# GenRecon web app — phone video → walkable 3D (PWA)

Front-to-back app for **genrecon.suby.pl**: upload a phone video → server reconstructs a 3D
scene with [GenRecon](https://github.com/kasothaphie/GenRecon) → walk through it first-person
in the browser. Built to drop onto the `genrecon` LXC from issue **i-opavuyaq**.

```
backend/    FastAPI service: upload → ffmpeg → COLMAP → reconstruct_scene → chunked_to_glb → scene.glb
frontend/   PWA: upload UI, live progress, three.js first-person walkthrough (Octree collision)
            vendor/ = three.js r161 (vendored, offline-capable) · demo.glb = bundled sample scene
sample/     a sample walkable scene.glb (also used as the MOCK output)
```

## Run locally WITHOUT a GPU (mock mode)

```bash
python3 -m venv .venv && . .venv/bin/activate
pip install -r backend/requirements.txt
GENRECON_MOCK=1 GENRECON_JOBS_DIR="$PWD/_jobs" \
  GENRECON_FRONTEND_DIR="$PWD/frontend" GENRECON_SAMPLE_GLB="$PWD/sample/scene.glb" \
  uvicorn app:app --app-dir backend --host 0.0.0.0 --port 8000
# open http://localhost:8000  → "Zobacz przykładową scenę 3D", or upload any file (mock ignores content)
```

## Deploy on the `genrecon` machine (real pipeline)

Prereqs from issue i-opavuyaq Step 2 (GenRecon at `/opt/genrecon/src`, checkpoints at
`/opt/genrecon/ckpts`, conda env `genrecon`, COLMAP + ffmpeg installed).

```bash
git clone <this-repo> /opt/genrecon/app
pip install -r /opt/genrecon/app/backend/requirements.txt      # into the system/base python
cp /opt/genrecon/app/backend/genrecon-app.service /etc/systemd/system/
systemctl enable --now genrecon-app
# then add the Caddy route on the suby edge:  genrecon.suby.pl → <genrecon-ip>:8000
```

Service config lives in `backend/genrecon-app.service` (env vars: jobs dir, GenRecon paths,
conda env, `GENRECON_CUDA_DEVICE` — pin to a free GPU, e.g. `5`). Set `GENRECON_MOCK=1` there to
serve the demo scene before the model is ready.

## API

| Method | Path | |
|---|---|---|
| POST | `/api/jobs` | multipart `video` → `{id}` |
| GET | `/api/jobs/{id}` | `{status, stage, progress, message, error}` |
| GET | `/api/jobs/{id}/scene.glb` | the reconstructed scene |
| GET | `/api/jobs/{id}/log` | raw pipeline log |
| GET | `/api/health` | |

## ⚠️ Verify against the real GenRecon `Iphone` layout

`backend/pipeline.py` assembles `<job>/images/` + `<job>/colmap/{cameras,images,points3D}.txt`
(what the README implies). During the i-opavuyaq Step-2 smoke test, confirm `reconstruct_scene.py
--mode Iphone` actually finds that layout; adjust `run_colmap` / `_reconstruct_cmd` if not.

## Deploy: web-optimized walkthrough (optional but recommended)

Reconstructed meshes are 100s of MB. `backend/optimize_glb.sh` (called automatically
by the glb stage) shrinks them to a browser-loadable size (simplify + webp). It needs
a one-time local gltf-transform install:

```sh
bash backend/setup_gltftool.sh   # installs to /opt/genrecon/gltftool
```

Without it the pipeline still works — it just serves the full-res `scene.glb`.
Tune via `GENRECON_SIMPLIFY_RATIO` (default 0.12) and `GENRECON_TEX_SIZE` (2048).