I didnt sleep
Nahhh I just dont think it holds any weight when someone suggests using AI for something and then showcases a chat bot image, sure its doable, but unless zou got over 24gb Vram, you're gunna get low face counts, no retopolgy, bad textures without true albedo, normals or depth.
If someone wants to do it right, visit Hunyuan, get comfyUI and spend 7k on a PNY 6000 GPU.
If anyone has a beast machine and understands ComfyUI, I have an OpenPose app that loads the .libs and generates / tweens frames between poses, meaning we can up the frames.
Also looked into (thanks
@Jev ) missing mounts, again, posing the original mount images, using controlnet to get stzle reference from the missing mount armour and generating.
But its not consistent due to my lack of VRAM.