Fuser

AnimateDiff

Provided by Fal — Learn More

AnimateDiff allows you to transform text and video prompts into dynamic videos with a wide range of motion types.

Preview

Inputs

Prompt
text
The prompt to use for generating the video. Be as descriptive as possible for best results.
Video
video
The video to transform.
optional: true
Negative Prompt
text
The negative prompt to use. Use it to address details that you don't want in the image.
default: (bad quality, worst quality:1.2), ugly faces, bad anime
Number of Frames
number
The number of frames to generate. Value is ignored when video is provided.
default: 16
First N Seconds
number
The first N number of seconds of video to animate.
default: 3
Steps
number
The number of inference steps to perform. 4-12 is recommended for turbo mode.
default: 8
Strength
number
The strength of the input video in the final output.
default: 0.7
Guidance Scale
number
The CFG scale is a measure of how close you want the model to stick to your prompt.
default: 1
Seed
seed
The same seed and prompt will output the same image every time.
default: 8919
FPS
number
Number of frames per second to extract from the video.
default: 8
Motions
dropdown
The motions to apply to the video.
optional: trueAccepts: Zoom Out, Zoom In, Pan Left, Pan Right, Tilt Up, Tilt Down
Video Size
dropdown
The size of the video to generate.
default: square_hdAccepts: Square HD, Square, Portrait 4:3, Portrait 16:9, Landscape 4:3, Landscape 16:9
Use Turbo
toggle
Turbo mode is faster but may produce lower quality results.
default: true

Outputs

Generated Video
video
The generated video from the model.