SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2212.08070 · 2022

NeRF-Art: Text-Driven Neural Radiance Fields Stylization

arXiv · PDF · Open in the Atlas

Code that ran

8 of 9 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; 0 honoured the contract we drafted, 0 violated it and 7 have no contract verdict; 1 failure of our instrument; 1 no recorded run), measured 2026-11-21; 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/2212.08070.json · badge: /api/ran/2212.08070.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
cassiePython/NeRF-Art canonical (classifier label)
classifier basis not recorded on the link
8 ran · 1 no recorded run
FunctionStatusWhere it lives
CosineAnnealWarmUpSchedulerLambda Ran
ran on a synthesized input
cassiePython/NeRF-Art/models/base.py
code served (permissive licence) · get_code("dbd4db730f0b61af")
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 function. 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 numpy as np

ARGS = {
    'total_steps': 1000,
    'warmup_steps': 100,
    'min_factor': 0.1
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
get_embedder Ran
ran; our contract draft was wrong, not the code
cassiePython/NeRF-Art/models/base.py
code served (permissive licence) · get_code("8fa39b1ba521d6b5")
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, 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 torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'multires': 4,
    'input_dim': 3
}
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, 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
get_optimizer Ran
ran on a synthesized input
cassiePython/NeRF-Art/models/base.py
code served (permissive licence) · get_code("bcf3ce36658dfbb4")
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 Adam. 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 torch.nn as nn
from torch import optim
import numbers


torch.manual_seed(0)


class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.fc1 = nn.Linear(4, 3)
        self.fc2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.fc1(x)
        x = self.fc2(x)
        return x


class TrainingConfig:
    def __init__(self, lr):
        self.lr = lr


class Args:
    def __init__(self, training):
        self.training = training


# Test case 1: numeric learning rate
model1 = SimpleModel()
training_config1 = TrainingConfig(lr=0.001)
args1 = Args(training=training_config1)

# Test case 2: dictionary learning rate with default and module-specific rates
model2 = SimpleModel()
training_config2 = TrainingConfig(lr={
    'default': 0.001,
    'fc1': 0.002,
    'fc2': 0.0005
})
args2 = Args(training=training_config2)

# Use the first test case
ARGS = {
    'args': args1,
    'model': model1
}
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 Adam — a Adam output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Adam output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
load_cam Ran
ran on a synthesized input
cassiePython/NeRF-Art/dataio/BlendedMVS.py
code served (permissive licence) · get_code("df48902042eca6a8")
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 (2, 4, 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 numpy as np
import tempfile
import os

# Create a temporary camera file with the required format
# The function expects a file with specific word positions for extrinsic and intrinsic matrices
temp_dir = tempfile.mkdtemp()
filepath = os.path.join(temp_dir, "camera.txt")

# Create a camera file with 31 words (maximum case for original_blendedmvs)
# Words 0: placeholder
# Words 1-16: extrinsic matrix (4x4, row-major)
# Words 17: placeholder
# Words 18-26: intrinsic matrix (3x3, row-major)
# Words 27-30: depth range parameters (min_depth, depth_interval, max_depth, max_depth_computed)

words_list = [
    "0",  # word 0: placeholder
    # Extrinsic matrix (words 1-16): 4x4 identity-like
    "1.0", "0.0", "0.0", "0.1",
    "0.0", "1.0", "0.0", "0.2",
    "0.0", "0.0", "1.0", "0.3",
    "0.0", "0.0", "0.0", "1.0",
    # word 17: placeholder
    "0",
    # Intrinsic matrix (words 18-26): 3x3
    "320.0", "0.0", "160.0",
    "0.0", "320.0", "120.0",
    "0.0", "0.0", "1.0",
    # Depth parameters (words 27-30)
    "425.0", "2.5", "97.5", "700.0"
]

with open(filepath, "w") as f:
    f.write(" ".join(words_list))

ARGS = {
    "filepath": filepath,
    "interval_scale": 1,
    "original_blendedmvs": True
}
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 (2, 4, 4) — shape (2, 4, 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
normalize Ran
ran on a synthesized input
cassiePython/NeRF-Art/render.py
code served (permissive licence) · get_code("f654b6ff433cdb1b")
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 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 numpy as np

# The normalize function takes a vector and normalizes it along a specified axis
# It uses:
# - vec: a numpy array
# - axis: an integer parameter (default -1)

# Create a small test vector with distinct dimensions
vec = np.random.randn(2, 4)  # Small batch size 2, feature dim 4

# Define arguments
ARGS = {
    "vec": vec,
    "axis": -1
}
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 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
poses_avg Ran
ran on a synthesized input
cassiePython/NeRF-Art/render.py
code served (permissive licence) · get_code("3306b86e39ec4155")
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 2, dtype family float, shape (4, 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 numpy as np

# Construct valid arguments for poses_avg function
# The function expects poses to be an array where:
# - poses[0, :3, 3] extracts the camera location (3D vector)
# - poses[0, :3, 2] extracts the forward direction (3D vector)
# - poses[0, :3, 1] extracts the up direction (3D vector)
# So poses should be at least shape (N, 4, 4) or similar

