Update app.py

app.py

@@ -1,69 +1,56 @@
 import os
-import shutil
-import subprocess
-import tempfile
-import time
-from pathlib import Path
-from typing import List, Tuple
 
-import gradio as gr
-import numpy as np
-import open3d as o3d
-from PIL import Image
-
-APP_DIR = Path(__file__).parent.resolve()
-OUT_DIR = APP_DIR / "outputs"
-RUNS_DIR = APP_DIR / "runs"
-OUT_DIR.mkdir(exist_ok=True)
-RUNS_DIR.mkdir(exist_ok=True)
-
-
-def _run(cmd: List[str], cwd: Path, logfile: Path) -> Tuple[int, str]:
-"""Run a shell command, tee output to logfile, return (code, text)."""
-proc = subprocess.Popen(cmd, cwd=str(cwd), stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True)
-lines = []
-with logfile.open("a") as f:
-for line in iter(proc.stdout.readline, ""):
-f.write(line)
-lines.append(line)
-proc.wait()
-out = "".join(lines)
-return proc.returncode, out
-
-
-def _save_images_to(dirpath: Path, images: List[Image.Image], max_size: int) -> None:
-dirpath.mkdir(parents=True, exist_ok=True)
-for i, im in enumerate(images):
-im = im.convert("RGB")
-im.thumbnail((max_size, max_size))
-im.save(dirpath / f"im_{i:03d}.jpg", quality=92)
-
-
-def _colmap_pipeline(img_dir: Path, work_dir: Path, num_threads: int = 4) -> Path:
-"""Run COLMAP SfM+MVS. Returns path to fused point cloud (PLY)."""
-os.environ.setdefault("OMP_NUM_THREADS", str(num_threads))
-db = work_dir / "database.db"
-sparse = work_dir / "sparse"
-dense = work_dir / "dense"
-logs = work_dir / "logs.txt"
-
-sparse.mkdir(exist_ok=True)
-dense.mkdir(exist_ok=True)
-
-# 1) Feature extraction
-code, _ = _run([
-"colmap", "feature_extractor",
-"--database_path", str(db),
-"--image_path", str(img_dir),
-"--ImageReader.single_camera", "1",
-"--SiftExtraction.max_image_size", "2400",
-"--SiftExtraction.num_threads", str(num_threads),
-], cwd=work_dir, logfile=logs)
-if code != 0:
-raise RuntimeError("COLMAP feature_extractor failed. See logs.txt")
-
-# 2) Exhaustive matching
-code, _ = _run([
-"colmap", "exhaustive_matcher",
-"--database_path", str(db),
-demo.launch()
+def reconstruct(files: List[gr.File], max_img_size: int, threads: int) -> Tuple[str, str, str]:
+"""Main entry for Gradio. Returns (obj_path, obj_download, ply_download)."""
+ts = time.strftime("%Y%m%d_%H%M%S")
+run_dir = RUNS_DIR / f"run_{ts}"
+img_dir = run_dir / "images"
+run_dir.mkdir(parents=True, exist_ok=True)
+
+paths = [Path(f.name) for f in files]
+_save_images_to(img_dir, paths, int(max_img_size))
+
+fused = _colmap_pipeline(img_dir, run_dir, num_threads=int(threads))
+obj = _poisson_mesh_from_ply(fused)
+
+logs = run_dir / "logs.txt"
+if logs.exists():
+shutil.copy2(logs, OUT_DIR / f"logs_{ts}.txt")
+fused_out = OUT_DIR / f"fused_{ts}.ply"
+shutil.copy2(fused, fused_out)
+
+return str(obj), str(obj), str(fused_out)
+
+
+def ui():
+with gr.Blocks(title="Sparse2City3D: Urban massing from few photos") as demo:
+gr.Markdown(
+"""
+# Sparse2City3D
+Upload 5–30 photos of a street/building. The app runs **COLMAP** to reconstruct a dense point cloud and converts it into a coarse **OBJ** mesh for urban‑planning massing.
+"""
+)
+with gr.Row():
+with gr.Column(scale=2):
+files = gr.Files(label="Upload images", file_types=["image"], file_count="multiple")
+max_size = gr.Slider(800, 3200, value=2000, step=100, label="Max image size (px)")
+threads = gr.Slider(1, 8, value=4, step=1, label="# CPU threads")
+run_btn = gr.Button("Reconstruct 3D", variant="primary")
+with gr.Column(scale=3):
+model3d = gr.Model3D(label="3D preview (OBJ)")
+dl_mesh = gr.File(label="Download mesh (OBJ)")
+dl_ply = gr.File(label="Download point cloud (PLY)")
+
+def _wrap(files, max_size, threads):
+if not files:
+raise gr.Error("Please upload at least 5 images.")
+obj, obj_dl, ply_dl = reconstruct(files, int(max_size), int(threads))
+return obj, obj_dl, ply_dl
+
+run_btn.click(_wrap, inputs=[files, max_size, threads], outputs=[model3d, dl_mesh, dl_ply])
+return demo
+
+
+if __name__ == "__main__":
+demo = ui()
+demo.launch()