Results and Weights link 404 #1

Activity

1. DI-LEE commented on May 20, 2026

   

   DI-LEE

   on May 20, 2026

   Collaborator

   More actions

   Thank you for your interest in our paper!
   I've updated the download link.
   Hope this helps!

   👍React with 👍1Reacted by akeeei

   😄React with 😄1Reacted by akeeei
2. akeeei commented on May 20, 2026

   

   akeeei

   on May 20, 2026

   Author

   More actions

   Thank you for your interest in our paper! I've updated the download link. Hope this helps!

   Hi, thank you for sharing this interesting work. I tried to reproduce the results, but the outputs I obtained seem to differ from those shown on the project page. I did not modify any of the default parameter settings. Could you please advise what might be causing this discrepancy? Any suggestions would be greatly appreciated.

   video_rgb.mp4
3. DI-LEE commented on May 21, 2026

   

   DI-LEE

   on May 21, 2026

   Last edited by DI-LEE

   Collaborator

   More actions

   First of all, the artifacts appearing right in front of the camera can be resolved by modifying the rendering trajectory (ex) zoom in.)

   However, the “red shirt” results are quite different from our results.
   I have attached the corresponding cfg file we used!

   Namespace(sh_degree=3, source_path='../../data/DyNeRF/sear_steak', model_path='../output/1107_ours_full/sear_steak_red_shirt', images='images', resolution=-1, mask_target='person', white_background=True, data_device='cuda', eval=True, render_process=False, add_points=False, extension='.png', llffhold=8, checkpoint_path='../../data/DyNeRF/sear_steak/output', edit_fps=1, layer_range='0,29', pivot_axis='time', batch_size=4, overlap_num=1, square_window=True, dataloader=True, zerostamp_init=False, custom_sampler=None, iterations=14000, edit_iterations=10000, editing_warmup_iter=100, DU_interval=50, local_vis_log_interval=100, coarse_iterations=3000, position_lr_init=0.00016, position_lr_final=1.6e-06, position_lr_delay_mult=0.01, position_lr_max_steps=20000, deformation_lr_init=0.00016, deformation_lr_final=1.6e-05, deformation_lr_delay_mult=0.01, grid_lr_init=0.0016, grid_lr_final=0.00016, feature_lr=0.0025, opacity_lr=0.05, scaling_lr=0.005, rotation_lr=0.001, percent_dense=0.01, lambda_dssim=0, lambda_lpips=0, weight_constraint_init=1, weight_constraint_after=0.2, weight_decay_iteration=5000, opacity_reset_interval=60000, densification_interval=100, densify_from_iter=500, densify_until_iter=10000, densify_grad_threshold_coarse=0.0002, densify_grad_threshold_fine_init=0.0002, densify_grad_threshold_after=0.0002, pruning_from_iter=500, pruning_interval=100, opacity_threshold_coarse=0.005, opacity_threshold_fine_init=0.005, opacity_threshold_fine_after=0.005, add_point=False, convert_SHs_python=False, compute_cov3D_python=False, debug=False, net_width=128, timebase_pe=4, defor_depth=0, posebase_pe=10, scale_rotation_pe=2, opacity_pe=2, timenet_width=64, timenet_output=32, bounds=1.6, plane_tv_weight=0.0002, time_smoothness_weight=0.001, l1_time_planes=0.0001, kplanes_config={'grid_dimensions': 2, 'input_coordinate_dim': 4, 'output_coordinate_dim': 16, 'resolution': [64, 64, 64, 150]}, multires=[1, 2], no_dx=False, no_grid=False, no_ds=False, no_dr=False, no_do=False, no_dshs=False, empty_voxel=False, grid_pe=0, static_mlp=False, apply_rotation=False, seed=1004, ip='127.0.0.1', port=6009, debug_from=-1, detect_anomaly=False, test_iterations=[3000, 7000, 14000], save_iterations=[10000, 14000, 20000, 24000, 29000, 31000, 34000, 40000, 50000, 60000, 30000], quiet=False, checkpoint_iterations=[], start_checkpoint='14000', expname='sear_steak_red_shirt', configs='../src/4DGaussians/arguments/dynerf/sear_steak.py', editing_flag=True, prompt='Change his clothes into a red shirt', original_prompt='A photo of a person wearing a dark apron cooking a steak with tongs at a wooden kitchen counter', target_prompt='A photo of a person wearing a red shirt cooking a steak with tongs at a wooden kitchen counter', idx=62, write_local_log=True, densification_iterations=7000, editing_iterations=20000, vanilla_editing=False, rendering_flag=True)
   
