Open 2621690255 opened 3 months ago
I'm interested too, I've tried to upscale the input image to 4096x4096 and then run inference with parameter --texture-resolution=4096. But other than a quite big increase of GPU memory usage, I've not noticed any differences
Hi, all! stable-fast-3d have any parameters that can improve the quality of the generated model? such as increasing the number of triangles used in inference.