The method to convert v3 into Diffusers format
Hello, I have been greatly inspired by your work. I am currently working on creating a new ControlNet following the x-flux methodology. I have successfully created a ControlNet, but unfortunately, I have not been able to convert it into a Diffusers-compatible format.
While x-flux includes inference code, my goal is to make the ControlNet I created using x-flux compatible with the Diffusers library.
During this process, I noticed that you successfully converted canny-v3 into a Diffusers version. Could you share the steps or methods you used to achieve this?
When I inspected the safetensors file, I found that the v3 model contains 86 weight files, whereas the Diffusers version contains 102 weight files. Could you explain where this difference comes from?
https://github.com/huggingface/diffusers/discussions/10258
I found solution!