Input a character image, generate multiple images with different perspectives
You can directly click the link below to try out the AI application for this workflow. No prior knowledge of ComfyUI is required; it's ready to use right away. If the results are satisfactory, you can deploy it to your local environment.
https://www.runninghub.ai/ai-detail/2007425273741971458/?inviteCode=rh-v1315
More workflow parameter settings, please refer to my settings on rh:
https://www.runninghub.ai/post/2005880336003334146/?inviteCode=rh-v1315
Sign up via this link claim 1000 RH Coins and generate tons of images/videos for FREE!
My AI models/apps are on RH now! Try them out & support my work — every click counts. Thank you!
Thank you all for your support. Your support is the motivation for me to update more interesting and useful workflows. Thank you so much!!!
Description
FAQ
Comments (17)
where I can find the openpose images you are using in this workflow ?
Just download it and extract it
@Zovetry Some are missing
This is fantastic. Running it locally with 16GB Vram, but I had to use the merged model/lora for it to fit.
Hi!
I just wanted to say that I really like your workflow ,everything works great on my side and the results are excellent !!
I had a quick question about character proportions and expressions. I work with different types of characters: some are smaller or shortstack, others have certain body parts that are more developed, and I also pay a lot of attention to facial expressions.
Since this is something I’m still learning, I was wondering if you have any advice on how to help the workflow take these variations into account more accurately and consistently.
Thanks again for sharing your work — it’s been very helpful!
Nice. Improving back view consistency will make it much better.
CR Text node simply will not download for me, and I cannot find it online. Please help
You don't need it. It's simple String (Multiline) node it's already inside ComfyUI. Comfyroll is tons of nodes and I get problems with it sometimes. My advice : Stay away from Comfyroll nodes as much as you can
Not easy to use, and the whole workflow took 21min to finish on my RTX 3090, that's very slow.
It would be nice to have a screenshot for the openpose placement, otherwise even if we have the openpose files the only way to know where to put each pose is to watch the Youtube video again and try to do a pause when it's showed. Not straightforward at all
about 1hr on 5070ti
that weird, it took me about 6 min with a 3090 with the fp8 mixed model and 4steps lighting lora
Nice workflow. 17m36s with my RTX 3070.
Model used was a gguf Q8.
excellent workflow, Some of the example images are missing and i was having trouble with the int node but i was able to just replace it with the ComfyUI built in node. right click>Add Node>utils>primitive>Int. then just run the strings to the new node instead of the old one. Also you need to reset the value in the Int node to 1024 each time you run it. Thanks for this amazing workflow!
Hi, the front 45 degree side view one insists on making different cloth. Is there a fix to this?
same issue
It's not perfect, but I changed the prompt for that box (the first red one) to make it closer to the prompt the rear view 45° image uses. I translated the rear view prompt to english, changed it to front view, then translated it back to chinese.