Only export model

app.py

@@ -20,7 +20,6 @@ from utils.app_utils import (
 import imageio
 
 from scene.gaussian_predictor import GaussianSplatPredictor
-from gaussian_renderer import render_predicted
 
 import gradio as gr
 
@@ -30,11 +29,8 @@ from huggingface_hub import hf_hub_download
 
 def main():
 
-    os.system("sudo find /usr -name nvcc")
-
     if torch.cuda.is_available():
         device = "cuda:0"
-        print("Found cuda")
     else:
         device = "cpu"
 
@@ -94,31 +90,10 @@ def main():
             None,
             activate_output=False)
 
-        reconstruction = {k: v[0].contiguous() for k, v in reconstruction_unactivated.items()}
-        reconstruction["scaling"] = model.scaling_activation(reconstruction["scaling"])
-        reconstruction["opacity"] = model.opacity_activation(reconstruction["opacity"])
-
-        # render images in a loop
-        world_view_transforms, full_proj_transforms, camera_centers = get_target_cameras()
-        background = torch.tensor([1, 1, 1] , dtype=torch.float32, device=device)
-        loop_renders = []
-        t_to_512 = torchvision.transforms.Resize(512, interpolation=torchvision.transforms.InterpolationMode.NEAREST)
-        for r_idx in range( world_view_transforms.shape[0]):
-            image = render_predicted(reconstruction,
-                                     world_view_transforms[r_idx].to(device),
-                                     full_proj_transforms[r_idx].to(device), 
-                                     camera_centers[r_idx].to(device),
-                                     background,
-                                     model_cfg,
-                                     focals_pixels=None)["render"]
-            image = t_to_512(image)
-            loop_renders.append(torch.clamp(image * 255, 0.0, 255.0).detach().permute(1, 2, 0).cpu().numpy().astype(np.uint8))
-        loop_out_path = os.path.join(os.path.dirname(ply_out_path), "loop.mp4")
-        imageio.mimsave(loop_out_path, loop_renders, fps=25)
         # export reconstruction to ply
         export_to_obj(reconstruction_unactivated, ply_out_path)
 
-        return ply_out_path, loop_out_path
+        return ply_out_path
 
     with gr.Blocks() as demo:
         gr.Markdown(
@@ -159,13 +134,11 @@ def main():
             with gr.Column():
                 with gr.Row():
                     with gr.Tab("Reconstruction"):
-                        with gr.Column():
-                            output_video = gr.Video(value=None, width=512, label="Rendered Video", autoplay=True)
-                            output_model = gr.Model3D(
-                                height=512,
-                                label="Output Model",
-                                interactive=False
-                            )
+                        output_model = gr.Model3D(
+                            height=512,
+                            label="Output Model",
+                            interactive=False
+                        )
 
         submit.click(fn=check_input_image, inputs=[input_image]).success(
             fn=preprocess,
@@ -174,7 +147,7 @@ def main():
         ).success(
             fn=reconstruct_and_export,
             inputs=[processed_image],
-            outputs=[output_model, output_video],
+            outputs=[output_model],
         )
 
     demo.queue(max_size=1)

requirements.txt

@@ -8,5 +8,4 @@ moviepy
 markupsafe==2.0.1
 gradio==4.25.0
 rembg
-spaces
-cuda-python
+spaces