ComfyUI Tutorial Series Ep 70: Nunchaku Qwen Loras - Relight, Camera Angle & Scene Change
https://www.youtube.com/watch?v=9sD5Ekavjgo

https://redd.it/1ov8r21
@rStableDiffusion

Advice: Building Believable Customer Avatars with FaceSeek

I came up with a great trick here if you were to create story based content or faceless brands

Before inserting my AI generated faces in videos or thumbnails I normally use FaceSeek to see how real they look

I create characters in Midjourney and then I upload the image to FaceSeek. If it can't find the closest match I assume that the face is the most unique one, and therefore I can use it

If the face matching gets similar persons I change the AI prompt a bit until it looks good

Through this, they give me the option of not using AI faces that could look very much like a real person with whom
I am not endorsing a product it is just a great tool to content check if you are storytelling with AI images

https://redd.it/1ov9mov
@rStableDiffusion

XDiT finally release their ComfyUI node for Parallel Multi GPU worker.

https://redd.it/1ov9ns1
@rStableDiffusion

Continue to update the solution for converting 3D images into realistic photos in Qwen

https://redd.it/1ovccgz
@rStableDiffusion

Change Video Background Using Comfy UI | Works on Low V-RAM
https://youtu.be/eNTRzl20JWQ

https://redd.it/1ovmisi
@rStableDiffusion

[Qwen Edit 2509] Anything2Real Alpha

https://redd.it/1ovt8h8
@rStableDiffusion