How ListingLint checks your photos

By the ListingLint team · Updated 23 August 2026

Most "is your photo compliant?" advice is a checklist you eyeball. ListingLint instead reads the actual pixels of your image and measures each rule numerically. This page explains exactly how each check is computed, the tolerances we apply, and — importantly — which results are exact and which are estimates you should sanity-check yourself.

Everything here runs in your browser. Your image is decoded to a <canvas>, measured locally, and never uploaded. Close the tab and the pixels are gone. That's why the checks are instant and private.

The pipeline, in one paragraph

When you drop in a photo, the browser decodes it and draws it onto an off-screen canvas. From there we can read the raw RGBA value of every pixel. Each check is a small, deterministic function over those pixel values: some scan the border, some scan the whole frame, some look at contrast between neighbouring pixels. There is no server, no machine-learning model, and no external API — just arithmetic on the pixels, which is why the same image always produces the same result.

Check 1 — Background colour (exact)

The white-background rule is the one that trips up the most Amazon sellers, so we treat it precisely. We sample the pixels around the outer edge of the image — the region a marketplace's own automated check is most likely to read — and look at their R, G, and B values.

What "pure white" means here: a pixel counts as pure white only at 255,255,255. We report the measured value so you can see exactly how far off you are — for example, a lightbox background that reads 249,250,248 is visibly flagged as not pure white even though it looks white on screen.
Near-white detection: to decide whether a background can be rescued by snapping, we treat pixels at roughly 244+ on every channel as "near-white." Above that threshold, a one-click snap to 255,255,255 is safe; below it, the backdrop is genuinely coloured or grey and needs replacing, not snapping.
Why the edge, not the centre: the centre of the frame is usually the product. The border is where the background lives, so that's where a "is the background white?" question is actually answered.

This check is exact: it reports the real pixel values. See why near-white backgrounds get rejected for the practical side of this.

Check 2 — Dimensions and zoom eligibility (exact)

Image width and height come straight from the decoded file, so this is simply a comparison against each marketplace's thresholds:

For Amazon we flag the longest side against 1000px (the point at which zoom becomes available) and note whether you clear the 1600px comfort mark.
For eBay we check the 500px minimum and the same 1600px zoom recommendation.
For Etsy we look at the shortest side against a 2000px target, because Etsy's audience expects crisp, zoomable detail.

We deliberately ignore DPI/PPI metadata. For on-screen marketplace images only the pixel dimensions matter — a "72 DPI" and a "300 DPI" file of the same pixel size are identical to a browser. More on that in Amazon image size requirements.

Check 3 — Product frame-fill (measured, with assumptions)

Amazon asks that the product fill at least 85% of the frame. To estimate fill, we separate "product" pixels from "background" pixels: starting from the assumption that the background is the light, uniform region around the edges, we find the bounding box of the non-background content and compare its area to the total frame.

On a clean white background this is highly reliable — the product is obvious.
On a busy or low-contrast background it's an approximation, because the product is harder to separate from its surroundings. We surface the percentage so you can judge it against what you see.

Check 4 — Edges, shadows, text and borders (estimates)

Some rules can't be measured to the pixel without understanding the image, so we treat them as heuristics — prompts to look closely, not verdicts:

Edge shadow: we look for a grey gradient creeping in from the border, which often means a shadow is falling on the backdrop and dragging it below pure white.
Possible text or logo: we look for small, high-contrast clusters typical of overlaid text or a watermark. A busy real-world product can trigger this, which is exactly why it's labelled an estimate.
Hard border: we check whether the outermost ring of pixels forms a continuous frame of a single non-background colour.

When a check is an estimate, the tool says so. We would rather tell you "worth a look" than give you false confidence.

What the one-click fixes actually do

Snap background to white pushes near-white pixels (the 244+ band) up to 255,255,255 while leaving darker product pixels untouched, so the product keeps its edges.
Resize scales the image up or down to hit a marketplace's recommended dimensions using the browser's canvas resampling.
Every fix is applied to a copy on the canvas; your original file on disk is never modified. You download the fixed version only if you choose to.

Why measure instead of guess?

The whole reason marketplace rejections are frustrating is the feedback loop: you upload, wait, and only then learn something was off by a few pixel values you could never have seen. Measuring the image before you upload closes that loop. It also means our advice is falsifiable — if the tool says your background is 255,255,255, that's a fact about your file, not an opinion.

Check a product photo now → or browse the full guide library.

Frequently asked questions

Does ListingLint guarantee my photo will be approved?

No, and no honest tool can. Marketplaces run their own checks and change them without notice, and some rules (like "is this the actual product?") require human judgement. ListingLint catches the measurable failures — the majority of silent rejections — before you upload.

Are my images really never uploaded?

Correct. All decoding, measuring, and fixing happens in your browser via the Canvas API. There is no upload step and no server that receives your photos. See our privacy policy.

Why is one check "exact" and another an "estimate"?

Exact checks read a value that is unambiguous in the file (a pixel's colour, the image's width). Estimates infer something that depends on interpreting the image (where the product ends, whether a mark is text). We label them differently so you know how much to trust each.

This page describes ListingLint's own approach for transparency. Marketplace policies are set by Amazon, eBay, and Etsy and can change at any time; always confirm against the official source. ListingLint is not affiliated with or endorsed by any marketplace.