Score two or three designs before you upload — contrast, color, clutter
| Metric | Target Band | Weight | Why It Matters |
|---|---|---|---|
| Contrast (luminance spread) | std-dev ≥ 0.18 | 30% | Sets subject off background; weak contrast blends into feeds |
| Brightness (mean luminance) | 0.35 – 0.75 | 25% | Too dark disappears in dark mode; too bright reads as an ad |
| Clutter (edge density) | 0.05 – 0.15 | 25% | Subject must read at 168 px — busy edges become noise |
| Saturation | 0.35 – 0.75 | 20% | Color pops without reading as spammy neon |
Luminance uses Rec. 709 weights (0.2126 R + 0.7152 G + 0.0722 B) on a 0-1 scale. Edge density is the mean adjacent-pixel luminance difference.
| Requirement | Value |
|---|---|
| Resolution | 1280 × 720 px |
| Aspect ratio | 16:9 |
| Minimum width | 640 px |
| Max file size | 2 MB |
| Formats | JPG, PNG, GIF |
| Feed display size | ~168 px (mobile) to ~360 px (desktop) |
Specs are YouTube's published requirements. The tester previews your file at feed scale because that's where impressions are actually won.
Every image you drop in gets decoded to pixels and measured on four qualities that predict whether a thumbnail survives the feed. The math runs in your browser — nothing uploads, nothing expires, no account.
Each image is downsampled to a small grid and every pixel is converted to luminance (Rec. 709) and saturation (max-minus-min over max channel). Mean luminance becomes the brightness metric; the standard deviation of luminance becomes contrast — the same RMS contrast measure photographers use. Edge density is the average absolute difference between neighboring pixels: low means flat, high means visually busy. Each metric maps to a 0-100 sub-score via its target band, then the weighted total (contrast 30%, brightness 25%, clutter 25%, saturation 20%) becomes the attention score.
Export your candidates at 1280×720, drop two or three in, and read the bars: the sub-score that's dragging is the sub-score to fix — a contrast bar in the reds means separate your subject from the background; a clutter bar over target means remove elements until the subject reads at the small preview. Then look at the mobile-size preview yourself, because legibility is ultimately a judgment the metrics only approximate. Take the winner (and one runner-up) into YouTube Studio's Test & Compare for the real vote.
Three reference images show how the formula responds. Design A is the ideal: mean luminance 0.62, contrast 0.18, saturation 0.55, edge density 0.08 — every metric inside its band, composite 100.0/100. Design B is the classic dark-and-flat miss: luminance 0.28, contrast 0.06, saturation 0.15, edges 0.02. Its brightness scores 80, contrast 33.3, saturation 42.9, clutter 40, and the weighted total lands at 48.6 — the score says what your eye would: muddy, invisible next to dark-mode UI. Design C overshoots: luminance 0.82, contrast 0.22, saturation 0.80, edge density 0.28. Contrast earns 100, but over-bright brightness scores 72, over-saturated color 66.7, and busy edges 0 — composite 61.3. All three totals come straight from the scoring function and are reproducible with a calculator.
Two ways. Before publishing, run a pre-test yourself: score each design on measurable qualities — contrast, saturation, brightness, clutter — and show them at mobile feed size to check legibility. After publishing, use YouTube Studio's built-in Test & Compare, which serves up to three thumbnails to real viewers and reports which won. This tool handles the pre-test; Studio handles the real experiment.
Measurably: strong luminance contrast between subject and background, mean brightness in the middle band (very dark thumbnails vanish in dark mode, blown-out ones get skipped), saturated but not neon color, and low visual clutter so the subject reads at 168 pixels wide. Beyond the metrics: one focal point, a face or object with clear emotion or stakes, and 3-5 words of text at most.
1280x720 pixels, 16:9 aspect ratio, under 2 MB, saved as JPG, PNG, or GIF — those are YouTube's published specs. Anything smaller than 640 pixels wide will look soft in the feed. This tester also previews your upload at mobile feed scale, where most impressions actually happen.
No — and be suspicious of any tool that claims it does. Real click-through can only be measured with real viewers, which is what YouTube's Test & Compare does. What this tool measures are the design qualities that predict scannability: contrast, color, brightness, and clutter. Strong metrics get you into the contest; the audience decides the winner.
The feed shrinks your 1280x720 art to roughly 168-360 pixels wide depending on device and placement. Fine detail, thin text, and busy backgrounds that looked good full-size disappear at that scale. Testing at small size — which this tool does — catches the problem before upload.
Three here, which matches YouTube's own Test & Compare limit of three thumbnails per video. Each design gets a 0-100 attention score with per-metric breakdowns, and the highest total is flagged as the pick for a Studio experiment.