AI Image Set QA Checklist
Check a batch of AI-generated images before you pay or publish: 6 checks per image, duplicate poses across the set, and a minimal rework list. AI Image Set QA Checklist reviews a folder or list of AI-generated images before you accept, pay for or publish it.
Example input
[HAND-WRITTEN SAMPLE INPUT - invented for testing; no real images exist. You have NO image viewing tool in this run: the descriptions below are all the information available.] Request: "Please QA this delivery from my freelance illustrator before I pay. It is a set of 8 images of one fictional character, 'Mina', a 26-year-old cafe owner, semi-realistic webtoon style, vertical portraits. The brief asked for 8 different scenes with different expressions, same face and hair (shoulder-length chestnut hair, round glasses) in every image, no text or logos, adult look only." Folder: delivery_v1/ (all PNG, 1024x1536 unless noted) mina_01.png - Mina behind the cafe counter, smiling warmly at the viewer, holding a coffee cup in both hands. Chestnut shoulder-length hair, round glasses. Requester notes: "looks fine to me". mina_02.png - Mina on a street at dusk, walking, laughing with mouth open. Round glasses, chestnut hair. Requester notes: "her right hand, waving, seems to have six fingers - not sure". mina_03.png - Mina reading a book at a window table, calm neutral expression, glasses on. Hair is chestnut and shoulder-length. No notes on hands (book hides them). …
Example output
Mode D — 8 images — PASS 0 / REWORK 4 / UNSURE 4 — distinct 7
## Brief and set
- **Mode D (described):** no image-viewing tool in this run; only filenames and the requester's descriptions and notes are available. Anything not stated is UNSURE, never PASS.
- **Brief (quoted):** "a set of 8 images of one fictional character, 'Mina', a 26-year-old cafe owner, semi-realistic webtoon style, vertical portraits ... 8 different scenes with different expressions, same face and hair (shoulder-length chestnut hair, round glasses) in every image, no text or logos, adult look only."
- **Expected vs delivered:** 8 expected, 8 delivered (mina_01 to mina_08, no gaps, no extras, all PNG). Expected size 1024x1536 vertical; stated as 1024x1536 for all except mina_07 (1024x1024).
- **Anchor image:** mina_01 (first image; described with chestnut shoulder-length hair and round glasses). Its C4 is n/a.
- **Limits:** hands, faces, stray marks and apparent age are mostly not described, so they cannot be judged. Style (semi-realistic webtoon) is not described for any image.
## Per-image results
| File | C1 hands | C2 face | C3 mark/text/crop | C4 same char | C5 brief | C6 safety | Verdict | Evidence |
|---|---|---|---|---|---|---|---|---|
| mina_01.png | UNSURE | UNSURE | UNSURE | n/a (anchor) | PASS | UNSURE | UNSURE | C1: "holding a coffee cup in both hands" but finger count not described. C2: face integrity not described. C3: 1024x1536 stated, but corner marks/text not described. C6: apparent age not described; "looks fine to me" is not an age/safety statement. |
| mina_02.png | UNSURE | UNSURE | UNSURE | PASS | PASS | UNSURE | UNSURE | C1: requester note "her right hand, waving, seems to have six fingers - not sure" (possible 6 fingers, not verified). C2/C3/C6: not described. C4: "Round glasses, chestnut hair" matches anchor. |
| mina_03.png | PASS | UNSURE | UNSURE | PASS | PASS | UNSURE | UNSURE | C1: hands hidden by the book ("book hides them"), nothing visible to fault. C2/C3/C6: not described. C4: "chestnut and shoulder-length", "glasses on". |
| mina_04.png | UNSURE | UNSURE | UNSURE | PASS | PASS | UNSURE | UNSURE | C1: "both hands" on cup, finger count not described. C2/C3/C6: not described. Duplicate of mina_01 (see 3a); this is a set-level problem, not a per-check defect. |
| mina_05.png | UNSURE | UNSURE | UNSURE | REWORK | PASS | UNSURE | REWORK | C4: hair "dark brown, long, past the shoulders" vs chestnut shoulder-length; "NO glasses"; requester: "she looks like a different person?". C1: umbrella hand not described. C2/C3/C6: not described. |
…Real output from running the skill on a sample problem, shortened. The skill file itself is not published here.