Route xformers memory_efficient_attention through SDPA on Blackwell

Pin torch to 2.10.0 to match prebuilt xformers wheel ABI

Unpin timm (BiRefNet dynamic module needs timm.layers from 0.9+)

Add einops (StableNormal hub repo needs it)

Apply appendix bundle: build diff_gaussian_rasterization + nvdiffrast on GPU, preload libcudart.so.13

Pin torch to 2.11.0 to match CUDA 13.0 toolchain on Blackwell

Unpin torch/torchvision to get ZeroGPU's preinstalled torch (CUDA 13 ABI)

Drop unused gradio_litmodel3d (caps gradio<5)

Support new ZeroGPU (Blackwell) hardware: bump torch, swap to xformers, build nvdiffrast at runtime

Update requirements.txt

Update requirements.txt

Update requirements.txt

Update app.py

Update README.md

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

fix: Update Images

Merge branch 'main' of https://huggingface.co/spaces/Stable-X/Hi3DGen

fix: Update app

fix:Update app

fix: Update cube2mesh

fix: Add color render

fix: Add color rendering

fix: Update app

clean code