Sign In
home
models
images
videos
posts
articles
bounties
tools
challenges
events
shop

Stable Cascade Basic Workflow [ComfyUI] ArgusV10

39
driftjohnson
Grandfather Badge
Feb 25, 2024
news
comfyuireleasedcascadecomfy
Stable Cascade Basic Workflow [ComfyUI] ArgusV10

Update: v82-Cascade Anyone

The Checkpoint update has arrived !

New Checkpoint Method was released. All Workflows were refactored.
https://huggingface.co/stabilityai/stable-cascade/tree/main/comfyui_checkpoints
put both inside /models/checkpoints/


v30-txt2img
- updated workflow for new checkpoint method.
- Text 2 Image.

https://civitai.com/models/310409?modelVersionId=351798


v40-img2img
- updated workflow for new checkpoint method.

- Image to Image with prompting, Image Variation by empty prompt.
https://civitai.com/models/310409?modelVersionId=351805

v42-img2img
- updated workflow for new checkpoint method.
- added Lora Loader for testing new trained Lora's

- Image to Image with prompting, Image Variation by empty prompt.
https://civitai.com/models/310409?modelVersionId=351805

v50-img2vision
- updated workflow for new checkpoint method.

- Image to CLIP Vision + Text Prompt.

https://civitai.com/models/310409?modelVersionId=351809


v60-img2remix

- updated workflow for new checkpoint method.

- Multi-Image to CLIP Vision + Text Prompt.

https://civitai.com/models/310409?modelVersionId=351815

v70-img2remix-faceswap

- updated workflow for new checkpoint method.

- Multi-Image to CLIP Vision + Text Prompt.
- Use an HD Face image with Reactor.

https://civitai.com/models/310409?modelVersionId=353109

v82-Cascade-Anyone

- Add high quality Face image with 4 character reference images using prompts.

- built from v70 to estimate custom characters without training or Cnet

https://civitai.com/models/310409?modelVersionId=353583

I want to stress that you MUST update your comfyUI to the latest version, you should also update ALL your custom nodes because there is no way to know which ones might have affect the UNET, CLIP and VAE spaces which cascade is now using to generate our images.

In addition, i have disabled a lot of custom nodes i did not need on that run. it's easy, just add ".disabled" to the folder name. This is what the button does in the manager. it's very easy to "switch off" some custom nodes in this way.

~

Everything Below applied to the UNET/CLIP Method for loadings all the models here: official repo: https://huggingface.co/stabilityai/stable-cascade

~ i will leave the Early method here, for anyone wishing to use it :)

The UltraBasic Stable Cascade UNET Workflows for ComfyUI:

reddit: https://www.reddit.com/r/StableDiffusion/comments/1auzhzm/comfy_img2img_with_stable_cascade/

these older workflows were deprecated on day 4 by a new method, however still work fine.

v10 = txt2img Stable Cascade here: https://civitai.com/models/310409?modelVersionId=348385

v12 = v10 txt2img without custom nodes: https://civitai.com/models/310409?modelVersionId=351470
v16 = img2img (stage C) Stable Cascade Workflow here: https://civitai.com/models/310409?modelVersionId=351400
v17 = v16 img2img without custom nodes for scaling: https://civitai.com/models/310409?modelVersionId=351464
v18 = v16 img2img (stage B and C ) now supported by new default node: https://civitai.com/models/310409?modelVersionId=351658

You can squeeze it only any GPU if you use the correct combination.
These notes are in the Workflow also ;)

Cascade Combos:

  • stage_b + stage_c ~ 22GB

  • stage_b_bf16 + stage_c_bf16 ~ 12GB

  • stage_b_lite + stage_c_lite ~ 8GB

  • stage_b_lite_bf16 + stage_c_lite_bf16 ~ 5GB

I put together to paths you need to put all the models in case you had to manually DL each of them, due to a poor connection or whatever :)

Huggingface has the models we need, follow the chart below to find where they go

https://huggingface.co/stabilityai/stable-cascade

Text Encoder

ComfyUI Path: models\clip\Stable-Cascade\

HF Filename: /text_encoder/model.safetensors

text encoder CLIP = 1.39GB

Stage C

ComfyUI Path: models\unet\Stable-Cascade\

HF Filename: stage_c.safetensors

  • stage_c = 14.4GB

  • stage_c_bf16 = 7.18GB

  • stage_c_lite = 4.12GB

  • stage_c_lite_bf16 = 2.06GB

Stage B

ComfyUI Path: models\unet\Stable-Cascade\

HF Filename: stage_b.safetensors

  • stage_b = 6.25GB

  • stage_b_bf16 = 3.13GB

  • stage_b_lite = 2.8GB

  • stage_b_lite_bf16 = 1.4GB

Stage A

ComfyUI Path: models\vae\Stable-Cascade\

HF Filename: stage_a.safetensors

stage_a = 73.7mb

Effnet Encoder

ComfyUI Path: models\vae\Stable-Cascade\

HF Filename: effnet_encoder.safetensors

img2img VAE encoder = 81.5mb

39

Comments

DuckFace's Avatar
DuckFace
Day of the Dead Badge

good work man!

aldeco's Avatar
aldeco

Thanks for explaining the differences between the models.

I have Nvidia Quadro P2000 video card that has 5Gb VRAM. To my surprise I was able to use all available models up to stage_c and stage_b. The time to generate one 1024x1024 image is about 2 minutes (stage_c - 20 steps, stage_b - 5 steps).

346542742384560934687's Avatar
346542742384560934687

SUGGESTION: Rename things to: "UNETLoader C", "UNETLoader B" and "Load VAE A"
Works on 1080 Ti.
I had to put the files in the ComfyUI folder. It wouldn't work in the Automatic1111 subfolders.

duskfallcrew's Avatar
duskfallcrew

Suggestion: If the training scripts are available, see if you can make a tutorial on how that works, even if ti has to be CLI, i know they have a "instructions" on the thing, but it kinda makes my head hurt XD -- I'm not wanting to fine tune the whole thing just do loras

driftjohnson
Grandfather Badge
op

I believe that OneTrainer has added support for training Lora with Cascade, I'm looking into it and will report back :D and yes the advantage of Cascade is because it had three stages we only train on Stage C, this will make training much less painful i think.

duskfallcrew's Avatar
duskfallcrew

OOh wow o.o ok, well hopefully one trainer isn't windows only XD

driftjohnson
Grandfather Badge
op

it's windows and linux afaik

duskfallcrew's Avatar
duskfallcrew

So eventually i could get it to work on Vast, just maybe not this week LOL

pdogdappaule's Avatar
pdogdappaule

thanks for the great workflows and putting me onto stable-cascade... For some reason which i have yet to figure out, the v30-txt2img works perfectly, but each of the other cascade versions give me a blank grid (no nodes visible at all) but no errors being logged... Any ideas why that would happen? I'm using the stage_b + stage_c checkpoints.

driftjohnson
Grandfather Badge
op

unless they are just off the edge of the screen i'm not sure. They are all originally part of a single workflow that has them all in one. i'll release that at v100. you can disable sections with ctrl+b so it's easier to use when testing. I'll take a look later