Mickmumpitz
patreon
[ADVANCED VERSION] Create HYPERREALISTIC Consistent AI Characters - FREE & LOCAL!
🕑 Added 2025-10-07 16:15:15 +0000 UTCAdvanced workflows for: https://youtu.be/PhiPASFYBmk
Thank you so much for your support and making these videos possible!
If you have issues, questions, or ideas please let me know on our discord channel!
🔗 Download the ADVANCED VERSIONS & EXAMPLE FILES below!
Have fun generating!
251119 UPDATE 3.7
Unfortunately, due to a ComfyUI update, I had to replace the image saver node, so the data set is no longer automatically saved in the correct folder. Instead, the images end up in the ComfyUI output folder. If you want to change this, you can manually enter an absolute path where your final data set will be saved.
![[ADVANCED VERSION] Create HYPERREALISTIC Consistent AI Characters - FREE & LOCAL!](https://img5.patrefans.com/storage/1/is/qf/563145-019e8569-d9fe-7aa6-b2ec-440946f8b5ee.png)
![[ADVANCED VERSION] Create HYPERREALISTIC Consistent AI Characters - FREE & LOCAL!](https://img5.patrefans.com/storage/10/jj/jk/563145-019e8569-da06-715e-9c09-a686df7339cd.jpg)
Comments
Jeremy Bedingfield
Wow, your attitude kind of stinks doesn't it? Not sure how you get through life being quite so 'entitled'.. Still if you switch the bottom section to 'on' and then hit run again I am sure you will come to realize just how rude you have just been. How about you thank a person for there efforts in creating something you blatantly couldn't?
Vojta
I can't see the dataset creation button, I even bought the advanced for it, SUPER THANKS FOR NOTHING
Jan
None of those workflows works.
Derek
Also, I just ran the workflow on my Mac M3 and some of the nodes require cuda so it's not compatible, unfortunately.
Derek
Can this workflow be modified to work with an SDXL model instead of Qwen?
Philipp
Could I enhance the workflow that I can put two images into it: one closeup of the person and one full body shot to get better results?
The Falxon
Thank you for this! This was just enough cheese and pepperoni to figure out how to make dough for the pizza. You sir are a genius, I'm well on my way in my AI journey.
LUC Club
in the 251007_MICKMUMPITZ_WAN-2-2-VID_ADV workflow, when loading Diffusion model at 41%, i getting Reconnection.. toast notification and the log shows: model weight dtype torch.float16, manual cast: None model_type FLOW Requested to load WAN21 loaded partially; 24056.31 MB usable, 24056.31 MB loaded, 3195.10 MB offloaded, lowvram patches: 331 (RES4LYF) rk_type: res_2s 100%|██████████| 3/3 [01:07<00:00, 22.59s/it]. the workflow pause there like forever. Am I hitting the VRAM limit? if it is what model should i change to so my RTX5090 32VRAM can run this workflow?
Tyler
Where is / what is the stock photography Lora?
Ricky Lawrence
Just joined. While I'm here, to help this community out, Claude will fix your errors for you, it has to know your setup though, everyone environments are different, be aware, now unto my question: got _CCC_3-7 workflow, installed on pc with 4090, is this adv.json a optional version?
LUC Club
film grain and seed generator nodes are missing?
Bruce Landry
So far 3 nodes don't load in the WF. I can load 2 (GGUF Loader, USOStyleRefference) of them manually but USO Reference Conditioning, I can nod find or load. What's I'm doing wrong here?
