Blown windows
The defect bracketing exists to solve. Measured as the share of the frame clipped to pure white.
Free · No Signup
Listing Photo Checker
Drop the photos you are about to publish and see what a buyer's screen will show. Every finding is measured from the pixels, and every one names the fix.
Drop your listing photos here
or click to browse
JPG, PNG or WEBP · Up to 40 photos · Max 15 MB each
How the check works
Measured, not guessed
- 01
Drop the set
The photos never leave as anything but pixels to measure. There is no account, no credit and no model call.
- 02
Each frame is measured
Exposure, clipped highlights, detail surviving in the brightest region, shadow depth and edge sharpness, against what a delivered listing photo looks like.
- 03
You get findings, not a grade
The score is a summary. The useful part is the list, because each entry says what is wrong and which tool fixes it.
What it checks
Only things that can be measured from the file
Exposure
Average brightness against the 124 to 152 band a delivered listing photo sits in.
Window detail
Whether the brightest region still carries texture, or whether the glass went to paper white.
Sharpness
Edge energy, which catches missed focus, camera shake and small files scaled up.
File and framing
Resolution against portal minimums, and orientation against how portals lay a listing out.
The set as a whole
How many photos there are, and whether any of them fits a portal's landscape hero slot.
Why a checklist beats an opinion
Every photographer has been told a photo is too dark by someone who could not say how dark, or by how much. A number settles it. The point of measuring rather than judging is that you can disagree with a threshold, look at the value, and decide for yourself — and that the same photo gets the same answer twice.
- 0
- credits
- 40
- photos per check
- 124–152
- target brightness
Checker questions
What the report can and cannot tell you
Does this use AI?
No, and that is deliberate. Every finding is a statistic computed from the pixels, so it is repeatable and you can check it yourself. A model would let the report say more interesting things and would also let it be confidently wrong, which is the one thing a checklist must not be.
Why does it not tell me if a photo is crooked?
Because the measurement was not good enough to ship. It was built, tested against photos rotated by known amounts, and was wrong by up to three degrees, with a confidence signal that did not track when it was wrong. Telling someone with a level photo to rotate it is worse than saying nothing.
Why is there no colour cast check?
The obvious measurement reads the scene rather than the photograph. On a test set the strongest green readings came from a lawn and a garden, which are photos full of grass rather than photos with a cast. A real cast test needs a surface known to be neutral, and nothing in a listing photo identifies one.
Is my listing set stored?
The photos are measured in the request and not kept. If you ask for the full report by email, the address and the summary numbers are stored so the report can be sent; the photos are not.
What is the score out of 100?
A blunt summary: every photo starts at 100 and each finding takes a fixed bite. It is there to order a set and give you a headline. The findings are the substance.
Found something to fix?
Every finding names the tool that addresses it. Blown windows go to bracket merge, everything else to the enhancer.
Open the studio