Space: set app_file to app.py (full NeAR); update README notes
Browse files
README.md
CHANGED
|
@@ -6,7 +6,7 @@ colorTo: indigo
|
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: 6.9.0
|
| 8 |
python_version: "3.10"
|
| 9 |
-
app_file:
|
| 10 |
pinned: false
|
| 11 |
license: apache-2.0
|
| 12 |
short_description: "Relightable 3D from one image: SLAT, neural renderer, HDRI"
|
|
@@ -48,7 +48,7 @@ This repository combines:
|
|
| 48 |
|
| 49 |
## ZeroGPU Runtime Notes
|
| 50 |
|
| 51 |
-
- The Space
|
| 52 |
- **`app_hyshape.py`** (when used as entry): defaults to **`NEAR_HYSHAPE_GEOMETRY_CPU_PRELOAD_AT_START=1`** β background **CPU** Hunyuan load at start; **Generate Mesh** pays **GPU move + inference** in `@spaces.GPU`.
|
| 53 |
- The full `app.py` Space keeps **page-load image defaults** and **HDRI preview** on lightweight CPU paths so the first page visit does not spend the first ZeroGPU allocation on model initialization.
|
| 54 |
- **`app.py`**: optional background **CPU** preload of Hunyuan + NeAR (`NEAR_MODEL_CPU_PRELOAD_AT_START`); **`@spaces.GPU`** callbacks move each pipeline to CUDA once, then run inference. **gsplat** is used when the pipeline renders (no separate app-level warmup pass).
|
|
@@ -102,9 +102,9 @@ Compared with a standard image-to-3D pipeline, NeAR focuses on:
|
|
| 102 |
Key files and directories:
|
| 103 |
|
| 104 |
- `example.py` β minimal end-to-end inference example.
|
| 105 |
-
- `
|
| 106 |
-
- `
|
| 107 |
-
- `
|
| 108 |
- `setup.sh` β environment setup helper.
|
| 109 |
- `checkpoints/` β local pipeline configuration and model checkpoints.
|
| 110 |
- `trellis/pipelines/near_image_to_relightable_3d.py` β main NeAR inference pipeline.
|
|
|
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: 6.9.0
|
| 8 |
python_version: "3.10"
|
| 9 |
+
app_file: app.py
|
| 10 |
pinned: false
|
| 11 |
license: apache-2.0
|
| 12 |
short_description: "Relightable 3D from one image: SLAT, neural renderer, HDRI"
|
|
|
|
| 48 |
|
| 49 |
## ZeroGPU Runtime Notes
|
| 50 |
|
| 51 |
+
- The Space **`app_file`** in the YAML header above is **`app.py`** (full NeAR app). Switch to **`app_gsplat.py`** or **`app_hyshape.py`** for isolated diagnostics.
|
| 52 |
- **`app_hyshape.py`** (when used as entry): defaults to **`NEAR_HYSHAPE_GEOMETRY_CPU_PRELOAD_AT_START=1`** β background **CPU** Hunyuan load at start; **Generate Mesh** pays **GPU move + inference** in `@spaces.GPU`.
|
| 53 |
- The full `app.py` Space keeps **page-load image defaults** and **HDRI preview** on lightweight CPU paths so the first page visit does not spend the first ZeroGPU allocation on model initialization.
|
| 54 |
- **`app.py`**: optional background **CPU** preload of Hunyuan + NeAR (`NEAR_MODEL_CPU_PRELOAD_AT_START`); **`@spaces.GPU`** callbacks move each pipeline to CUDA once, then run inference. **gsplat** is used when the pipeline renders (no separate app-level warmup pass).
|
|
|
|
| 102 |
Key files and directories:
|
| 103 |
|
| 104 |
- `example.py` β minimal end-to-end inference example.
|
| 105 |
+
- `app.py` β full NeAR Gradio app; default **`app_file`** for the Hugging Face Space (see README YAML header).
|
| 106 |
+
- `app_gsplat.py` β **gsplat-only** ZeroGPU probe; set `app_file: app_gsplat.py` to test rasterization in isolation.
|
| 107 |
+
- `app_hyshape.py` β HyShape-only diagnostic; set `app_file: app_hyshape.py` for Hunyuan geometry in isolation.
|
| 108 |
- `setup.sh` β environment setup helper.
|
| 109 |
- `checkpoints/` β local pipeline configuration and model checkpoints.
|
| 110 |
- `trellis/pipelines/near_image_to_relightable_3d.py` β main NeAR inference pipeline.
|