Lei Zhao, Wei Xing, Zhanjie Zhang, Jiakai Sun, Guangyuan Li, Han Jiao
We lifted 10 functions out of this paper's own repositories and ran 5 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| SJoJoK/3DGStream | canonical | 5 of 10 |
| Function | Status | Where it lives |
|---|---|---|
| gaussian | Ran | SJoJoK/3DGStream/utils/loss_utils.py code served (permissive licence) · get_code("c56b7ef16f309a45") |
| l1_loss | Ran | SJoJoK/3DGStream/utils/loss_utils.py code served (permissive licence) · get_code("ac0e42d6fbcfbbe6") |
| l2_loss | Ran | SJoJoK/3DGStream/utils/loss_utils.py code served (permissive licence) · get_code("8c3b0f873ba11813") |
| normalize_activation | Ran | SJoJoK/3DGStream/lpipsPyTorch/modules/utils.py code served (permissive licence) · get_code("1dab900b2adbe38e") |
| read_next_bytes | Ran | SJoJoK/3DGStream/scene/colmap_loader.py code served (permissive licence) · get_code("56858e04e6fdb2ff") |
| get_network | Not yet run | SJoJoK/3DGStream/lpipsPyTorch/modules/networks.py code served (permissive licence) · get_code("07bd0da29c4dc7bb") |
| get_state_dict | Not yet run | SJoJoK/3DGStream/lpipsPyTorch/modules/utils.py code served (permissive licence) · get_code("b06f27c08cf5d0ca") |
| prepare_output_and_logger | Not yet run | SJoJoK/3DGStream/train_frames.py code served (permissive licence) · get_code("3500802cd2c45a66") |
| qvec2rotmat | Not yet run | SJoJoK/3DGStream/scene/colmap_loader.py code served (permissive licence) · get_code("6ce64cf0fbcd6bb1") |
| rotmat2qvec | Not yet run | SJoJoK/3DGStream/scene/colmap_loader.py code served (permissive licence) · get_code("659bc4e7e63ed8f9") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Constructing photo-realistic Free-Viewpoint Videos (FVVs) of dynamic scenes from multi-view videos remains a challenging endeavor. Despite the remarkable advancements achieved by current neural rendering techniques, these methods generally require complete video sequences for offline training and are not capable of real-time rendering. To address these constraints, we introduce 3DGStream, a method designed for efficient FVV streaming of real-world dynamic scenes. Our method achieves fast on-the-fly perframe reconstruction within 12 seconds and real-time rendering at 200 FPS. Specifically, we utilize 3D Gaussians (3DGs) to represent the scene. Instead of the naïve approach of directly optimizing 3DGs per-frame, we employ a compact Neural Transformation Cache (NTC) to model the translations and rotations of 3DGs, markedly reducing the training time and storage required for each FVV frame. Furthermore, we propose an adaptive 3DG addition strategy to handle emerging objects in dynamic scenes. Experiments demonstrate that 3DGStream achieves competitive performance in terms of rendering speed, image quality, training time, and model storage when compared with stateof-the-art methods.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2403.01444")
get_code_for_paper("2403.01444")
have("2403.01444")
Connect an agent — have() is free.