Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Feature Request]: Model Presets & Scene/Style Presets #11

Open
1 task done
GRwald opened this issue Oct 30, 2024 · 3 comments
Open
1 task done

[Feature Request]: Model Presets & Scene/Style Presets #11

GRwald opened this issue Oct 30, 2024 · 3 comments
Labels
enhancement New feature or request

Comments

@GRwald
Copy link

GRwald commented Oct 30, 2024

Is there an existing issue for this?

  • I have searched the existing issues and checked the recent builds/commits

Summary

-With so many models it is hard to keep track what the sweet spot is for any model, with Flux and SD3.5 it gets even more complicated.
-When crafting styles/settings/scene it is nice to keep these prompts in a library.

Fooocus is the software with this feature, I use right now, Invoke has this too and also the krita plugin.
But I think if you go the external way (Json files) Fooocus is more flexible.

The prompt is tied to the model, it reacts differently to instructions, and its limitations, per model. So if you can add a prompt to a model you can really re-use your workflow especially with re-using the technical setup with Distry-flow!
Fooocus also got styles to swap in and out pieces of prompts. Not like wildcards but chosen by the user because he is seeking for a night scene and just want too load it into the prompt, and found a while back some keywords that worked.

I really like what you did for comfyUi, it really changes everything, it could become a standard install like the manager, because it solves that comfyUi is often not as comfy as it sounds :)

Great if model makers can share for instance a json file with best settings recipe, ComfyUi will become even as simple as Fooocus

Description

So:
-A way to store model recipes, and perhaps share it (by community or maker)
-A way to re-use well crafted prompts in a library. Logical to be combined and attached to the model recipe. Could also be as a style library. (+lora`s ? )

Would be very nice to do this internally inside of Disty-Flow but external is also ok right? Like Fooocus does...

I think this folds nice with your great Ui concept

Thank you so much for your work! And making this awesome ui! (make comfui comfy again! (sorry) )
Tnx <3
Gerwald

Additional information

see Fooocus presets here:
https://github.com/lllyasviel/Fooocus/tree/main/presets
external site to create presets:
https://ai.kleebz.com/fooocus/presets/

See Fooocus styles here:
https://github.com/lllyasviel/Fooocus/tree/main/sdxl_styles

@GRwald GRwald added the enhancement New feature or request label Oct 30, 2024
@diStyApps
Copy link
Owner

Thank you for great suggestions, i will check it out.

@GRwald
Copy link
Author

GRwald commented Nov 4, 2024 via email

@mBIMMER
Copy link

mBIMMER commented Nov 7, 2024

FIRSTLY! THANK YOU SO MUCH FOR THE COOL ADDON
I agree with Fooocus (oe similar DeFoocus or SimpleSDXL) so far the only one who did it, the ability to visually mix, reference photo + style and colors of photo + face replacement, ETC
2024-11-07_180250
2024-11-07_180535

it would be cool to see this similar options under options with Flux dev ,GUFF
2024-11-07_181804
2024-11-07_181717

PS I would like to see more options for Flux dev ,GUFF :)
in one interface inpaint and outpaint option
with the ability quick selection(without reloading the interface and models) create an image+ IMG2IMG together in one interface+ improve the image quality, Refine and+ Upscale
and it is clearly important to be able to choose the size of the image you create.
Now in some versions that I used example for img2img, there is no choice of the size of the final image.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

No branches or pull requests

3 participants