   
4. akeeei commented on May 22, 2026

   

   akeeei

   on May 22, 2026

   Author

   More actions

   Hi, I checked the cfg file from my run (shown below). However, the current codebase does not seem to include definitions for parameters such as square_window, mask_target, pivot_axis, etc. Could you please verify if the uploaded code version is correct?

   Namespace(sh_degree=3, source_path='../../data/DyNeRF/sear_steak', model_path='../output/editor_/sear_steak_red_shirt', images='images', resolution=-1, white_background=True, data_device='cuda', eval=True, render_process=False, add_points=False, extension='.png', llffhold=8, checkpoint_path='/opt/data/private/3d/4DGaussians/output/dynerf/sear_steak', edit_fps=1, layer_range='0,29', dataloader=True, zerostamp_init=False, custom_sampler=None, iterations=14000, edit_iterations=10000, editing_warmup_iter=100, DU_interval=50, local_vis_log_interval=100, coarse_iterations=3000, position_lr_init=0.00016, position_lr_final=1.6e-06, position_lr_delay_mult=0.01, position_lr_max_steps=20000, deformation_lr_init=0.00016, deformation_lr_final=1.6e-05, deformation_lr_delay_mult=0.01, grid_lr_init=0.0016, grid_lr_final=0.00016, feature_lr=0.0025, opacity_lr=0.05, scaling_lr=0.005, rotation_lr=0.001, percent_dense=0.01, lambda_dssim=0, lambda_lpips=0, weight_constraint_init=1, weight_constraint_after=0.2, weight_decay_iteration=5000, opacity_reset_interval=60000, densification_interval=100, densify_from_iter=500, densify_until_iter=10000, densify_grad_threshold_coarse=0.0002, densify_grad_threshold_fine_init=0.0002, densify_grad_threshold_after=0.0002, pruning_from_iter=500, pruning_interval=100, opacity_threshold_coarse=0.005, opacity_threshold_fine_init=0.005, opacity_threshold_fine_after=0.005, add_point=False, convert_SHs_python=False, compute_cov3D_python=False, debug=False, net_width=128, timebase_pe=4, defor_depth=0, posebase_pe=10, scale_rotation_pe=2, opacity_pe=2, timenet_width=64, timenet_output=32, bounds=1.6, plane_tv_weight=0.0002, time_smoothness_weight=0.001, l1_time_planes=0.0001, kplanes_config={'grid_dimensions': 2, 'input_coordinate_dim': 4, 'output_coordinate_dim': 16, 'resolution': [64, 64, 64, 150]}, multires=[1, 2], no_dx=False, no_grid=False, no_ds=False, no_dr=False, no_do=False, no_dshs=False, empty_voxel=False, grid_pe=0, static_mlp=False, apply_rotation=False, seed=1004, ip='127.0.0.1', port=6009, debug_from=-1, detect_anomaly=False, test_iterations=[3000, 7000, 14000], save_iterations=[10000, 14000, 15000, 16000, 20000, 24000, 29000, 30000, 34000, 40000, 50000, 60000, 30000], quiet=False, checkpoint_iterations=[], start_checkpoint='14000', expname='sear_steak_red_shirt', configs='../src/4DGaussians/arguments/dynerf/sear_steak.py', editing_flag=True, prompt='Change his shirt to a red shirt', idx=62, write_local_log=True, densification_iterations=7000, editing_iterations=20000, vanilla_editing=False, rendering_flag=True)