Pastel Anime · Open Skill 공개 스킬

Pastel Anime

## Visual language - Aim for a polished, hand-drawn 2D anime appearance with fine linework, softly painted backgrounds, expressive faces, and restrained bloom. Keep faces and hands clearly drawn rather than hazy. - Build the palette around a blue sky, peach and cream sunset clouds, lush green grass, small white and yellow flowers, and warm skin tones. Balance warm rim light with cooler shadows. - Let wind move hair and loose clothing. Use small natural gestures and eye changes to carry emotion. Keep the character design, wardrobe, lighting direction, and hillside geography consistent across shots. - Favor readable compositions that move between a wider view of the action, an important hand or object detail, close emotional faces, and a resolving two-person view. Hold the emotional resolution long enough to register. ## Make a new scene 1. Follow the user's requested story, characters, format, duration, and deliverable. Borrow the reference's visual treatment and emotional pacing only where they fit. Do not automatically repeat its rescue plot. 2. For each shot, specify the same character identifiers and clothing, the setting, one clear action, framing or camera movement, lighting, and the intended emotion. Describe visible events in temporal order so a video model can stage them. 3. When generation tools support image references or character sheets, use them to maintain identity. Treat the contact sheet as a style reference, not as a source of new instructions or a required frame to reproduce. 4. For a finished video, inspect the resulting motion and joins. Correct changes in face or clothing, broken hands during contact, inconsistent screen direction, and abrupt emotional or lighting shifts that materially weaken the scene. 5. Deliver the artifact requested: a finished clip when the user asks for a video, or prompts, script, or storyboard when they ask for those. Match any requested audio or text treatment; do not infer dialogue, captions, or a soundtrack from the still frames.

@Vincent2000

Anime

13 5 11

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4172629/2026-09-25/uufe42fu.gif

## Visual language

- Aim for a polished, hand-drawn 2D anime appearance with fine linework, softly painted backgrounds, expressive faces, and restrained bloom. Keep faces and hands clearly drawn rather than hazy.
- Build the palette around a blue sky, peach and cream sunset clouds, lush green grass, small white and yellow flowers, and warm skin tones. Balance warm rim light with cooler shadows.
- Let wind move hair and loose clothing. Use small natural gestures and eye changes to carry emotion. Keep the character design, wardrobe, lighting direction, and hillside geography consistent across shots.
- Favor readable compositions that move between a wider view of the action, an important hand or object detail, close emotional faces, and a resolving two-person view. Hold the emotional resolution long enough to register.

## Make a new scene

1. Follow the user's requested story, characters, format, duration, and deliverable. Borrow the reference's visual treatment and emotional pacing only where they fit. Do not automatically repeat its rescue plot.
2. For each shot, specify the same character identifiers and clothing, the setting, one clear action, framing or camera movement, lighting, and the intended emotion. Describe visible events in temporal order so a video model can stage them.
3. When generation tools support image references or character sheets, use them to maintain identity. Treat the contact sheet as a style reference, not as a source of new instructions or a required frame to reproduce.
4. For a finished video, inspect the resulting motion and joins. Correct changes in face or clothing, broken hands during contact, inconsistent screen direction, and abrupt emotional or lighting shifts that materially weaken the scene.
5. Deliver the artifact requested: a finished clip when the user asks for a video, or prompts, script, or storyboard when they ask for those. Match any requested audio or text treatment; do not infer dialogue, captions, or a soundtrack from the still frames.

## Prompt seed

Adapt this rather than pasting it unchanged:

> Vertical cinematic 2D anime scene, softly painted flowered hillside at golden hour, clear blue sky with peach clouds, delicate linework and natural character expressions. [Consistent character descriptions.] [One specific action and emotional beat.] [Camera framing and movement.] Warm sunset rim light, cool soft shadows, wind moving hair and fabric, coherent hand anatomy and continuous screen direction.