Skip to content

Conversation

StAlKeR7779
Copy link
Contributor

@StAlKeR7779 StAlKeR7779 commented Apr 25, 2023

Done as I said in title, but need to test(and understand) how cli works, as previously it uses single prompt and now it's positive and negative.

@blessedcoolant
Copy link
Collaborator

Works as advertised.

But I think we need to change a few things unrelated to this PR.

  1. Get a model loader node in which outputs three values - Text Encoder, UNet, VAE
  2. Feed the Text Encoder to the Compel Node instead of the Model
  3. Feed the UNet to the Text To Latent instead of Model
  4. Feed the VAE to the Latent instead of the Model

@StAlKeR7779 StAlKeR7779 requested a review from damian0815 April 25, 2023 10:20
@damian0815 damian0815 self-requested a review April 25, 2023 11:34
@damian0815
Copy link
Contributor

thanks, looks good. not sure the latents storage is the right place to put the embeddings tho

Copy link
Contributor

@Kyle0654 Kyle0654 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Some questions and comments about usability. Try using it in the CLI to validate usability =)

@StAlKeR7779 StAlKeR7779 requested a review from Kyle0654 May 4, 2023 18:16
@StAlKeR7779 StAlKeR7779 requested a review from Kyle0654 May 4, 2023 19:08
@blessedcoolant
Copy link
Collaborator

Functionality is good. But why did you merge both Positive and Negative conditioning into the same output? I think they should both be outputted as individual outputs so they can be plugged as needed across multiple nodes if need be.

@StAlKeR7779 StAlKeR7779 requested a review from blessedcoolant May 5, 2023 18:47
@blessedcoolant blessedcoolant merged commit 463f635 into invoke-ai:main May 6, 2023
@StAlKeR7779 StAlKeR7779 deleted the feat/compel_node branch May 6, 2023 01:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants