Various ControlNet experiments for Anima Preview 3.
Inference:
https://github.com/Haoming02/sd-webui-forge-classic/commit/612fe1c8a8f87482c8889eb9183dbe2443eb2ef2
https://github.com/kohya-ss/ComfyUI-Anima-LLLite
Training:
https://github.com/kohya-ss/sd-scripts/tree/feat-anima-lllite
Description
FAQ
Comments (11)
ohhhhhh
Tried it and I'm unsure how to make it work. Sent an image through an AIO aux preprocessor with pyrahcanny, then into the apply controlnet node specific for anima with your canny file and it did nothing. Could you provide instructions on usage?
Workflow uploaded + preview + safetensors reuploaded with ComfyUI metadata. Please try again.
@inpaint I tried it again. I was unable to get the metadata from the files but I had my own workflows already, adapted part of yours into the workflow. It seems to be working, I'm using .9 on str and end percent personally. Good job! In the workflow you posted, I'd personally add a "get image size" node and link the width and height outputs to an "empty later image" node, canny and similar like using the exact same image size. Regardless, its working.
EDIT: Issue fixed with update, and workflows included, ignore this.
Does not work. A workflow or some examples would be swell.
Workflow uploaded + preview + safetensors reuploaded with ComfyUI metadata. Please try again.
@inpaint Alright friend, now I got it working, seems it might have been the safetensors because once I redownloaded them everything clicked into place. Works very well. You've done god's work here, I can't tell you how many times I've been hurting for a good CN on an Anima project.
Where should I place this file to get it working? The GitHub page is down.
Anima ControlNet LLLite.json. I’m using Forge Neo.
JSON workflows are for ComfyUI only. Check the image in the post for usage in Forge Neo. You place the Safetensors file in the models/ControlNet folder and load it from the ControlNet field. Ensure that your Forge Neo is updated to at least the commit where support was added, i have linked the commit in the post


