Skip to content
Search
FRAME 0308

Chibi dance of joy — composed GIF

9 frames, 1.2 seconds, 443×443px; repeats indefinitely. GPT Image 2.6 frame-sheet + local GIF encoding.

Generated with GPT Image 2.6Generation prompt language · Japanese · Run recorded Nov 9, 2026 · Updated Nov 20, 2026Reference image required · 1 reference image used

Upload the 1 project-generated reference image(s) in the displayed order (R17), then use the complete adapted prompt.

Chibi dance of joy
GIF generated with GPT Image 2.6443 × 443Full-size image

Reference images

Project-generated character reference
Project-generated character reference1216 × 1293Full-size image
Prompt · Japanese
次の内容で GIF を作成して: 踊らせる、喜びの舞 【制作適応:今回の出力】上記は動きの意図です。この呼び出しではGIFファイルではなく、横4列×縦2行、合計8コマの単一PNGスプライトシートだけを描いてください。左上から右へ、次に下の段へ進む順序。各コマは同じ正方形で、区切り線、余白、番号、時刻、文字は不要。固定カメラ、固定の大きさと地面線、十分な余白、コマ間でキャラクターの同一性を維持し、実際に異なる連続動作を描く。純白背景。後で各コマを切り出してGIFにします。 8コマで左右のステップ、腕を上げる、体をひねる、小さく跳ぶ、着地、元の姿勢。全身と足元をすべてのコマに収める。
294 characters

How to use

Upload the 1 project-generated reference image(s) in the displayed order (R17), then use the complete adapted prompt.

Reuse the archived GPT Image 2.6 sheet and resample its action to nine frames. Sources with eight poses split one held pose across two encoded frames without adding a new pose; longer sequences select nine chronological poses. Preserve total duration and loop settings. Small source cells are enlarged for display without adding detail. The GIF is composed locally; the image tool returned a still PNG. The source sheet, exact crops, frame delays and first-frame poster are retained in the campaign archive.

Result and limitations

9 frames, 1.2 seconds, 443×443px; repeats indefinitely. GPT Image 2.6 frame-sheet + local GIF encoding.

Reviewed all nine-frame sequences in four contact sheets: action order, identity, captions and final poses retained. This output reuses archived GPT Image 2.6 source pixels; no new provider call. One held pose is split into two frames; eight distinct source poses, nine encoded frames. Total duration and playback count preserved. Original source review (frame counts describe source only): Eight poses show side steps, arm lift, jump, landing and return. Full body fits each cell. Small scale and ground-line variations remain visible.

Source

Adapted by Prompt Seen from @thibi_nanashi's original prompt. The author labels the source work Image 2.6. This adaptation uses GPT Image 2.6. The tool returned no exact API model ID. The prompt preserves its submitted language on both page editions. Likes describe the whole source post, shared by variants, and are not engagement for this example.

Test prints

0308-A · Displayed run

Model
GPT Image 2.6
Recorded
Nov 9, 2026
Provider
OpenAI + local GIF encoding

What we saw: Reviewed all nine-frame sequences in four contact sheets: action order, identity, captions and final poses retained. This output reuses archived GPT Image source pixels; no new provider call. One held pose is split into two frames; eight distinct source poses, nine encoded frames. Total duration and playback count preserved. Original source review (frame counts describe source only): Eight poses show side steps, arm lift, jump, landing and return. Full body fits each cell. Small scale and ground-line variations remain visible.

Reference images
  • Project-generated character reference
Output
  • Chibi dance of joyGIF generated with GPT Image 2.6

Source and licence

Site text licence (CC BY 4.0) · Testing method

Project-generated character reference
Generated imageSource unknown
Chibi dance of joy
Creator: Prompt SeenAI-generated demonstration · Prompt SeenLicence: CC BY 4.0Generated image

Inspiration & source evidence

This prompt is an independent adaptation. The image shown here is our own recorded result.

Original content
X · thibi_nanashi
Source snapshot
Whole original post; shared across variants, not this adaptationSource post likes: 25Observed 2026-11-07
Evidence
Archived public X snapshot and complete author promptIndependent adaptation; full original preserved separately

These figures belong to the original post or video. They are dated snapshots, not live counters or ratings of our image.