Background Validator

Spec v1 · Graduated cool-grey backdrop · Drop a final product image. Drag any numbered dot off the dress/model — dots must sample background only. Images are analyzed in your browser and never uploaded.

Drop image here
or click to choose a file

What this checks: background must be a cool-neutral grey (no warm/pink/beige cast — automatic fail), the gradient must run lighter-at-floor to darker-at-top, and each zone must land in a broad brightness band (targets: floor ≈#E8EAEB, wall ≈#DDE0E3, top ≈#C9CDD1 — "in the family" is good enough).
Manual checks (not automated): lace edges at 200% vs original · dress color vs physical sample · crop match across colorways

GPT Edit Prompt — use this exact text

I am attaching two images: [1] the photo to edit and [2] the approved background reference. Replace the background of image 1 so it matches the background of image 2 exactly — same brightness, same grey tone, same gradient. The hex values below describe the target, but the reference image is the source of truth.

Change ONLY the background and floor of image 1. The dress, lace pattern, train, model, skin tone, hair, and pose must remain pixel-identical — do not regenerate, sharpen, smooth, or recolor the subject in any way.

Replace the background with a professional graduated studio backdrop:
- Smooth seamless neutral-COOL light grey. No texture, no visible wall/floor seam.
- Floor: approx #E8EAEB
- Wall at the model's height: approx #DDE0E3
- Darkening gradually toward the top of frame to approx #C9CDD1 (soft vertical gradient, subtle vignette in the upper corners)
- Absolutely no warm, beige, cream, pink, or mauve tint — neutral to slightly cool grey only
- Keep a soft, realistic contact shadow under the hem and train, falling slightly behind the model
- Even, diffused lighting on the backdrop; no hard shadows
- Photorealistic high-end bridal e-commerce studio photography

Preserve exact lace and train edge detail where fabric meets the floor — no halos, no blur, no invented lace.
  1. In GPT, attach both the original photo and the approved reference image (in Dropbox: /Photo-Standards/), then paste the prompt above.
  2. Drop GPT's output into the validator on this page. Most images pass first try.
  3. PASS → continue to manual checks. FAIL → a correction prompt appears below the results. Copy it, paste it into the same GPT chat, re-drop the new output. If it still fails after a couple of tries, flag it to the photo lead rather than burning more rounds.
  4. Manual checks before upload: zoom 200% on lace/train edges vs the original (AI can invent lace detail), and confirm dress color against the physical sample.
  5. Crop to 4:7 if needed (the validator flags this), name the file per the brand guide, and upload.