# Create a batch of pose matrices (4x4 homogeneous transformation matrices)
# Using distinct sizes: batch_size=2, matrix is 4x4
poses = np.array([
    # First pose matrix
    [
        [1.0, 0.0, 0.0, 0.5],  # right vector + x translation
        [0.0, 1.0, 0.0, 0.3],  # up vector + y translation
        [0.0, 0.0, 1.0, 2.0],  # forward vector + z translation (camera location z)
        [0.0, 0.0, 0.0, 1.0],  # homogeneous row
    ],
    # Second pose matrix
    [
        [0.8, -0.6, 0.0, 1.0],
        [0.6, 0.8, 0.0, -0.5],
        [0.0, 0.0, 1.0, 3.0],
        [0.0, 0.0, 0.0, 1.0],
    ]
], dtype=np.float32)

ARGS = {
    "poses": poses
}
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 2, dtype family float, shape (4, 4) — shape (4, 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
run_secant_method Ran
ran on a synthesized input
cassiePython/NeRF-Art/models/ray_casting.py
code served (permissive licence) · get_code("f46f4b5fcbfe8e9c")
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
import torch.nn as nn

torch.manual_seed(0)

# Create a simple implicit surface query function
class ImplicitSurfaceModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(3, 1)
    
    def forward(self, p):
        # p shape: (batch_size, 3)
        return self.linear(p)

implicit_surface_fn = ImplicitSurfaceModel()

# Parameters
batch_size = 2
ray_dim = 3

# Create ray origins and directions
rays_o_masked = torch.randn(batch_size, ray_dim)
rays_d_masked = torch.randn(batch_size, ray_dim)
rays_d_masked = rays_d_masked / (torch.norm(rays_d_masked, dim=1, keepdim=True) + 1e-8)

# Create initial function values and depths
f_low = torch.randn(batch_size)
f_high = torch.randn(batch_size)
d_low = torch.randn(batch_size)
d_high = torch.randn(batch_size)

# Ensure d_low < d_high for valid bounds
d_low = torch.min(d_low, d_high)
d_high = torch.max(d_low, d_high)

def implicit_surface_query_fn(p):
    return implicit_surface_fn(p)

n_secant_steps = 3
logit_tau = 0.5

ARGS = {
    "f_low": f_low,
    "f_high": f_high,
    "d_low": d_low,
    "d_high": d_high,
    "rays_o_masked": rays_o_masked,
    "rays_d_masked": rays_d_masked,
    "implicit_surface_query_fn": implicit_surface_query_fn,
    "n_secant_steps": n_secant_steps,
    "logit_tau": logit_tau,
}
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
view_matrix Ran
ran on a synthesized input
cassiePython/NeRF-Art/render.py
code served (permissive licence) · get_code("5343bc10cd19d430")
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 2, dtype family float, shape (4, 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 numpy as np

# Set seed for reproducibility
np.random.seed(0)

# Create small 3D vectors as numpy arrays
forward = np.array([0.0, 0.0, 1.0])
up = np.array([0.0, 1.0, 0.0])
cam_location = np.array([1.0, 2.0, 3.0])

ARGS = {
    "forward": forward,
    "up": up,
    "cam_location": cam_location,
}
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 2, dtype family float, shape (4, 4) — shape (4, 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
root_finding_surface_points Not yet run
unverified (harvested, no recorded run)
cassiePython/NeRF-Art/models/ray_casting.py
code served (permissive licence) · get_code("a2bcb809253c085b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 8 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
get_embedder 1 other paper
1 dated earlier
2104.10078 (2021-04)
UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction

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 2026-11-21.

Repositories linked to this paper

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

Abstract

As a powerful representation of 3D scenes, the neural radiance field (NeRF) enables high-quality novel view synthesis from multi-view images. Stylizing NeRF, however, remains challenging, especially on simulating a text-guided style with both the appearance and the geometry altered simultaneously. In this paper, we present NeRF-Art, a text-guided NeRF stylization approach that manipulates the style of a pre-trained NeRF model with a simple text prompt. Unlike previous approaches that either lack sufficient geometry deformations and texture details or require meshes to guide the stylization, our method can shift a 3D scene to the target style characterized by desired geometry and appearance variations without any mesh guidance. This is achieved by introducing a novel global-local contrastive learning strategy, combined with the directional constraint to simultaneously control both the trajectory and the strength of the target style. Moreover, we adopt a weight regularization method to effectively suppress cloudy artifacts and geometry noises which arise easily when the density field is transformed during geometry stylization. Through extensive experiments on various styles, we demonstrate that our method is effective and robust regarding both single-view stylization quality and cross-view consistency. The code and more results can be found in our project page: https://cassiepython.github.io/nerfart/.

For agents

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

get_harvested_code_for_paper("2212.08070")
get_code_for_paper("2212.08070")
have("2212.08070")

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

curl https://syntology.ai/api/ran/2212.08070.json

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

[![Syntology run record](https://syntology.ai/api/ran/2212.08070.svg)](https://syntology.ai/paper/2212.08070)

Connect an agent — have() is free.