Tags: animation video-generation layer-compositing Video Template: Markdown Parameterized Input: Character_reference_images (image) Input: First_frame_reference_image (image) Input: characters (text) · optional · default: cat, crocodile, horses with cowboys, and the main hero Input: background_style (text) · optional · default: desert, cacti, clean shapes Input: creator_credit (text) · optional · default: voxelplot Output: Character_reference_images (image) Output: First_frame_layout (image) Output: Layer_images (image) Output: Individually_animated_layers (video) Output: Final_composited_video (video) midjourney google-nano-banana openai-gpt-2 openai-chatgpt bytedance-seedance-2 openai-gpt bytedance-seedance

Animate a Scene in Separate Character Layers

A staged workflow for creating character references, assembling a 16:9 layout, generating consistent layer images, animating each layer independently, and compositing the clips into a final video.

By Threads — Published on Oct 06

Example results

Download video result 1

Download video result 2

Screenshots

Animate a Scene in Separate Character Layers output 1

Animate a Scene in Separate Character Layers output 2

Animate a Scene in Separate Character Layers output 3

Animate a Scene in Separate Character Layers output 4

Animate a Scene in Separate Character Layers output 5

Animate a Scene in Separate Character Layers output 6

Animate a Scene in Separate Character Layers output 7

Animate a Scene in Separate Character Layers output 8

Prompt template

Create an animated scene by generating and animating its characters in separate layers, then compositing the clips.

Scene characters: ${characters:cat, crocodile, horses with cowboys, and the main hero}
Background style: ${background-style:desert, cacti, clean shapes}
Aspect ratio: 16:9
Creator credit: ${creator-credit:voxelplot}

1. Generate a separate reference image for each character using Midjourney or another image-generation tool.
2. Build the scene's first-frame layout using the character references and the supplied first-frame reference image. Match its camera angle and character placement. Establish the scene background using the specified background style and give the composition a consistent overall energy.
3. In the same ChatGPT conversation, generate three separate layer images from the layout:
   - Layer 1: cat and crocodile only.
   - Layer 2: horses with cowboys only.
   - Layer 3: main hero only.
   Keep the background identical across all three images. Do not change character positions or proportions. Use a 16:9 format.
4. Animate each layer separately, using a video-generation tool that works well with the chosen visual style.
5. Composite the clips in Seedance 2. First combine the cat-and-crocodile clip (video 1) with the horses-and-cowboys clip (video 2), using the shared scene composition (image 1) as the first-frame reference. Preserve the layer positions and proportions. Then place the main-hero clip over the resulting video in a second compositing pass.

Include this creator credit with the finished work: ${creator-credit:voxelplot}