Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

Reposer+ with bypass & composition (Patreon)

Published:
2024-04-07 13:47:40
Edited:
2024-04-21 08:57:46
Imported:
2026-06
Tags:

Downloads

Content

Just for giggles, I wanted to see if I could add composition to Reposer+ and it turns out it does kinda work! It's best if the composition image "makes sense". For example, if the composition image is of an alien landscape, then something similar should be in your prompt like "Alien landscape in the background". Of course, you can use any image and prompt mix to see what weird things happen, or simply bypass the composition group altogether. It also works quite well if the composition image is pretty general, such as a forest or waterfall scene, to fill in where your character isn't :)

You can get the IP Adapter models from their huggingface. Model.safetensors is the Clip Vision model used by most adapters - https://huggingface.co/h94/IP-Adapter/resolve/main/models/image_encoder/model.safetensors

Files

Comments

Ammar Salem

Thank you for creating such great workflows. I have a small request, if possible. Could you please add an image to the workflow so we can better understand what's going on just by looking at it? Thank you!

Finn Negrello

Ohhh my god yes. Needed this. Literally just spent a couple hours changing your old workflow then found this in my emails???? EPIC!

Finn Negrello
>>134568180

I'm just going to deleted the two "image batch get" nodes for now and test... any advice would be super-duper appreciated!

Gluthoric

In your workflow where are you getting the model for 1.5 Image Encoder (ip_adapter_sd15/model.safetensors)

KING LIFER

Hey boss just looking at your workflow I saw you had a few things that is now updated. You can now use the unified IPAdapter loader to save memory and style and composition now has its own node. (XL) but you can also add them via separate nodes. Love your work.

Plaidam

Can I bypass/mute the Supporting Image and still have the rest of the workflow run properly? I was able to bypass Composition but tried several different ways. Still, I wasn’t able to do it with the Supporting Image. I’d like to sometimes use just Face and Pose; other times, I add Supporting Image and Composition back in for different runs, ideally using something like Fast Groups Muter.

Noah Fatah

First of al I want to thank you for your work I want to apologize if you already explained this but I am a novice and new to this, can you explain where I get all these adapters from and what part I load which one? Much appriciated!

Daljeet Peteson
>>135731674

Have the model in the clip_vision folder (even renamed it per your tidy naming conventions!) ... but immediately get this error on run: Error occurred when executing CLIPVisionLoader: Error while deserializing header: HeaderTooLarge Am I doing something wrong here ?

Daljeet Peteson
>>135731674

Solved "Header Too Large Error" (wrong clip-vision model). All models are now relinked, but now getting this new ControlNet error: Requested to load ControlNet Loading 2 new models !!! Exception during processing!!! linear(): input and weight.T shapes cannot be multiplied (77x2048 and 768x320) Any troubleshooting help greatly appreciated!