems.

Suggested crops: Check ✂️ Crop, choose a shape, and click ✨ Suggest crops. Up to four suggestions appear below the preview, best first, each with the reasons it was chosen. Clicking one puts it in the crop box, where you can drag it to a new position or edit its numbers. Here are the three kinds of shape setting. Any shape (the default) leaves the proportions to the tool: it starts from a range of shapes between 1:2.5 and 2.5:1, then nudges each edge in or out for as long as the crop’s score improves, and it also offers a tighter and a looser alternative when those differ enough to be worth seeing. Common shapes limits the tool to the original shape and the familiar ratios in the list, and picks among them. A fixed shape (1:1, 4:5, a custom ratio, and so on) uses only that shape.

Compare all shapes: Shows one thumbnail per shape — Any shape, the original shape, and every ratio in the Shape menu (plus your custom ratio, once you have chosen one) — each with the best crop the tool finds for that shape at the current Tightness. The first card, Any shape, is the tool’s own choice with no limit on proportions. The cards stay in the order of the Shape menu so that each shape is easy to find; ★ marks the best overall, and each card gives its rank by score. Clicking a card puts its crop in the crop box and sets the Shape menu to match.

How the suggestions are made: The tool first builds an importance map of the image (check Show importance map to see it, drawn in the colorblind-friendly viridis colors, where yellow is most important and dark purple least). The map combines four clues: how much each area’s color differs from the image’s average color, the frequency-tuned approach of Achanta et al. (2009); how much it differs from the colors along the image’s edges, which are usually background; how much fine detail it holds; and whether it is skin-toned, using the color ranges of Chai and Ngan (1999). The tool then tries thousands of crops of each allowed shape, size, and position, and scores each on five things: how much of the importance the crop keeps; how much of the image’s area it uses; whether the subject falls on a rule-of-thirds intersection or dead center; whether the crop’s new edges cut through important detail (cutting along the top counts double, because that is where heads usually are); and how much of any skin-toned area it keeps, since skin is often a face. Scoring candidate crops against an importance map follows the approach of smartcrop.js by Jonas Wagner (2014). The rule of thirds places a subject one third of the way in from an edge rather than in the middle; the phrase was first written down by John Thomas Smith in 1797.

Tightness: At the Auto setting (the middle of the slider), a crop keeps growing for as long as each strip it adds holds at least roughly a third as much importance, area for area, as the image does on average. A small subject in a large empty scene therefore gets a tight crop, while an image whose content is spread throughout is barely cropped. Move the slider toward Loose to keep more surroundings, or toward Tight to close in.

Marking the subject: If the suggestions miss what matters, click 📍 Mark subject and then click on the subject in the preview (each click adds another mark). Every suggestion must then include every mark.

Limitations: The tool does not know what anything in the image is. Skin tones are only a stand-in for faces: the tool does not recognize faces as faces (wood, fur, and orange objects can pass for skin, and faces in grayscale images are not detected), nor which way a subject is looking, or where a horizon is, so it cannot leave room in front of a face or level a tilted horizon. Thin, faint subjects (a distant mast or antenna, for example) can be missed. Treat suggestions as starting points.

Trim margins: Removes a uniform border — white, black, any flat color, or transparency — and leaves the padding you choose. The background color is read from the image’s corners, and small speckles such as JPEG compression noise are tolerated. This is mainly useful for figures and graphs exported with extra white space; on a photograph with a cloudless sky or a plain backdrop, it will trim those areas too.

Selection size: The line above the preview gives the selection’s size in pixels and the largest print size it supports at the DPI set under Output Format & DPI. Cropping discards pixels, so a tight crop of a small image may be too small to print sharply.