Get the latest global workflows on my RunningHub homepage and run them free on RTX4090.
Sign Up with my invite link for 1000 bonus points + 100 daily:
Run Directly:
Please credit and link to my homepage if reprinting.国内合作加Q：549791525
It uses the latest nanchaku qwen edit 2509 model, which has detailed instructions.
Since you have to get the prompt word through the api, your picture will not work normally if it is R18 +.
Description
FAQ
Comments (6)
nothing to do with seedream or nano banana, qwen 2509 totally inconsistent in faces
My test results show that most faces hold well. You can try it with the original BF16.
Very noob question but can you tell me what
"When running locally, replace the "Runninghub LLM API Node" with your own commonly used hybrid large model API node."
means? I understand I need a new node, but I am unsure what type of node it is. I am running locally but I never used a hybrid large model api node. Can you recommend one, I have Ryzen 9950x and rtx5090 locally if that helps.
Gemini
Get a Huggingface account, set your PC config so you'll know the compatibilities of the models before you get them. Then get the Wan2.2 version you can run
@joxaro9187376 This comment doesnt help me in the least, seeing as I was asking about a node issue, specifically "Hybrid large model api node" which has nothing to to do with wan, seeing as its for a character gen dataset workflow.
I appreciate your enthusiasm, but I wanted to know about nodes within comfyui. I already have everything needed to run wan/flux/chroma whatever locally, but ive never messed with an api node