yorickvp / temporalnet-sdxl

A ControlNet model designed to enhance the temporal consistency of generated outputs

  • Public
  • 132 runs
  • L40S
  • GitHub
  • Paper
  • License
Iterate in playground

Input

*string
Shift + Return to add a new line

The stable diffusion prompt. Does what it says on the tin.

*file
Preview

The input video file.

file

Path to the initial conditioning image. It is recommended you get the first frame, modify it to a good starting look with stable diffusion, and use that as the first generated frame, if unspecified it will use the first video frame (not recommended)

integer

Seed. Use this to get the same result

integer
(minimum: 0)

Only use the first N frames of the output video. 0 to use all frames.

Default: 0

boolean

Return the output as a video. Otherwise, all frames are returned separately.

Default: true

Output

Generated in

This output was created using a different version of the model, yorickvp/temporalnet-sdxl:d54d7f5a.

Run time and cost

This model runs on Nvidia L40S GPU hardware. We don't yet have enough runs of this model to provide performance information.

Readme

[warning: files generated seem to not be playing on firefox, but work well in chrome]

This is TemporalNet1XL, it is a re-train of the controlnet TemporalNet1 with Stable Diffusion XL.

This does not use the control mechanism of TemporalNet2 as it would require some additional work to adapt the diffusers pipeline to work with a 6-channel input.