Why listing photos fail VQS: the five most common failures.

Listing photos fail VQS for five common reasons: colour tints from mixed lighting, over-cooked contrast, crushed shadows, clipped windows, and leaning verticals. Each is measured objectively, with every image scored 0-100 and a FAIL recorded at 69 or below. Because a set's score is governed by its lowest image, one failing photo can cost the whole listing its certification.

VQS runs every measured check on every image, across six categories: exposure, white balance, colour, geometry, artifacts and semantic checks, plus set consistency. In practice, most failed images trip the same few checks. Here are the five we see most often, what each looks like on screen, why it happens on real shoots, what the check actually measures, and how to keep it out of your next set. The full list of checks is covered in the plain-English guide.

1. A colour tint you stopped noticing

What it looks like. Whites that are not quite white. A kitchen with a faint green shift, a bedroom leaning magenta, or a living room where the daylight through the glass is a different colour to the room around it.

Why it happens. Interiors mix light sources: daylight through windows, warm globes, cool downlights. Auto white balance splits the difference and lands nowhere. A preset synced across the whole shoot locks the error in, and an uncalibrated editing screen hides it until a buyer's screen does not.

What VQS measures. White balance is checked several ways: overall temperature (white_balance), tint (white_balance_tint), the daylight visible through glass (white_balance_windows) and an overall colour_cast check. Each check scores 0-100. Anything at 69 or below is a FAIL.

How to avoid it. Standardise the light before you shoot. Turn lamps consistently on or off, correct each room on its own rather than syncing one setting across the set, and edit on a calibrated screen. If the whites in the frame read white and the view through the window still looks like daylight, you are most of the way there.

2. Over-cooked contrast

What it looks like. Punchy to the point of crunchy. Deep shadows and hot highlights in the same frame, and rooms that read more like a render than a photograph.

Why it happens. Contrast sells at thumbnail size, so sliders drift upward over time. Multi-exposure blends pushed hard for drama have the same effect. What looks bold on a phone looks harsh at full size.

What VQS measures. The contrast check measures tonal range objectively rather than by taste. Related checks catch the collateral damage: shadows_crushed where blacks block up, and exposure_bright where highlights run hot.

How to avoid it. Back the sliders off and view the image at full size before export. The target is the room as a buyer would see it standing in the doorway, not the version that wins a half-second scroll test.

3. Crushed shadows

What it looks like. Black holes under furniture, detail gone from dark corners, timber floors and joinery that fall away into nothing.

Why it happens. Underexposure is the honest cause. The more common one is processing: pulling blacks down to add punch, which throws away shadow detail the camera actually captured.

What VQS measures. The shadows_crushed check looks at whether shadow areas retain detail, and exposure_dark catches frames that are simply too dark overall.

How to avoid it. Expose with the shadows in mind, bracket in difficult rooms, and lift rather than crush in post. A dark corner with visible texture will always score better than a stylish void.

4. Clipped windows

What it looks like. Bright white rectangles where the garden, the view or the street should be. In a listing, the view is often the point of the room.

Why it happens. The gap between interior and exterior brightness is bigger than a single exposure can hold. Meter for the room and the windows blow out. Over-correct and you get the opposite problem, windows darker than the room they light.

What VQS measures. Three checks work together here: window_clipped catches blown window highlights, window_dark catches over-corrected glass, and window_artifacts catches blending errors around the frames. If a blend leaves double edges elsewhere in the image, ghosting_artifacts will find those too.

How to avoid it. Bracket your exposures and blend with restraint. The window should read as slightly brighter than the room, with the view recognisable. Not a white panel, and not a second photograph pasted behind the glass.

5. Leaning verticals

What it looks like. Door frames and wall edges that tilt inward or outward, and rooms that feel subtly wrong even to viewers who cannot say why.

Why it happens. Tilting the camera up or down in tight rooms, usually with a wide lens, converges the verticals. It is easy to miss on a small screen and easy to leave uncorrected in a fast edit.

What VQS measures. The geometry_vertical check measures how far verticals deviate from true, and geometry_horizontal does the same for level lines like benchtops and horizons.

How to avoid it. Level the camera at capture, raise or lower the whole rig rather than tilting, and apply perspective correction in post before export. Straight lines are one of the cheapest score improvements available.

One bad frame fails the set

The SET score is governed by the lowest image in the run. A listing is only as good as its worst photo. Set-level gates apply on top: in one real example, all 19 images passed individually with a mean of 98 and a lowest score of 93, yet two soft frames tripped the focus gate and the set failed, with no badge issued. You can see how gates and per-image results appear in the sample report.

The statuses themselves are simple. PASS is 90 or above, FLAG is 70-89, FAIL is 69 or below, applied per check and per image. Certification follows the composite: Platinum at 100, Certified at 90-99, Provisional at 70-89, and no mark below 70. A failing set never certifies, whatever its composite.

Check your own work before the market does

The VQS runner is deterministic: same image, same score, byte-identical. Run a daytime shoot of 12-20 original JPEGs through the demo and you will get every measured value for every image, with nothing opined. Certification is automatic from the result, the badge is a live embeddable widget, and anyone can confirm it on the public verification register. Pricing is US$9.99 a month with one measurement run included, or US$129 for a single run with the badge valid for 12 months. The standard itself is versioned, and every release is documented on the releases page.