Guard _install_cuda_packages() startup call against quota exhaustion

Reduce startup GPU quota burn: skip CUDA pkg compile if already importable

Skip CPU texture bake fallback: raw mesh passes to rigging instead

Increase apply_texture GPU duration to 600s; use 512px UV in CPU fallback

Add CPU texture-bake fallback: xatlas UV-unwrap + numpy projection

Fix nvdiffrast: use CameraProjection context_type='cuda' (GL context fails on ZeroGPU)

Bypass TexturePipeline: use CameraProjection directly for UV baking

Replace bake_texture with TexturePipeline (mvadapter API change)

Remove hyperswap_1a_256.onnx from _ensure_ckpts: file not publicly hosted

Fix get_plucker_embeds_from_cameras_ortho: width is positional not keyword

Update apply_texture to MV-Adapter new API

Remove unused get_ipadapter_image import (removed from mvadapter.utils)

Fix TripoSG NaN cascade and remove stale autocast wrapper

Add api_name to run_full_pipeline and pshuman_face for Gradio API access

Replace runtime TripoSG string-patching with pre-patched files in patches/triposg/

Fix pymeshlab patch: always write file after import removal

Integrate PSHuman locally — multi-view diffusion in-process, no remote calls

Replace pymeshlab with trimesh in inference_triposg.py (no py3.13 wheels)

Move CUDA compilation into @spaces.GPU function — nvcc unavailable at APP_STARTING

Fix exec()+locals() for all 4 xinntao packages (basicsr/realesrgan/gfpgan/facexlib)

Detect CUDA home dynamically instead of hardcoded /cuda-image/... path

Fix basicsr setup.py patch: keep _ns inside with-block so f is still open

Pre-install meson-python+ninja before --no-build-isolation step

Fix basicsr Python 3.13 exec()+locals() crash at runtime startup

Fix basicsr/realesrgan/gfpgan build: move to runtime --no-build-isolation install (setup.py imports torch)

Port MeshForge features to ZeroGPU Space: FireRed, PSHuman, Motion Search

Patch all-zero alpha: fallback to full-foreground instead of raising ValueError

Fix empty_contours_guard patch: use gray_image (correct param name) not alpha

Patch find_bounding_box: guard empty contours (blank alpha fallback to full image bbox)

Fix queries dtype cast in inference_utils.py; remove autocast (caused NaN/no isosurface)

Wrap run_triposg in torch.autocast float16 to fix Float/Half dtype mismatch

Fix RMBG loading with torch.device(cpu) context; fix None fallback alpha shape [1,1,H,W]

Revert to 1038lab/RMBG-2.0 (briaai is gated)

Fix RMBG-2.0: use briaai/RMBG-2.0 (official) with standard load pattern

Patch image_process.py rmbg_net None guard; fix RMBG-2.0 loading with device_map

Add missing deps: sentencepiece, spandrel, imageio; fix SKEL moderngl-window==2.4.6 conflict

Fix diso undefined symbol (clone with --recurse-submodules); remove RMBG preview/controls from UI

Fix triposg: clone source at runtime (no setup.py); remove bad pip install attempt

Bust build cache; pin numpy>=2 at build + runtime upgrade after chumpy chain

Pin moderngl-window>=3.0.0 (supports numpy>=2); remove manual numpy upgrade

Upgrade numpy>=2 after chumpy install chain to satisfy opencv-python-headless

Fix Gradio 6 theme warning; ensure numpy>=2 before chumpy dep chain pulls old version

Fix CUDA compile: set TORCH_CUDA_ARCH_LIST=8.6 and CUDA headers path for A10G

Fix: --no-build-isolation for CUDA packages; numpy>=2 for opencv-python-headless

Remove explicit chumpy from runtime install; let hmr2 pull it as transitive dep

Move mvadapter+stablenormal to runtime --no-deps (over-pinned conflicting deps)

Move hmr2/skel/chumpy to runtime --no-build-isolation; document all ZeroGPU constraints

ZeroGPU: compile CUDA packages at runtime startup (no nvcc in build container)

Add nvidia-cuda-toolkit to packages.txt; restore CUDA packages to requirements.txt; clean up app.py

Move detectron2 to runtime install (CUDA build dep)