search

zzzyuqing / DreamMat

[SIGGRAPH2024] DreamMat: High-quality PBR Material Generation with Geometry- and Light-aware Diffusion Models
MIT License
319 stars 13 forks source link

about .glb file #13

Open lijunnankman opened 5 months ago

lijunnankman commented 5 months ago

Can you share some .glb files about objects for demo ?

zzzyuqing commented 5 months ago

The models in the Objaverse dataset are in glb format, which you can download as needed. However, it's worth noting that our code currently only supports input in obj format.

lijunnankman commented 5 months ago

Hi, @zzzyuqing. Why you let the render image in [-1,1]?(https://github.com/zzzyuqing/DreamMat/blob/main/controlnet_train/diffusers_dataset.py#L59). Have you try to let the render image in [0,1]?

zzzyuqing commented 5 months ago

Hi, @zzzyuqing. Why you let the render image in [-1,1]?(https://github.com/zzzyuqing/DreamMat/blob/main/controlnet_train/diffusers_dataset.py#L59). Have you try to let the render image in [0,1]?

I referred to this script (https://github.com/huggingface/diffusers/blob/main/examples/controlnet/train_controlnet.py#L687) to train the controlnet. The images for stable diffusion training need to be within the range of [-1, 1].

lijunnankman commented 5 months ago

@zzzyuqing Thanks for reply. I also be confused on the training the contronet: it will product diffusion_pytorch_model.safetensors. But you share the contronet model is diffusion_pytorch_model.bin (not .safetensors). How can I convert diffusion_pytorch_model.safetensors to diffusion_pytorch_model.bin ?