Update app.py

app.py

@@ -48,7 +48,7 @@ def process_image(image_path):
     except Exception:
         gltf_path = create_3d_obj(np.array(image), depth_image, image_path, depth=8)
     
-    return [Image.fromarray(depth_image), gltf_path, gltf_path]
+    return [Image.fromarray(depth_image), gltf_path]
 
 def create_3d_obj(rgb_image, depth_image, image_path, depth=10):
     depth_o3d = o3d.geometry.Image(depth_image)
@@ -60,13 +60,11 @@ def create_3d_obj(rgb_image, depth_image, image_path, depth=10):
     camera_intrinsic = o3d.camera.PinholeCameraIntrinsic()
     camera_intrinsic.set_intrinsics(w, h, 500, 500, w / 2, h / 2)
 
-
     pcd = o3d.geometry.PointCloud.create_from_rgbd_image(rgbd_image, camera_intrinsic)
     pcd.estimate_normals(
         search_param=o3d.geometry.KDTreeSearchParamHybrid(radius=0.01, max_nn=30))
     pcd.orient_normals_towards_camera_location(camera_location=np.array([0., 0., 1000.]))
 
-  
     with o3d.utility.VerbosityContextManager(o3d.utility.VerbosityLevel.Debug):
         mesh_raw, _ = o3d.geometry.TriangleMesh.create_from_point_cloud_poisson(
             pcd, depth=depth, width=0, scale=1.1, linear_fit=True)
@@ -74,7 +72,6 @@ def create_3d_obj(rgb_image, depth_image, image_path, depth=10):
     voxel_size = max(mesh_raw.get_max_bound() - mesh_raw.get_min_bound()) / 256
     mesh = mesh_raw.simplify_vertex_clustering(voxel_size=voxel_size)
 
-   
     bbox = pcd.get_axis_aligned_bounding_box()
     mesh_crop = mesh.crop(bbox)
     
@@ -86,22 +83,29 @@ def create_3d_obj(rgb_image, depth_image, image_path, depth=10):
 title = "Zero-shot Depth Estimation with DPT + 3D Point Cloud"
 description = "DPT model predicts depth from an image, followed by 3D Point Cloud reconstruction."
 
-iface = gr.Interface(
-    fn=process_image,
-    inputs=[gr.Image(type="filepath", label="Input Image")],
-    outputs=[
-        gr.Image(label="Predicted Depth", type="pil"),
-        gr.Model3D(label="3D Mesh Reconstruction", clear_color=[1.0, 1.0, 1.0, 1.0]),
-        gr.File(label="3D gLTF")
-    ],
-    title=title,
-    description=description,
-    allow_flagging="never"
-)
-
-if __name__ == "__main__":
-    iface.launch()
-
+with gr.Blocks() as iface:
+    gr.Markdown("# Zero-shot Depth Estimation with DPT + 3D Point Cloud")
+    with gr.Row():
+        image_input = gr.Image(type="filepath", label="Input Image")
+        depth_output = gr.Image(label="Predicted Depth", type="pil")
+    gltf_output = gr.File(label="Download 3D gLTF")
+
+    # Embed an iframe for previewing the .gltf
+    with gr.Row():
+        gr.HTML('<iframe id="gltf-viewer" width="100%" height="400px"></iframe>')
+
+    def update_gltf_viewer(image_path):
+        gltf_path = process_image(image_path)[1]
+        iframe_html = f'''
+        <script>
+            document.getElementById('gltf-viewer').src = 'https://gltf-viewer.donmccurdy.com/?url=file://{gltf_path}';
+        </script>
+        '''
+        return process_image(image_path)[0], gltf_path, iframe_html
+
+    image_input.change(update_gltf_viewer, inputs=[image_input], outputs=[depth_output, gltf_output, gr.HTML()])
+
+iface.launch()