Prompt Extension does not depend on input image.
#113
by
neuralityAI
- opened
Hi! I've connected the prompt extender like so, but one thing does not make sense to me. The input image should be fed to the prompt extender as well right when running I2V workflows? Because we can not run I2V Imagination properly if the image is never an input to the node?
comfy/ComfyUI/custom_nodes/ComfyUI-WanVideoWrapper/qwen/system_prompt.py
The above files show that you copied the prompts for the different settings. But for imagination and rewrite we should probably use the VL models with image input as well.