SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.11273 · 2024

BrightDreamer: Generic 3D Gaussian Generative Framework for Fast Text-to-3D Synthesis

arXiv · PDF · Open in the Atlas

Code that ran

10 of 13 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; 2 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 7 failures of our instrument; 3 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/2403.11273.json · badge: /api/ran/2403.11273.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
lutao2021/BrightDreamer canonical (classifier label)
classifier basis: census
10 ran · 3 no recorded run
FunctionStatusWhere it lives
RGB2SH Ran
ran, honoured the contract we drafted
lutao2021/BrightDreamer/generator/BrightDreamer.py
pointer only (licence: MIT); the stored copy get_code serves, from avinashpaliwal/coherentgs, is pointer only (licence: NOASSERTION) · get_code("5a8b1e0eb5ec2789")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from avinashpaliwal/coherentgs, 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.

adjust_text_embeddings Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
lutao2021/BrightDreamer/trainer.py
code served (permissive licence) · get_code("cb3ad4e37c507940")
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. 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 torch
import math

torch.manual_seed(0)

# Create a simple options class with the required attributes
class OptConfig:
    def __init__(self):
        self.front_decay_factor = 2.0
        self.side_decay_factor = 1.5
        self.negative_w = 0.5

# Create embeddings dictionary with small tensors
# Each embedding should have shape (batch_size, embedding_dim)
embeddings = {
    'front': torch.randn(2, 4),
    'side': torch.randn(2, 4),
    'back': torch.randn(2, 4),
}

# Create azimuth values as a list
# Test with values in different ranges to exercise both branches
azimuth = [45.0, -120.0]

# Create opt config
opt = OptConfig()

ARGS = {
    'embeddings': embeddings,
    'azimuth': azimuth,
    'opt': opt,
}
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 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
biased_softplus Ran
ran; our contract draft was wrong, not the code
lutao2021/BrightDreamer/activation.py
code served (permissive licence) · get_code("fb07704f3a4c0430")
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 tensor input
x = torch.randn(2, 4)

# Create bias parameter
bias = 0.5

# Construct ARGS dict with all parameters
ARGS = {
    'x': x,
    'bias': bias
}
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
format_time Ran
ran; our contract draft was wrong, not the code
lutao2021/BrightDreamer/dnnlib/util.py
pointer only (licence: MIT); the stored copy get_code serves, from ntuaislab/DRAG, is pointer only (licence: NOASSERTION) · get_code("053fc534bc6bb989")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ntuaislab/DRAG, 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.

format_time_brief Ran
ran; our contract draft was wrong, not the code
lutao2021/BrightDreamer/dnnlib/util.py
pointer only (licence: MIT); the stored copy get_code serves, from ntuaislab/DRAG, is pointer only (licence: NOASSERTION) · get_code("77f4aa0649e7f404")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ntuaislab/DRAG, 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.

gaussian_3d_coeff Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
lutao2021/BrightDreamer/generator/gaussian_utils/gaussian_model.py
code served (permissive licence) · get_code("56a932cc407a8ecb")
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 1, dtype family float, shape (2,) 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 tiny tensors with batch size 2
# xyzs: [batch_size, 3] - 3D coordinates
xyzs = torch.randn(2, 3)

# covs: [batch_size, 6] - 6 covariance matrix elements (a, b, c, d, e, f)
# representing symmetric 3x3 matrix [[a, b, c], [b, d, e], [c, e, f]]
covs = torch.randn(2, 6)

ARGS = {
    "xyzs": xyzs,
    "covs": covs,
}
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 1, dtype family float, shape (2,) — shape (2,) 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
get_pos_neg_text_embeddings Ran
ran; our contract draft was wrong, not the code
lutao2021/BrightDreamer/trainer.py
code served (permissive licence) · get_code("0f6ee172f484a171")
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. 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 torch
import math

torch.manual_seed(0)

# Create embeddings dict with small tensors
# The function expects 'front', 'side', and 'back' keys
# Each embedding should have the same batch size and dimension
embedding_dim = 4
batch_size = 2

embeddings = {
    'front': torch.randn(batch_size, embedding_dim),
    'side': torch.randn(batch_size, embedding_dim),
    'back': torch.randn(batch_size, embedding_dim),
}

