lucataco / wan-2.1-1.3b-vid2vid

Wan 2.1 1.3b Video to Video. Wan is a powerful visual generation model developed by Tongyi Lab of Alibaba Group

  • Public
  • 765 runs
  • GitHub
  • Weights
  • Paper
  • License
Iterate in playground

Input

*string
Shift + Return to add a new line

Text prompt describing what you want to generate or modify

string
Shift + Return to add a new line

Negative prompt to specify what to avoid in the generation

Default: "low quality, blurry, distorted, disfigured, text, watermark"

file

Input video for video-to-video generation

integer
(minimum: 1, maximum: 100)

Number of frames to generate in the output video

Default: 81

integer
(minimum: 5, maximum: 24)

Number of frames per second in the output video

Default: 16

number
(minimum: 0, maximum: 1)

Strength of denoising when using video-to-video mode. Higher values change more content.

Default: 0.7

string

Aspect ratio for the output video

Default: "832x480"

integer
(minimum: 10, maximum: 50)

Number of sampling steps (higher = better quality but slower)

Default: 40

number
(minimum: 0, maximum: 20)

Classifier free guidance scale (higher values strengthen prompt adherence)

Default: 6

integer

Random seed for reproducible results (leave blank for random)

boolean

Whether to use tiled sampling for better quality on larger videos

Default: true

Output

Generated in

Run time and cost

This model costs approximately $0.20 to run on Replicate, or 5 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia L40S GPU hardware. Predictions typically complete within 4 minutes. The predict time for this model varies significantly based on the inputs.

Readme

About

Cog implementation of modelscope’s video to video process