How to use it in ComfyUI?
#2
by
hellowind666
- opened
How to use it in ComfyUI?
hope to use the Edit in Comfy too
but... how minimum hardware for work?
but... how minimum hardware for work?
I managed to run on RTX 5070 16GB. I had to fix file a pipeline_longcat_image_edit.py and forcibly manage models in GPU memory.
The scheme turns out to be something like this:
- Loading models into RAM.
- Loading Qwen2.5-VL into the GPU and executing it.
- Unloading Qwen2.5-VL from GPU.
- Loading LongCat-Image-Edit into GPU.
- Performing LongCat-Image-Edit
RAM on PC - 64GB.
Work time - 30/30 [01:17<00:00, 2.58s/it
