---
url: https://superduperai.co/library/anime-cel-style-video-prompt/
title: Anime style: 2D cel look, limited animation, clean line weight
description: A filled 7-part prompt for a hand-drawn anime shot: flat cel shading, animation on twos, a held painted background, and the three things the style breaks in current video models.
category: generation
platforms: [any]
works_with: [kling, runway, veo, image-gen]
tags: [anime, 2d, cel-shading, style-prompt, text-to-video, image-to-video]
difficulty: intermediate
verified: false
created: 2026-11-01
---

# Anime style: 2D cel look, limited animation, clean line weight

A filled 7-part prompt for a hand-drawn anime shot: flat cel shading, animation on twos, a held painted background, and the three things the style breaks in current video models.

## Prompt

```text
SUBJECT: {subject, with 2-3 visual details: hair colour and shape, one accessory, outfit colours}
ACTION: {one action, one beat, e.g. 'turns from the window and lets the curtain fall'}
SETTING: a quiet 2D background painted in flat colour fields with a soft gradient sky, late afternoon, a suburban street with power lines and a vending machine; the background holds still while the character moves
CAMERA: medium shot, eye level, static; one slow horizontal pan at the end of the shot, no zoom
LENS & LIGHT: flat, no lens; cel shading with two tones per surface (base colour and one hard-edged shadow), a thin rim of light on the hair, no gradients on the skin
STYLE: hand-drawn 2D anime cel look, limited animation on twos (12 drawings per second), a consistent clean dark line weight around characters, thinner lines inside, backgrounds without outlines, slight paper-like texture, muted pastel palette with one saturated accent; no 3D, no photoreal, no CGI shading
SOUND: soft wind, distant cicadas, no music
DURATION: 5 seconds. No on-screen text, no subtitles, no logos, no speech.
```

## Tips

- What it breaks: mouths and speech (the lip flap drifts), fast action (the model smears instead of drawing in-betweens), and any text, including signs. Keep the character silent and the action slow.
- Faces drift within the shot more than in photoreal styles. One character, one beat, under 6 seconds; use the first-frame method with a still from an image model for anything that must match a previous clip.
- Aspect: 16:9 for a scene with a background; 9:16 only for a character close-up, because the painted background is what sells the style and a vertical crop removes it.
- Try first: Kling or Runway with an image-model still as the first frame; the 'on twos' instruction is honoured more often from a still than from text alone. A suggestion to start with, not a measurement.

## Works with

- Kling: Kling text/image-to-video prompts.
- Runway: Runway Gen prompts and camera controls.
- Veo: Google text-to-video / image-to-video prompts.
- Image models: Midjourney, Flux, Imagen: keyframes and style boards for image-to-video.