---
name: character-sheet
description: "Make the character sheet of the person's AI character from their photo(s) — face close-up + full body in one image, a turnaround, a pose sheet and an expression sheet from reference charts, a back view — with GPT Image 2.5 edit at quality low (the winner against Nano Banana 2 Lite: 10 of 10 poses, 9 of 9 expressions, the character kept in 10 of 10 film frames), two-sentence prompts with slots, sizes and checks. Use when the user says character sheet, model sheet, reference sheet, turnaround, pose sheet, expression sheet, back view, make a sheet from my photo, лист персонажа, лист из фото, разворот персонажа, позы, эмоции персонажа. Part of the character meta-skill."
---

These creative references describe source experiments with external tools. Use your own supported account and tools; this download installs no automation. Settings and timings are experimental observations, not current FlashBoards availability.

# character-sheet · the sheet from the person's photo

Inputs are the person's own: {photo} — their photo(s) of {character} (best: one front face close-up + one front full body, the whole outfit to the shoes). Output: the {character sheet} that goes into every later prompt and into their character card.

## The winner

**GPT Image 2.5 edit**, `openai/gpt-image-2.5/sunburst/edit` (or `…/flare/edit`), **`quality: "low"`** (set it — the edit schema defaults to `high`), one image, size set to the layout. Short prompts that name picture roles and the layout, nothing about the looks.

| Job (same inputs on both) | Nano Banana 2 Lite | GPT Image 2.5 Sunburst |
|---|---|---|
| Turnaround, 4 views | ✕ all views, but in its own order (front, profile, profile, back); softer | ✓ front, back, left, right in prompt order; sharper fabric and face (quality high in that run) |
| Pose sheet, 10 poses from a chart | ✕ 5–8 of 10 — the models of the references survive (heels, stockings, their clothes); more warnings broke the grid into 15 panels | ✓ 10 of 10, every run |
| Expression sheet, 9 faces from a chart | ✕ 0–7 of 9 (chart as a reference came back unchanged: 0 of 9) | ✓ 9 of 9, every run, no earrings / fringes of the references |
| Character in a film frame | ✕ the actor comes back in about 1 run of 6, even with a long role-by-role prompt | ✓ 10 of 10 frames across 5 films, one-line prompt |
| Seen from behind | front-only sheet: a new back every run (8 backs, no two alike) | sheet with a back panel: 9 of 9 backs the same |

Nano Banana 2 Lite is faster (6–9 s a view) — it is the losing option for sheets.

## Steps

1. **Check the photo.** One face photo gives no trousers, shoes or back. Ask for a front full-body photo; if there is none, generate the full body from the face photo first and let the person approve it before the sheet.
2. **The base sheet** (face close-up + full body, the picture used everywhere later). Picture 1 (and 2) = {photo}. Size 1536×1152 (4:3) or 2048×1152.
   ```
   Character sheet of {character} from pictures 1–2: two views side by side on a plain white background — an extreme close-up of the face and a full-body front view. Same person and same clothes in both views, no text.
   ```
   With one photo: "from picture 1". Plain white: only the person travels on.
3. **Turnaround** (optional; it also gives the back). Same inputs, 2016×864 (21:9):
   ```
   Character turnaround of {character} from pictures 1–2: four full-body views in one row on a plain white background — front, back, left profile, right profile. Same person and same clothes in every view, no text.
   ```
4. **Give the sheet a back** — needed whenever the character will be seen from behind or small in a wide shot. Picture 1 = the base sheet, picture 2 = the turnaround (its back figure); 1792×1024, medium in our run:
   ```
   Character sheet of {character} from pictures 1–2 on a plain white background, three panels in one row: an extreme close-up of the face, the full body from the front, the full body from exactly behind. Same person and same clothes in every panel; the face is exactly as in picture 1; the back view shows exactly the back of the hair and clothes as in picture 2. No text.
   ```
   Run it twice and keep the cleaner (seamless white, no lines between panels).
5. **Pose sheet / expression sheet from references.** Poses and faces are easier to show than to describe: glue the reference photos into one chart in the grid you want (a layout step with PIL/ffmpeg, not a generation) — 10 poses as 5 × 2, 9 expressions as 3 × 3. Picture 1 = the chart, picture 2 = {character sheet}. Size = the chart's grid (1600×1200 for 5 × 2, 1536×1536 for 3 × 3).
   ```
   Picture 1 is a chart of {N} poses in a {C}-column × {R}-row grid. Redraw it as a new photo session of {character} from picture 2 — their face, hair, clothes and shoes in every panel — each panel keeping exactly its pose, body angle and camera angle. The same {C} × {R} grid, one plain mid-grey studio backdrop and soft even light in every panel, no text.
   ```
   Expressions: "a chart of {N} facial expressions in a {C} × {R} grid … their face, hair and clothes in every panel — each panel keeping exactly its facial expression, head angle, gaze and framing". Keep it at ≤ 10 panels: more panels = less resolution each and a higher chance of a broken grid.
6. **Check** (crops, panel by panel): the same face in every panel (marks, eyes, hairline), the outfit complete to the shoes, the views in the asked order, nothing of the reference people (hair, jewellery, clothes, black-and-white), one backdrop, no text. Score it as "N of N panels"; rerun only the failures.
7. **Save** the chosen sheet(s) to the person's character card; they are picture 1 of `character-consistent`, `character-replace` and the `animate` skills.

