fix: use js_on_load for queue polling, direct CORS fetch to gradio share links

update app.py

fix: prevent page height growing infinitely and open redirect in new tab

chore: update app.py, add app_bak.py, update requirements and autotune cache

Fix: total_mem -> total_memory

Fix: move GPU diagnostics into init_models() to avoid early CUDA init before spaces import

Add GPU/CUDA diagnostics at startup

Fix progress tracking: replace SSE+queue with file-based polling for HF Spaces compatibility

fix: progress not showing on 2nd extract - recreate queue and re-fetch ref in SSE loop

feat: display FOV and distance in preview panel

Fix race condition: disable generate during preprocess, use preprocessed image, thread-safe progress

Use local RMBG model, fix device placement, improve tqdm progress tracking

Fix device mismatch, use remote RMBG client, improve progress tracking, translate comments to English

refactor: use remote BRIA-RMBG-2.0 for background removal (no local GPU needed)

fix: add @spaces.GPU decorator to preprocess for ZeroGPU compatibility

feat: add clear button, SSE progress tracking, multi-session support

UI updates: Pixal3D V1.1 branding, camera settings, readme, ref thumbnail, GPU duration 240s, fix EXR path

gradio server (#3)

Update title, add project page link, add autotune cache

Add ZeroGPU support: import spaces first, add @spaces.GPU decorators, add flex_gemm env vars

Update app.py

Update app.py

Upload folder using huggingface_hub