SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2405.17705 · NeurIPS · 2024

DC-Gaussian: Improving 3D Gaussian Splatting for Reflective Dash Cam Videos

Chenyang Lei, Wei Yin, Xiaoxiao Long, Kai Cheng, Linhan Wang, Shuo Lei, Shengkun Wang, Chang-Tien Lu

arXiv · PDF · Open in the Atlas

Code that ran

8 of 10 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 6 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2405.17705.json · badge: /api/ran/2405.17705.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
linhanwang/DC-Gaussian canonical (classifier label)
classifier basis: census
8 ran · 2 no recorded run
FunctionStatusWhere it lives
margin_l1_loss Ran
ran; our contract draft was wrong, not the code
linhanwang/DC-Gaussian/utils/loss_utils.py
code served (permissive licence) · get_code("7f9bf918a2148a33")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create small tensors for network_output and gt
# Shape: batch_size=2, features=4
network_output = torch.randn(2, 4)
gt = torch.randn(2, 4)

# Define margin as a small positive value
margin = 0.5

# Define return_mask parameter
return_mask = False

ARGS = {
    "network_output": network_output,
    "gt": gt,
    "margin": margin,
    "return_mask": return_mask,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
margin_l2_loss Ran
ran; our contract draft was wrong, not the code
linhanwang/DC-Gaussian/utils/loss_utils.py
code served (permissive licence) · get_code("88f9eb65bd144122")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create small tensors for network_output and ground truth
# Both should be the same shape for element-wise operations
network_output = torch.randn(2, 4)
gt = torch.randn(2, 4)

# Margin parameter - a scalar value
margin = 0.5

# return_mask parameter - boolean
return_mask = False

ARGS = {
    "network_output": network_output,
    "gt": gt,
    "margin": margin,
    "return_mask": return_mask,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
normalize Ran
ran; our contract draft was wrong, not the code
linhanwang/DC-Gaussian/utils/loss_utils.py
pointer only (licence: CC0-1.0); the stored copy get_code serves, from fictionarry/dngaussian, is pointer only (licence: NOASSERTION) · get_code("e3442ad36daea88b")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (CC0-1.0), but get_code serves the stored copy, first harvested from fictionarry/dngaussian, and GitHub could not classify the licence of that repository (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

normalize_activation Ran
ran; our contract draft was wrong, not the code
linhanwang/DC-Gaussian/lpipsPyTorch/modules/utils.py
code served (permissive licence) · get_code("1dab900b2adbe38e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create a small batch of activation values
# Shape: (batch_size=2, features=4)
x = torch.randn(2, 4)

ARGS = {
    "x": x,
    "eps": 1e-10
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
qvec2rotmat Ran
ran, violated the contract we drafted
linhanwang/DC-Gaussian/scene/colmap_loader.py
pointer only (licence: CC0-1.0); the stored copy get_code serves, from LeoQLi/VA-GS, is pointer only (licence: NONE) · get_code("6ce64cf0fbcd6bb1")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (CC0-1.0), but get_code serves the stored copy, first harvested from LeoQLi/VA-GS, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

readImages Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
linhanwang/DC-Gaussian/metrics.py
code served (permissive licence) · get_code("7033e507518f3b1a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
from pathlib import Path
from PIL import Image
import torch
import tempfile
import shutil

torch.manual_seed(0)

# Create temporary directories with dummy image files
temp_dir = tempfile.mkdtemp()
renders_dir = Path(temp_dir) / "renders"
gt_dir = Path(temp_dir) / "gt"

renders_dir.mkdir(exist_ok=True)
gt_dir.mkdir(exist_ok=True)

# Create small dummy images (4x3 RGB)
dummy_image_1 = Image.new('RGB', (4, 3), color=(255, 0, 0))
dummy_image_2 = Image.new('RGB', (4, 3), color=(0, 255, 0))

# Save dummy images
dummy_image_1.save(gt_dir / "image1.png")
dummy_image_2.save(gt_dir / "image2.png")
dummy_image_1.save(renders_dir / "image1_render.png")
dummy_image_2.save(renders_dir / "image2_render.png")

# Also save one with "orig" in the name to test filtering
dummy_image_1.save(gt_dir / "image_orig.png")
dummy_image_1.save(renders_dir / "image_orig_render.png")

ARGS = {
    "renders_dir": renders_dir,
    "gt_dir": gt_dir,
    "postfix": "_render"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
read_next_bytes Ran
ran, honoured the contract we drafted
linhanwang/DC-Gaussian/scene/colmap_loader.py
code served (permissive licence) · get_code("56858e04e6fdb2ff")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import struct
import io

# Create a binary file-like object with test data
# We'll pack some values and then read them back
test_data = struct.pack('<hfi', 42, 3.14, 100)
fid = io.BytesIO(test_data)

ARGS = {
    'fid': fid,
    'num_bytes': 10,  # 2 bytes (h) + 4 bytes (f) + 4 bytes (i) = 10 bytes
    'format_char_sequence': 'hfi',
    'endian_character': '<'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rotmat2qvec Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
linhanwang/DC-Gaussian/scene/colmap_loader.py
pointer only (licence: CC0-1.0); the stored copy get_code serves, from LeoQLi/VA-GS, is pointer only (licence: NONE) · get_code("659bc4e7e63ed8f9")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (CC0-1.0), but get_code serves the stored copy, first harvested from LeoQLi/VA-GS, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_network Not yet run
unverified (harvested, no recorded run)
linhanwang/DC-Gaussian/lpipsPyTorch/modules/networks.py
code served (permissive licence) · get_code("07bd0da29c4dc7bb")
How it ran

no invocation recorded.

get_state_dict Not yet run
unverified (harvested, no recorded run)
linhanwang/DC-Gaussian/lpipsPyTorch/modules/utils.py
code served (permissive licence) · get_code("b06f27c08cf5d0ca")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 7 are also present in other papers whose code we hold (7 with at least one paper dated earlier, 6 the same month, 6 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
read_next_bytes 70 other papers
40 dated earlier, 4 the same month, 26 later
1804.00650 (2018-04)
DeepMVS: Learning Multi-view Stereopsis
1902.10556 (2019-02)
Recurrent MVSNet for High-resolution Multi-view Stereo Depth Inference
1912.06378 (2019-12)
Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matching
and 67 more not named here
get_network 60 other papers
24 dated earlier, 9 the same month, 27 later
2203.04036 (2022-03)
StyleHEAT: One-Shot High-Resolution Editable Talking Face Generation via Pre-trained StyleGAN
2308.04079 (2023-08)
3D Gaussian Splatting for Real-Time Radiance Field Rendering
2309.13101 (2023-09)
Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction
and 57 more not named here
normalize_activation 53 other papers
20 dated earlier, 8 the same month, 25 later
2203.04036 (2022-03)
StyleHEAT: One-Shot High-Resolution Editable Talking Face Generation via Pre-trained StyleGAN
2308.04079 (2023-08)
3D Gaussian Splatting for Real-Time Radiance Field Rendering
2309.13101 (2023-09)
Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction
and 50 more not named here
get_state_dict 52 other papers
20 dated earlier, 8 the same month, 24 later
2203.04036 (2022-03)
StyleHEAT: One-Shot High-Resolution Editable Talking Face Generation via Pre-trained StyleGAN
2308.04079 (2023-08)
3D Gaussian Splatting for Real-Time Radiance Field Rendering
2309.13101 (2023-09)
Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction
and 49 more not named here
qvec2rotmat 39 other papers
21 dated earlier, 4 the same month, 14 later
2205.14870 (2022-05)
Compressible-composable NeRF via Rank-residual Decomposition
2207.10762 (2022-07)
MeshLoc: Mesh-Based Visual Localization
2303.03808 (2023-03)
Multiscale Tensor Decomposition and Rendering Equation Encoding for View Synthesis
and 36 more not named here
rotmat2qvec 34 other papers
15 dated earlier, 4 the same month, 15 later
2310.10642 (2023-10)
Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting
2311.13398 (2023-11)
Depth-Regularized Optimization for 3D Gaussian Splatting in Few-Shot Images
2311.17061 (2023-11)
HumanGaussian: Text-Driven 3D Human Generation with Gaussian Splatting
and 31 more not named here
normalize 1 other paper
1 dated earlier
2403.06912 (2024-03)
DNGaussian: Optimizing Sparse-View 3D Gaussian Radiance Fields with Global-Local Depth Normalization

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Figure 1: Given a sequence of video captured by a dash cam that may contain obstructions like reflections and occlusions, DC-Gaussian achieves high-fidelity novel view synthesis getting rid of the obstructions. (a) dash cam; (b) original video frame; (c) novel view rendering with obstruction removal.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2405.17705")
get_code_for_paper("2405.17705")
have("2405.17705")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2405.17705.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2405.17705.svg)](https://syntology.ai/paper/2405.17705)

Connect an agent — have() is free.