## Using the sheet in a film scene (measured, details in other skills)

- Into a film frame: the sheet first, the frame second, "shoot it again exactly the same… but the person in it is {character} from picture 1… Change nothing else" → `character-replace`. A small figure seen from behind kept the actor's jacket and canteen with the one-line prompt in 3 runs of 4; naming the character's clothes and saying that nothing of the actor's costume stays made it 4 of 4 in the earlier test. A later test on 9.10.2026, recorded in `character-replace`, also held 4 of 4 with the blind template and no clothes named: none of the actor's gear remained, though the pose straightened. A sheet with a back panel pins the back; the actor's leftover gear still needs that sentence.
- Video from the sheet (MiniMax H3 Max Turbo, `animate` meta-skill): **Way A** — the sheet on a 16:9 canvas as the start frame, the scene in words: the face holds, the room, lens and light are the model's reading. **Way B** — replace the actor in the film's first and last frame, animate between: framing, set, light and colour fixed by real frames; the way to stay close to the film. 9 s and 12 s, 768P, one generation each.
- Where it goes wrong: the frame comes back in the sheet's proportions (set the size to the frame's aspect); the head turns the wrong way (give every turn as a side of the frame); the sheet flashes in the video (say in the first line that the video opens on shot one and the reference never appears).
- A whole scene longer than 15 s: the face drifts between start and end frames → the face board on the end frame (`animate-scene-pairs`).

## Never

- Describe the person's face, hair or clothes in the prompt — the photo carries them.
- Give a reference chart as a plain reference to Nano Banana (0 of 9) or trust Nano Banana with reference people in the frame.
- Use a demo character's photos, sheet, charts or prompts for a person.

## Settings and times (measured)

| Request | Time |
|---|---|
| GPT Image 2.5 Sunburst · low · pose / expression sheet (1600×1200 / 1536×1536) | not recorded |
| GPT Image 2.5 · low · film frame with the character | 20–26 s |
| GPT Image 2.5 Flare edit · low · four-view sheet 2048×1152 · 1 / 2 references | — |
| Same · two references resized to 768 px wide before upload | not recorded |
| Nano Banana 2 Lite · a sheet view / a two-view sheet | 6–9 s |

External workflow: upload your own photo/chart through a supported image-edit tool, set the GPT Image endpoint and quality `low` explicitly, and pass the prompt and target dimensions. Nano Banana 2 is a comparison model, not the selected sheet workflow.

## Example run (reference only — never reuse its files, prompts, faces or frames for a person)

Demo character **Mara Vale**, 28–29.09.2026. Generated face and front-body anchors were used instead of a real person's photos. The Nano Banana 2 Lite base sheet was 4:3 and took 6 s; turnarounds compared Nano against GPT at quality high, 2016×864. The short chart prompts yielded 10/10 poses and 9/9 expressions. The back-sheet and drift studies supplied the later identity checks. Film tests used *The Shining*, *Don't Breathe*, *This Boy's Life*, *Do the Right Thing*, *The Color Purple*, *Mad Max: Fury Road* and the *Joker* stairs scene. The earlier *Mad Max* replacement named Mara's clothes; the later 9.10.2026 blind test in `character-replace` held 4/4 without naming them. These are separate tests, not a remeasurement of every sheet result. No example files are delivered or required.