Moving this up to the top, since it's important:
v1 does not support AIO models by default. (Models with VAE and/or CLIP baked-in). Your Anima/Krea2 models need to be placed in your models/diffusion_models folder to work with the (UNet) loaders in the v1 workflow.
This workflow has the following functions/features:
Text to Image
Image to Image (does not work well IMO)
Multiple Lora Loading
Global control for Seed
Wildcards
Upscale
Separate checkpoint, CLIP, and lora loaders for Upscale+Detailers
Detailers
Color Correction
Save Image with full resource metadata in CivitAI readable format
Support Policy
These workflows are provided as-is and without guaranteed support.
I do not provide support for third-party applications, modified versions, redistributed copies, reposts, bundles, or installations obtained from unofficial sources.
Inclusion of these workflows in another project does not imply my endorsement, involvement, or responsibility for that project.
Disclaimer
This workflow is not intended to purposefully generate any specific content. Responsibility for generated content falls on the user of the workflow.
Installation of ComfyUI and/or any related custom nodes for these workflows are at the installer's own risk.
Description
'Fixed' version uploaded at 1303 US Eastern time on 29 July 2026.
This just changes the 'default loaded' CLIP model on the USDU CLIP loader to the one for Anima.
Initial Release:
This is my current Anima workflow and is also compatible with Krea 2 Turbo.
I use Anima to generate the initial image and Krea 2 Turbo for the upscale.
Works with the following combinations: Anima + Anima, Krea 2 Turbo + Krea 2 Turbo, Anima + Krea 2 Turbo, and Krea 2 Turbo + Anima.
See the description below the image samples for more info. Some notes are included on the workflow canvas as well.
FAQ
Comments (10)
im getting an error
RuntimeError: mat1 and mat2 shapes cannot be multiplied (266x2560 and 1024x2048)
UltimateSDUpscaleCustomSample
Ultimate SD Upscale (Custom Sample)
Execution failed
Node threw an error during execution.
@hotpckxts726 no idea. I'm not a dev.
What models are you using? Anima for initial image and USDU? Are the CLIP loaders modes set correctly? Do you have the right text encoders loaded? This is the likely culprit.
UNet:
An Anima model
CLIP/text encoder:
qwen_3_06b_base.safetensors
CLIP Loader type:
stable_diffusion
UNet:
Krea 2 model
CLIP/text encoder:
qwen3vl_4b_fp8_scaled.safetensors
CLIP Loader type:
krea2
Error messages don't mean anything to me and if I can't recreate what you are experiencing, I can't troubleshoot it.
Otherwise, try posting your issue on the USDU project page or have AI help you troubleshoot. That's part of the reason I have all those links on the start of the workflow.
My own check with ChatGPT says the CLIP/text encoder mismatch is probably the cause. I can try updating the JSON, but the expectation is that you are checking and adjusting the models yourself.
Uploaded the replacement JSON.
@Gladas Yeah it was a text encoder issue. thank you
Hey so what is UNet loader? It looks like a checkpoint loader, but I'm not able to select any checkpoints.
It's for "diffusion_models", which is what Krea 2 and Anima models are. They have to be in the models/diffusion_models folder. This is mentioned below the sample images.
The reason for this is because the automatic switches in the workflow sometimes bug out when connecting CLIP loaders for text_encoder models. So I opted for the current loaders only instead of providing an option for AIO models.
The "Not so simple" series got nuked and this is the new Worflows series?
Yep, that's correct. If any of my workflows get nuked again, then I'll stop posting them.