Ruizhi Shao
Ruizhi Shao
Set the diffusion model to evaluation mode as a precaution. Currently, The training mode and evaluation mode will not affect the model's performance since stable diffusion utilizes GroupNorm in U-Net...
- Add experimental configs (controlnet-normal/canny, instructpix2pix, control4d-multiviews, magic3d-coarse/fine-multiviews). - Add multiview dataset which supports loading multi-view images and OpenCV camera model of NeRFStudio. - Add ControlNet and Instructpix2pix as new...
Using prompt [RAW photo, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3] and negative prompt [(overexposed, underexposed, out of focus, deformed iris, deformed pupils,...
Support utilization of volsdf density function in neus volume renderer. I have discovered that VolSDF is more stable than NeuS in sparse multi-view reconstruction tasks and 3D generation using VSD...
The GPU memory usage during training with high-resolution settings is currently unstable, leading to OOM errors. The primary factor contributing to this issue is the grid_prune function in Nerfacc, resulting...
- [x] ControlNet DU guidance - [ ] InstructNeRF2NeRF DU guidance - [ ] StableDiffusion DU guidance - [x] VSD DU guidance - [ ] DeepFloyd DU guidance - [...
It appears that the SMPLX coordinate system is not aligned with the threestudio coordinate system, resulting in incorrect view-dependent prompts during generation. For instance, an azimuth angle of around -180...