Consistant Multi Character AI Digital Models! (ComfyUI Tutorial)

Consistant Multi Character AI Digital Models! (ComfyUI Tutorial)

Say goodbye to duplicate faces and hello to unlimited character possibilities! This ComfyUI tutorial unlocks the power of IPadapter for generating consistent, unique characters with ease. Learn to create 2, 3, or even more distinct AI characters with consistent features, banish face-swapping struggles, and refine your designs for stunning multi-character AI art. Dive into the world of ComfyUI, IPadapter, and advanced character creation techniques – subscribe for weekly AI art breakthroughs!

[Aiconomist]

Let me know if you want part 2 where we can enhance face features, use controlnet and upscaling 👇👇

[itscout]

Please do a second advanced part, this part was helpful however the faces are not the best. Thank you for the great tutorials!

[lgpyrogod]

This is an amazing video!!! (I am 30 second in) haha love the work, thanks for sharing this workflow and methods!

[beetwing]

Thanks and please yes, do another part or even more. Good stuff!!

[nacho]

Your tutorials are always really well done and interesting. Thanks a lot!

[DmitriySokhach]

@Aiconomist oh man! Thank you! Watching 2nd video of yours -> subscribed immediately and seems now inspired to learn ComfyUI more!
Such a great and simple explanation!

[digital]

I don't think you need to load the model every time. You can connect the IP Adapter advanced nodes directly from the model output to the model input when chaining.

[QuizTimeeeeee]

great video love the work!! please make Part2 and maybe part3 with more details on changing clothes and poses of the models

[Designing-hcpz]

can you please make a video on how to swap face in comfy ui of single person/character cause using Fooocus web ui's faceswap didn't come out that much as expected...

[niemandsohn]

Cool video, but is it possible to do this with character loras instead of Reference images?

[Paulo-utli]

I’ll try to replicate this workflow using InstantID and open pose and depth CN to control positioning. mean while it would be good to see a video of you trying to create this. thanks!

[christofferbersau]

Thank you, but isn't faceid better for faces?

[timothywells]

thanks for the great tutorial as always! i really like the way you explain everything clearly and methodically! It a yes to a part 2 from me? Also i have a question, in the course will you be covering SDXL lora training? or if not would you be able to make a tutorial for getting good consistent photorealistic results?, So far I've tried to train 6 versions using a number of different guides and my results have been.... shall we say less than optimal.

[hkkjkvv]

Hey buddy, love your content...
Just a question....I'm making an ai model....face seems to coming perfect but how do I control the body measurements consistency?

[justincanton]

Thanks, is there a web based version of this I can use?

[patrickstonks]

Ive been searching for awhile, is there a way to use segment anything to create masks?

As in, create a generated image of multiple characters, pass that generation into a segment node, create masks for each character, then pass into the IP Adapter?

[shshsh-zyqq]

I am trying to make my own model. With this way, will the model's face and body type show the consistency? Amazing, tutorial. Thank you so much. I have already subscribed you :)

[thays]

This doens't work for PONY though, correct? I got IP Face adapter working great for any 1.5 image in Forge. In Pony, it seems to not really work at all. Is there a way to apply this to PONY models? Also, is there any way to better utilize ADetailer than the left-right scan and using [SEP] in the aDetailer prompt for faces? Or is that still the only/best option for applying adetailer?

[AndreGenesini]

I would like to buy the AI Models course, I dont mind it it will take one month or even more to complete, thanks

[CyberwizardProductions]

you said to rename the models, and to follow what it says on the github page. well for the clipvision models it just says download and rename NOT what to rename it to