Help in getting this to work in InvokeAI

#3
by wittmason - opened

I am working on getting this to work in InvokeAI and having very little luck. I've got the new model reflected in the models.yaml and have tried to manually load it and it's not working. Any help in getting this to run would be appreciated.

Not that I could help you out with any advanced debugging, but I put the ckpt in the models/ldm/stable.../ folder, more or less copied the first entry in the models yaml, just changing the name and pointing to the correct ckpt and ran the script with --model mdjrny-v4 (which is the name i defined in the yaml)

  1. Download and copy the ckpt model to:
<invokeAI folder>\models\ldm\stable-diffusion-v1\mdjrny-v4.ckpt
  1. Navigate and edit the <invokeAI folder>\configs\models.yaml
  2. Append this model definition in the config:
midjourney:
  weights: ./models/ldm/stable-diffusion-v1/mdjrny-v4.ckpt
  description: Midjourney Style, keyword mdjrny-v4 style
  config: ./configs/stable-diffusion/v1-inference.yaml
  width: 512
  height: 512  

Remark:
Use quotations if you are using special symbols in the path or description, if not the yaml parser will give you an error.

  1. Restart InvokeAI, and start the InvokeAI with:

Remark:
InvokeAI has a command line feature to import models !import_model <relative\path\to\model.ckpt>
However at the moment i wrote this, this process has a little bug that errors when it finds an existing models.yaml
and leaves the user hanging half way inthe process, so that's why i wrote manual step tutorial.

Good luck!

prompthero changed discussion status to closed

Sign up or log in to comment