Tighten ZeroGPU duration to 18s (measured 14s per rollout)

Gradio 6 API: theme/css move to launch(), Textbox buttons=['copy']

Take the CPU-only RNG branch when attaching the LoRA (ZeroGPU does not intercept torch.cuda.get_rng_state)

Add out-of-sight event modes (occluder / lights-out / camera pan), chunk schedule readout, all seven project-page presets

Delete pyproject.toml with huggingface_hub

Delete inference.py with huggingface_hub

Delete checkpoints.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload pipeline/checkpoints.py with huggingface_hub

Upload checkpoints.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload folder using huggingface_hub

initial commit