License Platform

Pokora icon

Pokora is a video creation platform that combines existing video clips with AI generated video clips, using Stable Diffusion, in a native SwiftUI interface, completely local with no internet access necessary. Pokora uses the frames of an input movie to run image to image processing with a Stable Diffusion model. Check out ml-stable-diffusion for the latest CoreML model changes and how to convert models.

?‍♂️ Pokora is named after Hans Pokora, author of many books on collectable psychedelic vinyl.


  • Load video from disk ✅
  • Process frames using Stable Diffusion (prompt, seed, strength) ✅
  • Export video including audio from original video ✅
  • Need icon ✅
  • Need easier install of models ✅
  • Adjust strength over the course of the video #7
  • Playback video in app #9
  • Add up rezzing using RealESRGAN #8
  • Persist between launches #16
  • Update to use ControlNET #10
  • Show preview while processing #34
  • Multiselect #36


  • Requires square video as input
  • Requires models converted to CoreML (see below)
  • Currently using ml-stable-diffusion v0.3.0


Built using below, but haven’t tested elsewhere yet.

  • macOS 13.3.1+
  • Xcode 14.3+


You will need to convert or download models in CoreML format. You can download from the HuggingFace org here.

NOTE: I had trouble with the v2.1 model, I think it doesn’t like the 768×768. I verified this model works here, however I have had better speeds with a model I converted myself.


This project is licensed under the MIT License – see the LICENSE file for details.


View Github