# Create a simple config object with the required attributes
class Config:
    def __init__(self):
        self.front_decay_factor = 2.0
        self.side_decay_factor = 1.5
        self.negative_w = 0.5

opt = Config()

# Test with an azimuth value in the [-90, 90) range
azimuth_val = 30

ARGS = {
    'embeddings': embeddings,
    'azimuth_val': azimuth_val,
    'opt': opt,
}
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 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
positional_encoding_2d Ran
ran, honoured the contract we drafted
lutao2021/BrightDreamer/generator/network_crossattn_generative.py
code served (permissive licence) · get_code("32ac24298457bdaf")
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 ndarray, ndim 3, dtype family float, shape (4, 3, 8) 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 numpy as np

# d_model must be divisible by 4
d_model = 8

# height and width are spatial dimensions
height = 4
width = 3

ARGS = {
    "d_model": d_model,
    "height": height,
    "width": width,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (4, 3, 8) — shape (4, 3, 8) 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
safe_normalize Ran
ran; our contract draft was wrong, not the code
lutao2021/BrightDreamer/trainer.py
code served (permissive licence) · get_code("1c9cf033d080b2e0")
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)

# The function safe_normalize takes a tensor x and normalizes it
# by dividing by the L2 norm computed along the last dimension
# Parameters:
# - x: a tensor of any shape
# - eps: a small epsilon value for numerical stability (default 1e-20)

# Create a small test tensor with shape (2, 4)
# This gives us a batch of 2 vectors, each of dimension 4
x = torch.randn(2, 4)

# eps uses the default value from the function signature
eps = 1e-20

ARGS = {
    "x": x,
    "eps": eps,
}
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
unflatten_with_named_tensor Ran
ran on a synthesized input
lutao2021/BrightDreamer/dpt.py
code served (permissive licence) · get_code("f2952bdbf9446f24")
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 3, dtype family float, shape (2, 3, 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
import torch.nn as nn

torch.manual_seed(0)

# Create a small named tensor to unflatten
# Shape: (2, 12) - batch size 2, flattened dimension of 12
input_tensor = torch.randn(2, 12)

# The dimension to unflatten (dim=1, the second dimension)
dim = 1

# The target sizes to unflatten into (3, 4) so 3*4=12
sizes = (3, 4)

ARGS = {
    "input": input_tensor,
    "dim": dim,
    "sizes": sizes,
}
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 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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
ask_yes_no Not yet run
unverified (harvested, no recorded run)
lutao2021/BrightDreamer/dnnlib/util.py
code served (permissive licence) · get_code("9d31d2c4cd16bb2d")
How it ran

no invocation recorded.

forward_flex Not yet run
unverified (harvested, no recorded run)
lutao2021/BrightDreamer/dpt.py
code served (permissive licence) · get_code("e243c3c7c7f3110a")
How it ran

no invocation recorded.

