Video to Video AI Style Transfer with Stable Diffusion and Keyframing in TouchDesigner - Tutorial
Added 2023-02-19 01:04:37 +0000 UTCHey! In this tutorial, we'll go over how how to do video to video style transfer with Stable Diffusion using a custom component in TouchDesigner. We'll cover how to keyframe in animations so you can drive your AI style generation with audio reactivity and swap out your model parameters at exact times in your video.
Link to early SD Tutorials: https://www.youtube.com/watch?v=mRXTR9vcHAs
Link to the Computerender API: https://computerender.com/
Prompt Engineering with Lexica: https://lexica.art/
Opt out of Ai Training: https://haveibeentrained.com/
Huge thank you to Peter Whidden for his support on this and his work on Computerender which makes this project possible!
Comments
Would love to see something like this with stable diffusion running locally. I know it will be a lot slower ( I can do about 1 image per second right now with RTX 3080 and automatic1111) but it would be so powerful in an installation where theres no internet connection. I've been trying to figure out ways to speed it up by using very low res image generation for example, but so far havent been able to make anything too great.
James Clarke
2023-12-09 16:37:59 +0000 UTCI second this!
Ben Meyers
2023-05-04 01:59:01 +0000 UTCwould love to see this work with the automatica 1111 api, could be a game changer
Daniel
2023-03-09 05:15:20 +0000 UTC
