Models ignore input images
Hello,
I have an issue I tried to solve for over 4 hours now. I used the workflow from the ComfyUI-Wiki (https://comfyui-wiki.com/en/tutorial/advanced/hunyuan-image-to-video-workflow-guide-and-example).
No matter what I do, the output will often have little in common with the original picture or change the image in a way that doesn't seem to be natural compared to other outputs I have seen. Do I expect too much and just have to generate a batch of picture of is there something wrong with the workflow?
Could you provide a own workflow for your models? I have probavly tried every single one I could find on the Internet. Either they crash (12GB VRAM card) or they will ignore the input image.
I have everything installed (ComfyUI manager) that the workflow tries to preload.
Kind regards.
Could you try the one from the official example page, namely the "v2 replace" one meant for the fixed models? https://comfyanonymous.github.io/ComfyUI_examples/hunyuan_video/#v2-replace
All of the models I upload are just tested with the workflows from there, with 0 changes other than replacing the default model loader with the GGUF unet loader, and optionally the clip loader if you use a quantized T5 model.
As for the OOM issue, there was an update to ComfyUI-GGUF today that might fix it.
Could you try the one from the official example page, namely the "v2 replace" one meant for the fixed models? https://comfyanonymous.github.io/ComfyUI_examples/hunyuan_video/#v2-replace
All of the models I upload are just tested with the workflows from there, with 0 changes other than replacing the default model loader with the GGUF unet loader, and optionally the clip loader if you use a quantized T5 model.
As for the OOM issue, there was an update to ComfyUI-GGUF today that might fix it.
After trying the workflow you provided, I received OOM issues too. So I decided to entirely remove ComfyUI and start from 0. Also updated the ComfyUI-GGUF. Now both your workflow work and the other workflow seems to follow the input image. I'm not sure what caused the issue. But I probably messed something up in ComfyUI, as I haven't been using it for too long. It unfortunately is quite confusing for someone that has only been using WebUI previously.
Thank you for your help, all runs fine now.