forward_vit Not yet run
unverified (harvested, no recorded run)
lutao2021/BrightDreamer/dpt.py
code served (permissive licence) · get_code("47594f898e209e22")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 11 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 4 the same month, 8 later) and 2 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
format_time 73 other papers
26 dated earlier, 4 the same month, 42 later, 1 undated
2004.00049 (2020-04)
In-Domain GAN Inversion for Real Image Editing
2004.11660 (2020-04)
Disentangled and Controllable Face Image Generation via 3D Imitative-Contrastive Learning
2108.00946 (2021-08)
StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generators
and 70 more not named here
ask_yes_no 69 other papers
26 dated earlier, 4 the same month, 38 later, 1 undated
2004.00049 (2020-04)
In-Domain GAN Inversion for Real Image Editing
2004.11660 (2020-04)
Disentangled and Controllable Face Image Generation via 3D Imitative-Contrastive Learning
2108.00946 (2021-08)
StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generators
and 66 more not named here
format_time_brief 68 other papers
22 dated earlier, 4 the same month, 41 later, 1 undated
2108.00946 (2021-08)
StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generators
2202.00273 (2022-02)
StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets
2206.07695 (2022-06)
VoxGRAF: Fast 3D-Aware Image Synthesis with Sparse Voxel Grids
and 65 more not named here
RGB2SH 12 other papers
1 dated earlier, 2 the same month, 9 later
2311.14521 (2023-11)
GaussianEditor: Swift and Controllable 3D Editing with Gaussian Splatting
2403.16095 (2024-03)
CG-SLAM: Efficient Dense RGB-D SLAM in a Consistent Uncertainty-aware 3D Gaussian Field
2403.19495 (2024-03)
CoherentGS: Sparse Novel View Synthesis with Coherent 3D Gaussians
and 9 more not named here
safe_normalize 6 other papers
5 dated earlier, 1 later
2209.14988 (2022-09)
DreamFusion: Text-to-3D using 2D Diffusion
2211.07600 (2022-11)
Latent-NeRF for Shape-Guided Generation of 3D Shapes and Textures
2308.08545 (2023-08)
TeCH: Text-guided Reconstruction of Lifelike Clothed Humans
and 3 more not named here
biased_softplus 4 other papers
3 dated earlier, 1 later
2302.10663 (2023-02)
RealFusion: 360° Reconstruction of Any Object from a Single Image
2306.17843 (2023-06)
Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors
2308.11473 (2023-08)
IT3D: Improved Text-to-3D Generation with Explicit View Synthesis
and 1 more not named here
forward_flex 4 other papers
4 dated earlier
2302.10663 (2023-02)
RealFusion: 360° Reconstruction of Any Object from a Single Image
2303.02091 (2023-03)
Delicate Textured Mesh Recovery from NeRF via Adaptive Surface Refinement
2306.17843 (2023-06)
Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors
and 1 more not named here
forward_vit 4 other papers
4 dated earlier
2302.10663 (2023-02)
RealFusion: 360° Reconstruction of Any Object from a Single Image
2303.02091 (2023-03)
Delicate Textured Mesh Recovery from NeRF via Adaptive Surface Refinement
2306.17843 (2023-06)
Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors
and 1 more not named here
unflatten_with_named_tensor 4 other papers
4 dated earlier
2302.10663 (2023-02)
RealFusion: 360° Reconstruction of Any Object from a Single Image
2303.02091 (2023-03)
Delicate Textured Mesh Recovery from NeRF via Adaptive Surface Refinement
2306.17843 (2023-06)
Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors
and 1 more not named here
gaussian_3d_coeff 1 other paper
1 later
2405.02280 (2024-05)
DreamScene4D: Dynamic Multi-Object Scene Generation from Monocular Videos
get_pos_neg_text_embeddings 1 other paper
1 later
2410.07155 (2024-10)
Trans4D: Realistic Geometry-Aware Transition for Compositional Text-to-4D Synthesis

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

Text-to-3D synthesis has recently seen intriguing advances by combining the text-to-image priors with 3D representation methods, e.g., 3D Gaussian Splatting (3D GS), via Score Distillation Sampling (SDS). However, a hurdle of existing methods is the low efficiency, per-prompt optimization for a single 3D object. Therefore, it is imperative for a paradigm shift from per-prompt optimization to feed-forward generation for any unseen text prompts, which yet remains challenging. An obstacle is how to directly generate a set of millions of 3D Gaussians to represent a 3D object. This paper presents BrightDreamer, an end-to-end feed-forward approach that can achieve generalizable and fast (77 ms) text-to-3D generation. Our key idea is to formulate the generation process as estimating the 3D deformation from an anchor shape with predefined positions. For this, we first propose a Text-guided Shape Deformation (TSD) network to predict the deformed shape and its new positions, used as the centers (one attribute) of 3D Gaussians. To estimate the other four attributes (i.e., scaling, rotation, opacity, and SH), we then design a novel Text-guided Triplane Generator (TTG) to generate a triplane representation for a 3D object. The center of each Gaussian enables us to transform the spatial feature into the four attributes. The generated 3D Gaussians can be finally rendered at 705 frames per second. Extensive experiments demonstrate the superiority of our method over existing methods. Also, BrightDreamer possesses a strong semantic understanding capability even for complex text prompts. The code is available in the project page.

For agents

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

get_harvested_code_for_paper("2403.11273")
get_code_for_paper("2403.11273")
have("2403.11273")

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

curl https://syntology.ai/api/ran/2403.11273.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.11273.svg)](https://syntology.ai/paper/2403.11273)

Connect an agent — have() is free.