update

app.py

@@ -6,12 +6,10 @@ import spaces
 # os.system('pip install iopath')
 # os.system("pip install -v -v -v  'git+https://github.com/facebookresearch/pytorch3d.git@stable'")
 # os.system("cd pytorch3d && pip install -e . && cd ..")
-# os.system("pip install 'git+https://github.com/facebookresearch/pytorch3d.git'")
 os.system("mkdir -p checkpoints/ && wget https://download.europe.naverlabs.com/ComputerVision/DUSt3R/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth -P checkpoints/")
 
 import gradio as gr
 import random
-from viewcrafter import ViewCrafter
 from configs.infer_config import get_parser
 from huggingface_hub import hf_hub_download
 
@@ -34,11 +32,19 @@ def download_model():
             hf_hub_download(repo_id=REPO_ID, filename=filename, local_dir='./checkpoints/', force_download=True)
     
 download_model()
+parser = get_parser() # infer_config.py
+opts = parser.parse_args() # default device: 'cuda:0'
+opts.save_dir = './'
+os.makedirs(opts.save_dir,exist_ok=True)
+test_tensor = torch.Tensor([0]).cuda()
+opts.device = str(test_tensor.device)
+os.system("pip install 'git+https://github.com/facebookresearch/pytorch3d.git'")
+from viewcrafter import ViewCrafter
 
 def viewcrafter_demo(opts):
     css = """#input_img {max-width: 1024px !important} #output_vid {max-width: 1024px; max-height:576px} #random_button {max-width: 100px !important}"""
     image2video = ViewCrafter(opts, gradio = True)
-    image2video.run_gradio = spaces.GPU(image2video.run_gradio, duration=800)
+    image2video.run_gradio = spaces.GPU(image2video.run_gradio, duration=300)
     with gr.Blocks(analytics_enabled=False, css=css) as viewcrafter_iface:
         gr.Markdown("<div align='center'> <h1> ViewCrafter: Taming Video Diffusion Models for High-fidelity Novel View Synthesis </span> </h1> \
                       <h2 style='font-weight: 450; font-size: 1rem; margin: 0rem'>\
@@ -100,15 +106,8 @@ def viewcrafter_demo(opts):
     return viewcrafter_iface
 
 
-if __name__ == "__main__":
-    parser = get_parser() # infer_config.py
-    opts = parser.parse_args() # default device: 'cuda:0'
-    opts.save_dir = './'
-    os.makedirs(opts.save_dir,exist_ok=True)
-    test_tensor = torch.Tensor([0]).cuda()
-    opts.device = str(test_tensor.device)
-    viewcrafter_iface = viewcrafter_demo(opts)
-    viewcrafter_iface.queue(max_size=10)
-    viewcrafter_iface.launch()
-    # viewcrafter_iface.launch(server_name='127.0.0.1', server_port=80, max_threads=1,debug=False)
+viewcrafter_iface = viewcrafter_demo(opts)
+viewcrafter_iface.queue(max_size=10)
+viewcrafter_iface.launch()
+# viewcrafter_iface.launch(server_name='127.0.0.1', server_port=80, max_threads=1,debug=False)
 

requirements.txt

@@ -28,5 +28,4 @@ omegaconf==2.3.0
 triton
 av
 xformers
-gradio
-git+https://github.com/facebookresearch/pytorch3d.git
+gradio

viewcrafter.py

@@ -143,7 +143,8 @@ class ViewCrafter:
                     r = [float(i) for i in lines[2].split()]
             else: 
                 phi, theta, r = self.gradio_traj
-            device = torch.device("cpu")
+            # device = torch.device("cpu")
+            device = self.device
             camera_traj,num_views = generate_traj_txt(c2ws, H, W, focals, principal_points, phi, theta, r,self.opts.video_length, device,viz_traj=True, save_dir = self.opts.save_dir)
             # camera_traj,num_views = generate_traj_txt(c2ws, H, W, focals, principal_points, phi, theta, r,self.opts.video_length, self.device,viz_traj=True, save_dir = self.opts.save_dir)
         else: