update

app.py

@@ -10,13 +10,20 @@ from configs.infer_config import get_parser
 from huggingface_hub import hf_hub_download
 
 traj_examples = [
-    ['0 40', '0 0', '0 0'],
-    ['0 -35', '0 0', '0 -0.1'],
-    ['0 -3 -15 -20 -17 -5 0', '0 -2 -5 -10 -8 -5 0 2 5 3 0', '0 0'],
-    ['0 3 10 20 17 10 0', '0 -2 -8 -6 0 2 5 3 0', '0 -0.02 -0.09 -0.16 -0.09 0'],
-    ['0 30', '0 -1 -5 -4 0 1 5 4 0', '0 -0.2'],
+    ['0 -35; 0 0; 0 -0.1'],
+    ['0 -3 -15 -20 -17 -5 0; 0 -2 -5 -10 -8 -5 0 2 5 3 0; 0 0'],
+    ['0 3 10 20 17 10 0; 0 -2 -8 -6 0 2 5 3 0; 0 -0.02 -0.09 -0.16 -0.09 0'],
+    ['0 30; 0 -1 -5 -4 0 1 5 4 0; 0 -0.2'],
 ]
 
+# img_examples = [
+#     ['test/images/boy.png'],
+#     ['test/images/car.jpeg'],
+#     ['test/images/fruit.jpg'],
+#     ['test/images/room.png'],
+#     ['test/images/castle.png'],
+# ]
+
 img_examples = [
     ['test/images/boy.png',0,1],
     ['test/images/car.jpeg',5,1],
@@ -62,11 +69,31 @@ print(f'>>> System info: {version_str}')
 from viewcrafter import ViewCrafter
 
 
+CAMERA_MOTION_MODE = ["Basic Camera Trajectory", "Custom Camera Trajectory"]
+
+
+def show_traj(mode):
+    if mode == 'Left':
+        return gr.update(value='0 -35; 0 0; 0 0',visible=True),gr.update(visible=False)
+    elif mode == 'Right':
+        return gr.update(value='0 35; 0 0; 0 0',visible=True),gr.update(visible=False)
+    elif mode == 'Up':
+        return gr.update(value='0 0; 0 -30; 0 0',visible=True),gr.update(visible=False)
+    elif mode == 'Down':
+        return gr.update(value='0 0; 0 20; 0 0',visible=True), gr.update(visible=False)
+    elif mode == 'Zoom in':
+        return gr.update(value='0 0; 0 0; 0 -0.4',visible=True), gr.update(visible=False)
+    elif mode == 'Zoom out':
+        return gr.update(value='0 0; 0 0; 0 0.4',visible=True), gr.update(visible=False)
+    elif mode == 'Customize':
+        return gr.update(value='0 0; 0 0; 0 0',visible=True), gr.update(visible=True)
+    elif mode == 'Reset':
+        return gr.update(value='0 0; 0 0; 0 0',visible=False), gr.update(visible=False), gr.update(visible=False)
+
 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_traj = spaces.GPU(image2video.run_traj, duration=50) # fixme
-    image2video.run_gen = spaces.GPU(image2video.run_gen, duration=260) # fixme
+    image2video.run_both = spaces.GPU(image2video.run_gen, duration=300) # fixme
     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'>\
@@ -84,71 +111,109 @@ def viewcrafter_demo(opts):
                      <a style='font-size:18px;color: #000000' href='https://www.youtube.com/watch?v=WGIEmu9eXmU'> [Video] </a> </div>")
 
 
-        with gr.Column():
-            # step 0: tutorial
-            gr.Markdown("## Step 0: Read tutorial", show_label=False)
-            gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>Please refer to the tutorial <a href='https://github.com/Drexubery/ViewCrafter/blob/main/docs/gradio_tutorial.md' target='_blank'>here</a> for best practice, which includes the cameara system defination and the renderer parameters.</div>")
-
-            # step 2: input an image
-            gr.Markdown("---\n## Step 1: Input an Image, selet an elevation angle and a center_scale factor", show_label=False, visible=True)
-            gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>1. Estimate an elevation angle  that represents the angle at which the image was taken; a value bigger than 0 indicates a top-down view, and it doesn't need to be precise. <br>2. The origin of the world coordinate system is by default defined at the point cloud corresponding to the center pixel of the input image. You can adjust the position of the origin by modifying center_scale; a value smaller than 1 brings the origin closer to you.</div>")
-            with gr.Row(equal_height=True):
-                with gr.Column(scale=2):
-                    with gr.Row():
-                        i2v_input_image = gr.Image(label="Input Image",elem_id="input_img")
-                    with gr.Row():
-                        i2v_elevation = gr.Slider(minimum=-45, maximum=45, step=1, elem_id="elevation", label="elevation", value=5)
-                        i2v_center_scale = gr.Slider(minimum=0.1, maximum=2, step=0.1, elem_id="i2v_center_scale", label="center_scale", value=1)
-            gr.Examples(examples=img_examples,
-                        inputs=[i2v_input_image,i2v_elevation,i2v_center_scale],
-                        examples_per_page=6
-            )
-            # step 2 - camera trajectory generation
-            gr.Markdown("---\n## Step 2: Input camera trajectory", show_label=False, visible=True)
-            gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> Input a d_phi sequence, a d_theta sequence, and a d_r sequence to generate a camera trajectory. In the sequences, a positive d_phi moves the camera to the right, a negative d_theta moves the camera up, and a negative d_r moves the camera forward. Ensure that each sequence starts with 0 and contains at least two elements (a start and an end). If you upload a new image, remember to conduct this step again. </div>")
-            with gr.Row():
-                with gr.Column():
-                    # camera_mode = gr.Radio(choices=CAMERA_MOTION_MODE, value=CAMERA_MOTION_MODE[0], label="Camera Motion Control Mode", interactive=True, visible=False)
-                    i2v_d_phi = gr.Text(label='d_phi sequence')
-                    i2v_d_theta = gr.Text(label='d_theta sequence')
-                    i2v_d_r = gr.Text(label='d_r sequence')
-                    i2v_start_btn = gr.Button("Generate trajectory")
-                    # camera_info = gr.Button(value="Proceed", visible=False)      
-                with gr.Column():
-                    i2v_traj_video = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True)          
-            gr.Examples(examples=traj_examples,
-                        inputs=[i2v_d_phi, i2v_d_theta, i2v_d_r],
-            )
+        with gr.Row():
+            with gr.Column():
+                # # step 1: input an image
+                # gr.Markdown("---\n## Step 1: Input an Image, selet an elevation angle and a center_scale factor", show_label=False, visible=True)
+                # gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>1. Estimate an elevation angle  that represents the angle at which the image was taken; a value bigger than 0 indicates a top-down view, and it doesn't need to be precise. <br>2. The origin of the world coordinate system is by default defined at the point cloud corresponding to the center pixel of the input image. You can adjust the position of the origin by modifying center_scale; a value smaller than 1 brings the origin closer to you.</div>")
+                with gr.Row():
+                    with gr.Column():
+                        with gr.Row():
+                            i2v_input_image = gr.Image(label="Input Image",elem_id="input_img")
+                        with gr.Row():
+                            i2v_elevation = gr.Slider(minimum=-45, maximum=45, step=1, elem_id="elevation", label="elevation", value=5)
+                            i2v_center_scale = gr.Slider(minimum=0.1, maximum=2, step=0.1, elem_id="i2v_center_scale", label="center_scale", value=1)
+                    with  gr.Column():
+                        with gr.Row():
+                            left = gr.Button(value = "Left")
+                            right = gr.Button(value = "Right")
+                        with gr.Row():
+                            up = gr.Button(value = "Up")
+                            down = gr.Button(value = "Down")                               
+                        with gr.Row():
+                            zin = gr.Button(value = "Zoom in")
+                            zout = gr.Button(value = "Zoom out")
+                        with gr.Row():
+                            custom = gr.Button(value = "Customize")
+                            reset = gr.Button(value = "Reset")
 
-            # step 3 - Generate video
-            gr.Markdown("---\n## Step 3: Generate video", show_label=False, visible=True)
-            gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> You can reduce the sampling steps for faster inference; try different random seed if the result is not satisfying. </div>")
-            with gr.Row():
-                with gr.Column():
-                    i2v_steps = gr.Slider(minimum=1, maximum=50, step=1, elem_id="i2v_steps", label="Sampling steps", value=50)
-                    i2v_seed = gr.Slider(label='Random seed', minimum=0, maximum=max_seed, step=1, value=0)
-                    i2v_end_btn = gr.Button("Generate video")
-                    # with gr.Tab(label='Result'):
                 with gr.Column():
-                    i2v_output_video = gr.Video(label="Generated Video",elem_id="output_vid",autoplay=True,show_share_button=True)
-
+                     with gr.Row():
+                        with gr.Column():
+                            i2v_pose = gr.Text(value = '0 0; 0 0; 0 0', label='poses(d_phi sequence; d_theta sequence; d_r sequence)',visible=False)
+                            with gr.Column(visible=False) as i2v_egs:
+                                gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>Please refer to the <a href='https://github.com/Drexubery/ViewCrafter/blob/main/docs/gradio_tutorial.md' target='_blank'>tutorial</a> for customizing camera trajectory.</div>")
+                                gr.Examples(examples=traj_examples,
+                                        inputs=[i2v_pose],
+                                    )     
 
 
-        i2v_start_btn.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale, i2v_d_phi, i2v_d_theta, i2v_d_r],
-                        outputs=[i2v_traj_video],
-                        fn = image2video.run_traj
-        )
-        
-        i2v_end_btn.click(inputs=[i2v_steps, i2v_seed],
-                        outputs=[i2v_output_video],
-                        fn = image2video.run_gen
+            # step 3 - Generate video
+            with gr.Column():
+                # gr.Markdown("---\n## Step 3: Generate video", show_label=False, visible=True)
+                # gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> You can reduce the sampling steps for faster inference; try different random seed if the result is not satisfying. </div>")
+                with gr.Row():
+                    with gr.Column():
+                        i2v_output_video = gr.Video(label="Generated Video",elem_id="output_vid",autoplay=True,show_share_button=True)
+                    with gr.Column():
+                        with gr.Row():
+                            i2v_steps = gr.Slider(minimum=1, maximum=50, step=1, elem_id="i2v_steps", label="Sampling steps", value=50)
+                            i2v_seed = gr.Slider(label='Random seed', minimum=0, maximum=max_seed, step=1, value=0)
+                        i2v_end_btn = gr.Button("Generate video")                    
+                    with gr.Column():    
+                        i2v_traj_video = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True)
+
+                
+        gr.Examples(examples=img_examples,
+            inputs=[i2v_input_image,i2v_elevation, i2v_center_scale,],
+            # examples_per_page=6
+        )            
+
+
+
+        i2v_end_btn.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale, i2v_pose, i2v_steps, i2v_seed],
+                        outputs=[i2v_output_video,i2v_traj_video],
+                        fn = image2video.run_both
         )
 
+        left.click(inputs=[left],
+                      outputs=[i2v_pose,i2v_egs],
+                      fn = show_traj
+                      )
+        right.click(inputs=[right],
+                      outputs=[i2v_pose,i2v_egs],
+                      fn = show_traj
+                      )
+        up.click(inputs=[up],
+                      outputs=[i2v_pose,i2v_egs],
+                      fn = show_traj
+                      )
+        down.click(inputs=[down],
+                      outputs=[i2v_pose,i2v_egs],
+                      fn = show_traj
+                      )
+        zin.click(inputs=[zin],
+                      outputs=[i2v_pose,i2v_egs],
+                      fn = show_traj
+                      )
+        zout.click(inputs=[zout],
+                      outputs=[i2v_pose,i2v_egs],
+                      fn = show_traj
+                      )
+        custom.click(inputs=[custom],
+                      outputs=[i2v_pose,i2v_egs],
+                      fn = show_traj
+                      )
+        reset.click(inputs=[reset],
+                      outputs=[i2v_pose,i2v_egs],
+                      fn = show_traj
+                      )
+
     return viewcrafter_iface
 
 
 viewcrafter_iface = viewcrafter_demo(opts)
 viewcrafter_iface.queue(max_size=10)
 viewcrafter_iface.launch() #fixme
-# viewcrafter_iface.launch(server_name='11.220.92.96', server_port=80, max_threads=10,debug=False)
+# viewcrafter_iface.launch(server_name='11.220.92.96', server_port=80, max_threads=10,debug=True)
 

app_2step.py

@@ -0,0 +1,227 @@
+import os
+import torch
+import sys
+# import spaces #fixme
+
+import random
+import gradio as gr
+import random
+from configs.infer_config import get_parser
+from huggingface_hub import hf_hub_download
+
+traj_examples = [
+    ['0 -35; 0 0; 0 -0.1'],
+    ['0 -3 -15 -20 -17 -5 0; 0 -2 -5 -10 -8 -5 0 2 5 3 0; 0 0'],
+    ['0 3 10 20 17 10 0; 0 -2 -8 -6 0 2 5 3 0; 0 -0.02 -0.09 -0.16 -0.09 0'],
+    ['0 30; 0 -1 -5 -4 0 1 5 4 0; 0 -0.2'],
+]
+
+# img_examples = [
+#     ['test/images/boy.png'],
+#     ['test/images/car.jpeg'],
+#     ['test/images/fruit.jpg'],
+#     ['test/images/room.png'],
+#     ['test/images/castle.png'],
+# ]
+
+img_examples = [
+    ['test/images/boy.png',0,1],
+    ['test/images/car.jpeg',5,1],
+    ['test/images/fruit.jpg',5,1],
+    ['test/images/room.png',10,1],
+    ['test/images/castle.png',-4,1],
+]
+
+max_seed = 2 ** 31
+
+def download_model():
+    REPO_ID = 'Drexubery/ViewCrafter_25'
+    filename_list = ['model.ckpt']
+    for filename in filename_list:
+        local_file = os.path.join('./checkpoints/', filename)
+        if not os.path.exists(local_file):
+            hf_hub_download(repo_id=REPO_ID, filename=filename, local_dir='./checkpoints/', force_download=True)
+    
+# download_model() #fixme
+parser = get_parser() # infer_config.py
+opts = parser.parse_args() # default device: 'cuda:0'
+tmp = str(random.randint(10**(5-1), 10**5 - 1))
+opts.save_dir = f'./{tmp}'
+os.makedirs(opts.save_dir,exist_ok=True)
+test_tensor = torch.Tensor([0]).cuda()
+opts.device = str(test_tensor.device)
+# opts.config = './configs/inference_pvd_1024_gradio.yaml' #fixme
+opts.config = './configs/inference_pvd_1024_local.yaml' #fixme
+
+# # install pytorch3d # fixme
+# pyt_version_str=torch.__version__.split("+")[0].replace(".", "")
+# version_str="".join([
+#     f"py3{sys.version_info.minor}_cu",
+#     torch.version.cuda.replace(".",""),
+#     f"_pyt{pyt_version_str}"
+# ])
+# print(version_str)
+# os.system(f"{sys.executable} -m pip install --no-index --no-cache-dir pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/{version_str}/download.html")
+# os.system("mkdir -p checkpoints/ && wget https://download.europe.naverlabs.com/ComputerVision/DUSt3R/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth -P checkpoints/")
+# print(f'>>> System info: {version_str}')
+
+
+from viewcrafter import ViewCrafter
+
+
+CAMERA_MOTION_MODE = ["Basic Camera Trajectory", "Custom Camera Trajectory"]
+
+
+def show_traj(mode):
+    if mode == 'Left':
+        return gr.update(value='0 -35; 0 0; 0 0',visible=True), gr.update(visible=True), gr.update(visible=True),gr.update(visible=False)
+    elif mode == 'Right':
+        return gr.update(value='0 35; 0 0; 0 0',visible=True), gr.update(visible=True), gr.update(visible=True),gr.update(visible=False)
+    elif mode == 'Up':
+        return gr.update(value='0 0; 0 -30; 0 0',visible=True), gr.update(visible=True), gr.update(visible=True),gr.update(visible=False)
+    elif mode == 'Down':
+        return gr.update(value='0 0; 0 20; 0 0',visible=True), gr.update(visible=True), gr.update(visible=True),gr.update(visible=False)
+    elif mode == 'Zoom in':
+        return gr.update(value='0 0; 0 0; 0 -0.4',visible=True), gr.update(visible=True), gr.update(visible=True),gr.update(visible=False)
+    elif mode == 'Zoom out':
+        return gr.update(value='0 0; 0 0; 0 0.4',visible=True), gr.update(visible=True), gr.update(visible=True),gr.update(visible=False)
+    elif mode == 'Customize':
+        return gr.update(value='0 0; 0 0; 0 0',visible=True), gr.update(visible=True), gr.update(visible=True),gr.update(visible=True)
+    elif mode == 'Reset':
+        return gr.update(value='0 0; 0 0; 0 0',visible=False), gr.update(visible=False), gr.update(visible=False),gr.update(visible=False)
+
+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_traj_basic = spaces.GPU(image2video.run_traj_basic, duration=50) # fixme
+    # image2video.run_traj = spaces.GPU(image2video.run_traj, duration=50) # fixme
+    # image2video.run_gen = spaces.GPU(image2video.run_gen, duration=260) # fixme
+    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'>\
+                        <a href='https://scholar.google.com/citations?user=UOE8-qsAAAAJ&hl=zh-CN'>Wangbo Yu</a>, \
+                        <a href='https://doubiiu.github.io/'>Jinbo Xing</a>, <a href=''>Li Yuan</a>, \
+                        <a href='https://wbhu.github.io/'>Wenbo Hu</a>, <a href='https://xiaoyu258.github.io/'>Xiaoyu Li</a>,\
+                        <a href=''>Zhipeng Huang</a>, <a href='https://scholar.google.com/citations?user=qgdesEcAAAAJ&hl=en/'>Xiangjun Gao</a>,\
+                        <a href='https://www.cse.cuhk.edu.hk/~ttwong/myself.html/'>Tien-Tsin Wong</a>,\
+                        <a href='https://scholar.google.com/citations?hl=en&user=4oXBp9UAAAAJ&view_op=list_works&sortby=pubdate/'>Ying Shan</a>\
+                        <a href=''>Yonghong Tian</a>\
+                    </h2> \
+                     <a style='font-size:18px;color: #000000' href='https://arxiv.org/abs/2409.02048'> [ArXiv] </a>\
+                     <a style='font-size:18px;color: #000000' href='https://drexubery.github.io/ViewCrafter/'> [Project Page] </a>\
+                     <a style='font-size:18px;color: #FF5DB0' href='https://github.com/Drexubery/ViewCrafter'> [Github] </a>\
+                     <a style='font-size:18px;color: #000000' href='https://www.youtube.com/watch?v=WGIEmu9eXmU'> [Video] </a> </div>")
+
+
+        with gr.Row():
+            with gr.Column():
+                # # step 1: input an image
+                # gr.Markdown("---\n## Step 1: Input an Image, selet an elevation angle and a center_scale factor", show_label=False, visible=True)
+                # gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>1. Estimate an elevation angle  that represents the angle at which the image was taken; a value bigger than 0 indicates a top-down view, and it doesn't need to be precise. <br>2. The origin of the world coordinate system is by default defined at the point cloud corresponding to the center pixel of the input image. You can adjust the position of the origin by modifying center_scale; a value smaller than 1 brings the origin closer to you.</div>")
+                with gr.Row():
+                    with gr.Column():
+                        with gr.Row():
+                            i2v_input_image = gr.Image(label="Input Image",elem_id="input_img")
+                        with gr.Row():
+                            i2v_elevation = gr.Slider(minimum=-45, maximum=45, step=1, elem_id="elevation", label="elevation", value=5)
+                            i2v_center_scale = gr.Slider(minimum=0.1, maximum=2, step=0.1, elem_id="i2v_center_scale", label="center_scale", value=1)
+                    with  gr.Column():
+                        with gr.Row():
+                            left = gr.Button(value = "Left")
+                            right = gr.Button(value = "Right")
+                        with gr.Row():
+                            up = gr.Button(value = "Up")
+                            down = gr.Button(value = "Down")                               
+                        with gr.Row():
+                            zin = gr.Button(value = "Zoom in")
+                            zout = gr.Button(value = "Zoom out")
+                        with gr.Row():
+                            custom = gr.Button(value = "Customize")
+                            reset = gr.Button(value = "Reset")
+
+                with gr.Column():
+                     with gr.Row():
+                        with gr.Column():
+                            i2v_pose = gr.Text(value = '0 0; 0 0; 0 0', label='poses(d_phi;d_theta;d_r)',visible=False)
+                            with gr.Column(visible=False) as i2v_egs:
+                                gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>Please refer to the <a href='https://github.com/Drexubery/ViewCrafter/blob/main/docs/gradio_tutorial.md' target='_blank'>tutorial</a> for customizing camera trajectory.</div>")
+                                gr.Examples(examples=traj_examples,
+                                        inputs=[i2v_pose],
+                                    )     
+                        with gr.Column():
+                            i2v_traj_btn = gr.Button("2.Generate camera trajectory",visible=False)
+                            i2v_traj_video = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True,visible=False)
+
+            # step 3 - Generate video
+            with gr.Column():
+                # gr.Markdown("---\n## Step 3: Generate video", show_label=False, visible=True)
+                # gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> You can reduce the sampling steps for faster inference; try different random seed if the result is not satisfying. </div>")
+                with gr.Row():
+                    with gr.Column():
+                        i2v_output_video = gr.Video(label="Generated Video",elem_id="output_vid",autoplay=True,show_share_button=True)
+                    with gr.Column():
+                        with gr.Row():
+                            i2v_steps = gr.Slider(minimum=1, maximum=50, step=1, elem_id="i2v_steps", label="Sampling steps", value=50)
+                            i2v_seed = gr.Slider(label='Random seed', minimum=0, maximum=max_seed, step=1, value=0)
+                        i2v_end_btn = gr.Button("3.Generate video")
+                        # with gr.Tab(label='Result'):
+                
+        gr.Examples(examples=img_examples,
+            inputs=[i2v_input_image,i2v_elevation, i2v_center_scale,],
+            # examples_per_page=6
+        )            
+
+
+        # generate trajectory buttn
+        i2v_traj_btn.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale, i2v_pose],
+                        outputs=[i2v_traj_video],
+                        fn = image2video.run_traj
+        )
+
+
+        i2v_end_btn.click(inputs=[i2v_steps, i2v_seed],
+                        outputs=[i2v_output_video],
+                        fn = image2video.run_gen
+        )
+
+        left.click(inputs=[left],
+                      outputs=[i2v_pose,i2v_traj_btn,i2v_traj_video,i2v_egs],
+                      fn = show_traj
+                      )
+        right.click(inputs=[right],
+                      outputs=[i2v_pose,i2v_traj_btn,i2v_traj_video,i2v_egs],
+                      fn = show_traj
+                      )
+        up.click(inputs=[up],
+                      outputs=[i2v_pose,i2v_traj_btn,i2v_traj_video,i2v_egs],
+                      fn = show_traj
+                      )
+        down.click(inputs=[down],
+                      outputs=[i2v_pose,i2v_traj_btn,i2v_traj_video,i2v_egs],
+                      fn = show_traj
+                      )
+        zin.click(inputs=[zin],
+                      outputs=[i2v_pose,i2v_traj_btn,i2v_traj_video,i2v_egs],
+                      fn = show_traj
+                      )
+        zout.click(inputs=[zout],
+                      outputs=[i2v_pose,i2v_traj_btn,i2v_traj_video,i2v_egs],
+                      fn = show_traj
+                      )
+        custom.click(inputs=[custom],
+                      outputs=[i2v_pose,i2v_traj_btn,i2v_traj_video,i2v_egs],
+                      fn = show_traj
+                      )
+        reset.click(inputs=[reset],
+                      outputs=[i2v_pose,i2v_traj_btn,i2v_traj_video,i2v_egs],
+                      fn = show_traj
+                      )
+
+    return viewcrafter_iface
+
+
+viewcrafter_iface = viewcrafter_demo(opts)
+viewcrafter_iface.queue(max_size=10)
+# viewcrafter_iface.launch() #fixme
+viewcrafter_iface.launch(server_name='11.220.92.96', server_port=80, max_threads=10,debug=True)
+

app_new1.py

@@ -0,0 +1,230 @@
+import os
+import torch
+import sys
+# import spaces #fixme
+
+import random
+import gradio as gr
+import random
+from configs.infer_config import get_parser
+from huggingface_hub import hf_hub_download
+
+traj_examples = [
+    ['0 40', '0 0', '0 0'],
+    ['0 -35', '0 0', '0 -0.1'],
+    ['0 -3 -15 -20 -17 -5 0', '0 -2 -5 -10 -8 -5 0 2 5 3 0', '0 0'],
+    ['0 3 10 20 17 10 0', '0 -2 -8 -6 0 2 5 3 0', '0 -0.02 -0.09 -0.16 -0.09 0'],
+    ['0 30', '0 -1 -5 -4 0 1 5 4 0', '0 -0.2'],
+]
+
+img_examples = [
+    ['test/images/boy.png',0,1],
+    ['test/images/car.jpeg',5,1],
+    ['test/images/fruit.jpg',5,1],
+    ['test/images/room.png',10,1],
+    ['test/images/castle.png',-4,1],
+]
+
+max_seed = 2 ** 31
+
+def download_model():
+    REPO_ID = 'Drexubery/ViewCrafter_25'
+    filename_list = ['model.ckpt']
+    for filename in filename_list:
+        local_file = os.path.join('./checkpoints/', filename)
+        if not os.path.exists(local_file):
+            hf_hub_download(repo_id=REPO_ID, filename=filename, local_dir='./checkpoints/', force_download=True)
+    
+# download_model() #fixme
+parser = get_parser() # infer_config.py
+opts = parser.parse_args() # default device: 'cuda:0'
+tmp = str(random.randint(10**(5-1), 10**5 - 1))
+opts.save_dir = f'./{tmp}'
+os.makedirs(opts.save_dir,exist_ok=True)
+test_tensor = torch.Tensor([0]).cuda()
+opts.device = str(test_tensor.device)
+# opts.config = './configs/inference_pvd_1024_gradio.yaml' #fixme
+opts.config = './configs/inference_pvd_1024_local.yaml' #fixme
+
+# # install pytorch3d # fixme
+# pyt_version_str=torch.__version__.split("+")[0].replace(".", "")
+# version_str="".join([
+#     f"py3{sys.version_info.minor}_cu",
+#     torch.version.cuda.replace(".",""),
+#     f"_pyt{pyt_version_str}"
+# ])
+# print(version_str)
+# os.system(f"{sys.executable} -m pip install --no-index --no-cache-dir pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/{version_str}/download.html")
+# os.system("mkdir -p checkpoints/ && wget https://download.europe.naverlabs.com/ComputerVision/DUSt3R/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth -P checkpoints/")
+# print(f'>>> System info: {version_str}')
+
+
+from viewcrafter import ViewCrafter
+
+
+CAMERA_MOTION_MODE = ["Basic Camera Trajectory", "Custom Camera Trajectory"]
+
+def proceed(mode):
+    if mode == "Basic Camera Trajectory":
+        return gr.update(visible=True), gr.update(visible=False)
+    else:
+        return gr.update(visible=False), gr.update(visible=True)
+
+
+
+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_traj_basic = spaces.GPU(image2video.run_traj_basic, duration=50) # fixme
+    # image2video.run_traj = spaces.GPU(image2video.run_traj, duration=50) # fixme
+    # image2video.run_gen = spaces.GPU(image2video.run_gen, duration=260) # fixme
+    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'>\
+                        <a href='https://scholar.google.com/citations?user=UOE8-qsAAAAJ&hl=zh-CN'>Wangbo Yu</a>, \
+                        <a href='https://doubiiu.github.io/'>Jinbo Xing</a>, <a href=''>Li Yuan</a>, \
+                        <a href='https://wbhu.github.io/'>Wenbo Hu</a>, <a href='https://xiaoyu258.github.io/'>Xiaoyu Li</a>,\
+                        <a href=''>Zhipeng Huang</a>, <a href='https://scholar.google.com/citations?user=qgdesEcAAAAJ&hl=en/'>Xiangjun Gao</a>,\
+                        <a href='https://www.cse.cuhk.edu.hk/~ttwong/myself.html/'>Tien-Tsin Wong</a>,\
+                        <a href='https://scholar.google.com/citations?hl=en&user=4oXBp9UAAAAJ&view_op=list_works&sortby=pubdate/'>Ying Shan</a>\
+                        <a href=''>Yonghong Tian</a>\
+                    </h2> \
+                     <a style='font-size:18px;color: #000000' href='https://arxiv.org/abs/2409.02048'> [ArXiv] </a>\
+                     <a style='font-size:18px;color: #000000' href='https://drexubery.github.io/ViewCrafter/'> [Project Page] </a>\
+                     <a style='font-size:18px;color: #FF5DB0' href='https://github.com/Drexubery/ViewCrafter'> [Github] </a>\
+                     <a style='font-size:18px;color: #000000' href='https://www.youtube.com/watch?v=WGIEmu9eXmU'> [Video] </a> </div>")
+
+
+        with gr.Column():
+            # # step 0: tutorial
+            # gr.Markdown("## Step 0: Read tutorial", show_label=False)
+            # gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>Please refer to the tutorial <a href='https://github.com/Drexubery/ViewCrafter/blob/main/docs/gradio_tutorial.md' target='_blank'>here</a> for best practice, which includes the cameara system defination and the renderer parameters.</div>")
+
+            # step 2: input an image
+            gr.Markdown("---\n## Step 1: Input an Image, selet an elevation angle and a center_scale factor", show_label=False, visible=True)
+            gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>1. Estimate an elevation angle  that represents the angle at which the image was taken; a value bigger than 0 indicates a top-down view, and it doesn't need to be precise. <br>2. The origin of the world coordinate system is by default defined at the point cloud corresponding to the center pixel of the input image. You can adjust the position of the origin by modifying center_scale; a value smaller than 1 brings the origin closer to you.</div>")
+            with gr.Row(equal_height=True):
+                with gr.Column(scale=2):
+                    with gr.Row():
+                        i2v_input_image = gr.Image(label="Input Image",elem_id="input_img")
+                    with gr.Row():
+                        i2v_elevation = gr.Slider(minimum=-45, maximum=45, step=1, elem_id="elevation", label="elevation", value=5)
+                        i2v_center_scale = gr.Slider(minimum=0.1, maximum=2, step=0.1, elem_id="i2v_center_scale", label="center_scale", value=1)
+            gr.Examples(examples=img_examples,
+                        inputs=[i2v_input_image,i2v_elevation,i2v_center_scale],
+                        examples_per_page=6
+            )
+
+            # step 2 - camera trajectory generation
+            gr.Markdown("---\n## Step 2: Input camera trajectory", show_label=False, visible=True)
+            gr.Markdown(f"\n - {CAMERA_MOTION_MODE[0]}: Select from 6 basic camera trajectory \
+                                        \n - {CAMERA_MOTION_MODE[1]}: Customize complex camera trajectory yourself \
+                                        \n - Click `Proceed` to go into next step", 
+                                        show_label=False, visible=True)
+            with gr.Row():
+                camera_mode = gr.Radio(choices=CAMERA_MOTION_MODE, value=CAMERA_MOTION_MODE[0], label="Camera trajectory mode", interactive=True, visible=True)
+                pro_btn = gr.Button("Proceed")
+            
+            with  gr.Column(visible=False) as ouput1:
+                gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> Select one cameras trajectory. </div>")
+                with gr.Row():
+                    with gr.Column():
+                        left = gr.Button(value = "Left")
+                        right = gr.Button(value = "Right")
+                        up = gr.Button(value = "Up")
+                        down = gr.Button(value = "Down")
+                        zoomin = gr.Button(value = "Zoom in")
+                        zoomout = gr.Button(value = "Zoom out")
+                   
+                    with gr.Column():
+                        i2v_traj_video1 = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True)          
+
+                    
+            with gr.Column(visible=False) as ouput2:
+                gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> Input a d_phi sequence, a d_theta sequence, and a d_r sequence to generate a camera trajectory. In the sequences, a positive d_phi moves the camera to the right, a negative d_theta moves the camera up, and a negative d_r moves the camera forward. Ensure that each sequence starts with 0 and contains at least two elements (a start and an end). If you upload a new image, remember to conduct this step again. </div>")
+                with gr.Row():
+                    with gr.Column():
+                        # camera_mode = gr.Radio(choices=CAMERA_MOTION_MODE, value=CAMERA_MOTION_MODE[0], label="Camera Motion Control Mode", interactive=True, visible=False)
+                        i2v_d_phi2 = gr.Text(label='d_phi sequence')
+                        i2v_d_theta2 = gr.Text(label='d_theta sequence')
+                        i2v_d_r2 = gr.Text(label='d_r sequence')
+                        i2v_traj_btn2 = gr.Button("Generate custom trajectory")
+                        # camera_info = gr.Button(value="Proceed", visible=False)      
+                    with gr.Column():
+                        i2v_traj_video2 = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True)          
+                with gr.Column():
+                    gr.Examples(examples=traj_examples,
+                                inputs=[i2v_d_phi2, i2v_d_theta2, i2v_d_r2],
+                    )
+
+
+            # with gr.Column():
+            #     i2v_traj_btn = gr.Button("Generate trajectory")
+            #     i2v_traj_video = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True)  
+
+            # step 3 - Generate video
+            gr.Markdown("---\n## Step 3: Generate video", show_label=False, visible=True)
+            gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> You can reduce the sampling steps for faster inference; try different random seed if the result is not satisfying. </div>")
+            with gr.Row():
+                with gr.Column():
+                    i2v_steps = gr.Slider(minimum=1, maximum=50, step=1, elem_id="i2v_steps", label="Sampling steps", value=50)
+                    i2v_seed = gr.Slider(label='Random seed', minimum=0, maximum=max_seed, step=1, value=0)
+                    i2v_end_btn = gr.Button("Generate video")
+                    # with gr.Tab(label='Result'):
+                with gr.Column():
+                    i2v_output_video = gr.Video(label="Generated Video",elem_id="output_vid",autoplay=True,show_share_button=True)
+
+
+        pro_btn.click(inputs=[camera_mode],
+                      outputs=[ouput1,ouput2],
+                      fn = proceed
+                      )
+
+
+        i2v_traj_btn2.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale, i2v_d_phi2, i2v_d_theta2, i2v_d_r2],
+                        outputs=[i2v_traj_video2],
+                        fn = image2video.run_traj
+        )
+
+
+        left.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,left],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+
+        right.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,right],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+        up.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,up],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+
+        down.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,down],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+        zoomin.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,zoomin],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+
+        zoomout.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,zoomout],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+        
+        i2v_end_btn.click(inputs=[i2v_steps, i2v_seed],
+                        outputs=[i2v_output_video],
+                        fn = image2video.run_gen
+        )
+
+    return viewcrafter_iface
+
+
+viewcrafter_iface = viewcrafter_demo(opts)
+viewcrafter_iface.queue(max_size=10)
+# viewcrafter_iface.launch() #fixme
+viewcrafter_iface.launch(server_name='11.220.92.96', server_port=80, max_threads=10,debug=True)
+

app_new2.py

@@ -0,0 +1,238 @@
+import os
+import torch
+import sys
+# import spaces #fixme
+
+import random
+import gradio as gr
+import random
+from configs.infer_config import get_parser
+from huggingface_hub import hf_hub_download
+
+traj_examples = [
+    ['0 40', '0 0', '0 0'],
+    ['0 -35', '0 0', '0 -0.1'],
+    ['0 -3 -15 -20 -17 -5 0', '0 -2 -5 -10 -8 -5 0 2 5 3 0', '0 0'],
+    ['0 3 10 20 17 10 0', '0 -2 -8 -6 0 2 5 3 0', '0 -0.02 -0.09 -0.16 -0.09 0'],
+    ['0 30', '0 -1 -5 -4 0 1 5 4 0', '0 -0.2'],
+]
+
+# img_examples = [
+#     ['test/images/boy.png'],
+#     ['test/images/car.jpeg'],
+#     ['test/images/fruit.jpg'],
+#     ['test/images/room.png'],
+#     ['test/images/castle.png'],
+# ]
+
+img_examples = [
+    ['test/images/boy.png',0,1],
+    ['test/images/car.jpeg',5,1],
+    ['test/images/fruit.jpg',5,1],
+    ['test/images/room.png',10,1],
+    ['test/images/castle.png',-4,1],
+]
+
+max_seed = 2 ** 31
+
+def download_model():
+    REPO_ID = 'Drexubery/ViewCrafter_25'
+    filename_list = ['model.ckpt']
+    for filename in filename_list:
+        local_file = os.path.join('./checkpoints/', filename)
+        if not os.path.exists(local_file):
+            hf_hub_download(repo_id=REPO_ID, filename=filename, local_dir='./checkpoints/', force_download=True)
+    
+# download_model() #fixme
+parser = get_parser() # infer_config.py
+opts = parser.parse_args() # default device: 'cuda:0'
+tmp = str(random.randint(10**(5-1), 10**5 - 1))
+opts.save_dir = f'./{tmp}'
+os.makedirs(opts.save_dir,exist_ok=True)
+test_tensor = torch.Tensor([0]).cuda()
+opts.device = str(test_tensor.device)
+# opts.config = './configs/inference_pvd_1024_gradio.yaml' #fixme
+opts.config = './configs/inference_pvd_1024_local.yaml' #fixme
+
+# # install pytorch3d # fixme
+# pyt_version_str=torch.__version__.split("+")[0].replace(".", "")
+# version_str="".join([
+#     f"py3{sys.version_info.minor}_cu",
+#     torch.version.cuda.replace(".",""),
+#     f"_pyt{pyt_version_str}"
+# ])
+# print(version_str)
+# os.system(f"{sys.executable} -m pip install --no-index --no-cache-dir pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/{version_str}/download.html")
+# os.system("mkdir -p checkpoints/ && wget https://download.europe.naverlabs.com/ComputerVision/DUSt3R/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth -P checkpoints/")
+# print(f'>>> System info: {version_str}')
+
+
+from viewcrafter import ViewCrafter
+
+
+CAMERA_MOTION_MODE = ["Basic Camera Trajectory", "Custom Camera Trajectory"]
+
+def proceed(mode):
+    if mode == "Basic Camera Trajectory":
+        return gr.update(visible=True), gr.update(visible=False)
+    else:
+        return gr.update(visible=False), gr.update(visible=True)
+
+
+
+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_traj_basic = spaces.GPU(image2video.run_traj_basic, duration=50) # fixme
+    # image2video.run_traj = spaces.GPU(image2video.run_traj, duration=50) # fixme
+    # image2video.run_gen = spaces.GPU(image2video.run_gen, duration=260) # fixme
+    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'>\
+                        <a href='https://scholar.google.com/citations?user=UOE8-qsAAAAJ&hl=zh-CN'>Wangbo Yu</a>, \
+                        <a href='https://doubiiu.github.io/'>Jinbo Xing</a>, <a href=''>Li Yuan</a>, \
+                        <a href='https://wbhu.github.io/'>Wenbo Hu</a>, <a href='https://xiaoyu258.github.io/'>Xiaoyu Li</a>,\
+                        <a href=''>Zhipeng Huang</a>, <a href='https://scholar.google.com/citations?user=qgdesEcAAAAJ&hl=en/'>Xiangjun Gao</a>,\
+                        <a href='https://www.cse.cuhk.edu.hk/~ttwong/myself.html/'>Tien-Tsin Wong</a>,\
+                        <a href='https://scholar.google.com/citations?hl=en&user=4oXBp9UAAAAJ&view_op=list_works&sortby=pubdate/'>Ying Shan</a>\
+                        <a href=''>Yonghong Tian</a>\
+                    </h2> \
+                     <a style='font-size:18px;color: #000000' href='https://arxiv.org/abs/2409.02048'> [ArXiv] </a>\
+                     <a style='font-size:18px;color: #000000' href='https://drexubery.github.io/ViewCrafter/'> [Project Page] </a>\
+                     <a style='font-size:18px;color: #FF5DB0' href='https://github.com/Drexubery/ViewCrafter'> [Github] </a>\
+                     <a style='font-size:18px;color: #000000' href='https://www.youtube.com/watch?v=WGIEmu9eXmU'> [Video] </a> </div>")
+
+
+        with gr.Row():
+            with gr.Column():
+                # # step 1: input an image
+                # gr.Markdown("---\n## Step 1: Input an Image, selet an elevation angle and a center_scale factor", show_label=False, visible=True)
+                # gr.Markdown("<div align='left' style='font-size:18px;color: #000000'>1. Estimate an elevation angle  that represents the angle at which the image was taken; a value bigger than 0 indicates a top-down view, and it doesn't need to be precise. <br>2. The origin of the world coordinate system is by default defined at the point cloud corresponding to the center pixel of the input image. You can adjust the position of the origin by modifying center_scale; a value smaller than 1 brings the origin closer to you.</div>")
+                with gr.Row():
+                    with gr.Column():
+                        with gr.Row():
+                            i2v_input_image = gr.Image(label="Input Image",elem_id="input_img")
+                        with gr.Row():
+                            i2v_elevation = gr.Slider(minimum=-45, maximum=45, step=1, elem_id="elevation", label="elevation", value=5)
+                            i2v_center_scale = gr.Slider(minimum=0.1, maximum=2, step=0.1, elem_id="i2v_center_scale", label="center_scale", value=1)
+
+            # with gr.Column():
+                # step 2 - camera trajectory generation
+                # gr.Markdown("---\n## Step 2: Input camera trajectory", show_label=False, visible=True)
+                # gr.Markdown(f"\n - {CAMERA_MOTION_MODE[0]}: Select from 6 basic camera trajectory \
+                #                             \n - {CAMERA_MOTION_MODE[1]}: Customize complex camera trajectory yourself \
+                #                             \n - Click `Proceed` to go into next step", 
+                #                             show_label=False, visible=True)
+                with gr.Column():
+                    camera_mode = gr.Radio(choices=CAMERA_MOTION_MODE, value=CAMERA_MOTION_MODE[0], label="Camera trajectory mode", interactive=True, visible=True)
+                    pro_btn = gr.Button("1.Select camera trajectory mode")
+                
+                with  gr.Column(visible=False) as ouput1:
+                    gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> 2.Click on one basic trajectory </div>")
+                    with gr.Row():
+                        left = gr.Button(value = "Left")
+                        right = gr.Button(value = "Right")
+                    with gr.Row():
+                        up = gr.Button(value = "Up")
+                        down = gr.Button(value = "Down")
+                    with gr.Row():
+                        zoomin = gr.Button(value = "Zoom in")
+                        zoomout = gr.Button(value = "Zoom out")
+                    
+                        with gr.Column():
+                            i2v_traj_video1 = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True)          
+
+                        
+                with gr.Column(visible=False) as ouput2:
+                    gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> Input a d_phi sequence, a d_theta sequence, and a d_r sequence, then click 'Generate custom trajectory' <a href='https://github.com/Drexubery/ViewCrafter/blob/main/docs/gradio_tutorial.md' target='_blank'>(Tutorial)</a> </div>")
+                    with gr.Row():
+                        with gr.Column():
+                            # camera_mode = gr.Radio(choices=CAMERA_MOTION_MODE, value=CAMERA_MOTION_MODE[0], label="Camera Motion Control Mode", interactive=True, visible=False)
+                            i2v_d_phi2 = gr.Text(label='d_phi sequence')
+                            i2v_d_theta2 = gr.Text(label='d_theta sequence')
+                            i2v_d_r2 = gr.Text(label='d_r sequence')
+                            i2v_traj_btn2 = gr.Button("2.Generate custom trajectory")
+                            # camera_info = gr.Button(value="Proceed", visible=False)      
+                        with gr.Column():
+                            i2v_traj_video2 = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True)          
+                with gr.Column():
+                    gr.Examples(examples=traj_examples,
+                                inputs=[i2v_d_phi2, i2v_d_theta2, i2v_d_r2],
+                    )
+
+                # with gr.Column():
+                #     i2v_traj_btn = gr.Button("Generate trajectory")
+                #     i2v_traj_video = gr.Video(label="Camera Trajectory",elem_id="traj_vid",autoplay=True,show_share_button=True)  
+
+            # step 3 - Generate video
+            with gr.Column():
+                # gr.Markdown("---\n## Step 3: Generate video", show_label=False, visible=True)
+                # gr.Markdown("<div align='left' style='font-size:18px;color: #000000'> You can reduce the sampling steps for faster inference; try different random seed if the result is not satisfying. </div>")
+                with gr.Row():
+                    with gr.Column():
+                        i2v_output_video = gr.Video(label="Generated Video",elem_id="output_vid",autoplay=True,show_share_button=True)
+                    with gr.Column():
+                        with gr.Row():
+                            i2v_steps = gr.Slider(minimum=1, maximum=50, step=1, elem_id="i2v_steps", label="Sampling steps", value=50)
+                            i2v_seed = gr.Slider(label='Random seed', minimum=0, maximum=max_seed, step=1, value=0)
+                        i2v_end_btn = gr.Button("3.Generate video")
+                        # with gr.Tab(label='Result'):
+                
+        gr.Examples(examples=img_examples,
+            inputs=[i2v_input_image,i2v_elevation, i2v_center_scale,],
+            # examples_per_page=6
+        )            
+
+        pro_btn.click(inputs=[camera_mode],
+                      outputs=[ouput1,ouput2],
+                      fn = proceed
+                      )
+
+        # generate trajectory buttn
+        i2v_traj_btn2.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale, i2v_d_phi2, i2v_d_theta2, i2v_d_r2],
+                        outputs=[i2v_traj_video2],
+                        fn = image2video.run_traj
+        )
+
+
+        left.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,left],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+
+        right.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,right],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+        up.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,up],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+
+        down.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,down],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+        zoomin.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,zoomin],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+
+        zoomout.click(inputs=[i2v_input_image, i2v_elevation, i2v_center_scale,zoomout],
+                        outputs=[i2v_traj_video1],
+                        fn = image2video.run_traj_basic
+        )
+        
+        i2v_end_btn.click(inputs=[i2v_steps, i2v_seed],
+                        outputs=[i2v_output_video],
+                        fn = image2video.run_gen
+        )
+
+    return viewcrafter_iface
+
+
+viewcrafter_iface = viewcrafter_demo(opts)
+viewcrafter_iface.queue(max_size=10)
+# viewcrafter_iface.launch() #fixme
+viewcrafter_iface.launch(server_name='11.220.92.96', server_port=80, max_threads=10,debug=True)
+

viewcrafter.py

@@ -160,9 +160,8 @@ class ViewCrafter:
                 render_results[-1] = self.img_ori
         # torch.Size([25, 576, 1024, 3]), [0,1]
         # save_pointcloud_with_normals([imgs[-1]], [pcd[-1]], msk=None, save_path=os.path.join(self.opts.save_dir,'pcd0.ply') , mask_pc=False, reduce_pc=False)
-        # diffusion_results = self.run_diffusion(render_results)
-        # save_video((diffusion_results + 1.0) / 2.0, os.path.join(self.opts.save_dir, 'diffusion0.mp4'))
-
+        diffusion_results = self.run_diffusion(render_results)
+        save_video((diffusion_results + 1.0) / 2.0, os.path.join(self.opts.save_dir, 'diffusion0.mp4'))
         return render_results
 
     def nvs_sparse_view(self,iter):
@@ -349,13 +348,14 @@ class ViewCrafter:
 
         return images, img_ori
     
-    def run_traj(self,i2v_input_image, i2v_elevation, i2v_center_scale, i2v_d_phi, i2v_d_theta, i2v_d_r):
+    def run_traj(self,i2v_input_image, i2v_elevation, i2v_center_scale, i2v_pose):
         self.opts.elevation = float(i2v_elevation)
         self.opts.center_scale = float(i2v_center_scale)
+        i2v_d_phi,i2v_d_theta,i2v_d_r = [i for i in i2v_pose.split(';')]
         self.gradio_traj = [float(i) for i in i2v_d_phi.split()],[float(i) for i in i2v_d_theta.split()],[float(i) for i in i2v_d_r.split()]
         transform = transforms.Compose([
-            transforms.Resize(576),
-            transforms.CenterCrop((576,1024)),
+            transforms.Resize((576,1024)),
+            # transforms.CenterCrop((576,1024)),
             ])
         torch.cuda.empty_cache()
         img_tensor = torch.from_numpy(i2v_input_image).permute(2, 0, 1).unsqueeze(0).float().to(self.device)
@@ -392,4 +392,34 @@ class ViewCrafter:
         gen_dir = os.path.join(self.opts.save_dir, "diffusion0.mp4")
         diffusion_results = self.run_diffusion(render_results)
         save_video((diffusion_results + 1.0) / 2.0, os.path.join(self.opts.save_dir, 'diffusion0.mp4'))
-        return gen_dir
+        return gen_dir
+
+    def run_both(self,i2v_input_image, i2v_elevation, i2v_center_scale, i2v_pose,i2v_steps, i2v_seed):
+        self.opts.ddim_steps = i2v_steps
+        seed_everything(i2v_seed)
+        self.opts.elevation = float(i2v_elevation)
+        self.opts.center_scale = float(i2v_center_scale)
+        i2v_d_phi,i2v_d_theta,i2v_d_r = [i for i in i2v_pose.split(';')]
+        self.gradio_traj = [float(i) for i in i2v_d_phi.split()],[float(i) for i in i2v_d_theta.split()],[float(i) for i in i2v_d_r.split()]
+        transform = transforms.Compose([
+            transforms.Resize((576,1024)),
+            # transforms.CenterCrop((576,1024)),
+            ])
+        torch.cuda.empty_cache()
+        img_tensor = torch.from_numpy(i2v_input_image).permute(2, 0, 1).unsqueeze(0).float().to(self.device)
+        img_tensor = (img_tensor / 255. - 0.5) * 2
+        image_tensor_resized = transform(img_tensor) #1,3,h,w
+        images = get_input_dict(image_tensor_resized,idx = 0,dtype = torch.float32)
+        images = [images, copy.deepcopy(images)]
+        images[1]['idx'] = 1
+        self.images = images
+        self.img_ori = (image_tensor_resized.squeeze(0).permute(1,2,0) + 1.)/2.
+
+        # self.images: torch.Size([1, 3, 288, 512]), [-1,1]
+        # self.img_ori:  torch.Size([576, 1024, 3]), [0,1]
+        # self.images, self.img_ori = self.load_initial_images(image_dir=i2v_input_image)
+        self.run_dust3r(input_images=self.images)
+        self.nvs_single_view(gradio=True)
+        traj_dir = os.path.join(self.opts.save_dir, "viz_traj.mp4")
+        gen_dir = os.path.join(self.opts.save_dir, "diffusion0.mp4")
+        return gen_dir,traj_dir,