SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.15627 · ICCV · 2021

Learning Generative Models of Textured 3D Meshes from Real-World Images

Aurelien Lucchi, Dario Pavllo, Thomas Hofmann, Jonas Kohler

arXiv · PDF · Open in the Atlas

Code that ran

6 of 11 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; 3 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 failures of our instrument; 5 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/2103.15627.json · badge: /api/ran/2103.15627.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
dariopavllo/textured-3d-gan canonical (classifier label)
classifier basis: census
6 ran · 5 no recorded run
FunctionStatusWhere it lives
diff_mask Ran
ran, honoured the contract we drafted
dariopavllo/textured-3d-gan/detection_tool.py
code served (permissive licence) · get_code("a13a79c65a8d64a4")
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 float, 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

torch.manual_seed(0)

# The function expects two tensors with the same shape
# It performs bitwise AND (&), sum(), and item() operations
# These operations work on boolean or integer tensors

parent = torch.tensor([[True, False, True, False],
                       [False, True, False, True]], dtype=torch.bool)

child = torch.tensor([[True, True, False, False],
                      [False, True, True, False]], dtype=torch.bool)

ARGS = {
    "parent": parent,
    "child": child
}
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 float, 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
divide_pred Ran
ran, honoured the contract we drafted
dariopavllo/textured-3d-gan/run_generation.py
code served (permissive licence) · get_code("86fe15b439328f16")
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 3, 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 numpy as np

torch.manual_seed(0)

# Test case 1: None input
pred_none = None

# Test case 2: Tensor input (batch size 4, feature dim 3)
pred_tensor = torch.randn(4, 3)

# Test case 3: List of tensors
pred_list = [torch.randn(4, 5), torch.randn(4, 7), None]

# We'll test with the tensor case as the primary argument
ARGS = {
    "pred": pred_tensor
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 3, 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
mean_iou Ran
ran; our contract draft was wrong, not the code
dariopavllo/textured-3d-gan/run_reconstruction.py
code served (permissive licence) · get_code("a3ccb1025ca05d3e")
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 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

torch.manual_seed(0)

# Create small tensors for alpha_pred and alpha_real
# Shape: (batch_size=2, height=4, width=3)
# Values between 0 and 1 to represent alpha/confidence scores
alpha_pred = torch.rand(2, 4, 3)
alpha_real = torch.rand(2, 4, 3)

ARGS = {
    "alpha_pred": alpha_pred,
    "alpha_real": alpha_real,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
mean_iou_noreduce Ran
ran; our contract draft was wrong, not the code
dariopavllo/textured-3d-gan/utils/losses.py
code served (permissive licence) · get_code("466a1a8724a2d313")
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)

# The function expects two tensors that can be compared with > 0.5,
# support bitwise operations (&, |), have a .float() method,
# and support .sum(dim=[-1, -2]) for summing over last two dimensions.
# This means they should be at least 3D tensors (batch, height, width).

# Create small tensors with shape (batch=2, height=4, width=3)
alpha_pred = torch.rand(2, 4, 3)
alpha_real = torch.rand(2, 4, 3)

ARGS = {
    "alpha_pred": alpha_pred,
    "alpha_real": alpha_real,
}
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
mean_miou_noreduce Ran
ran; our contract draft was wrong, not the code
dariopavllo/textured-3d-gan/utils/losses.py
code served (permissive licence) · get_code("de62206eae2ed5c5")
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 for pred and target
# The function expects tensors where it can sum over last 2 dimensions (dim=[-1, -2])
# and potentially sum over a batch dimension
# Shape: (batch=2, channels=4, height=3, width=3)
pred = torch.rand(2, 4, 3, 3)
target = torch.rand(2, 4, 3, 3)

# balanced parameter is a boolean
balanced = True

ARGS = {
    "pred": pred,
    "target": target,
    "balanced": balanced
}
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
positional_encoding Ran
ran, honoured the contract we drafted
dariopavllo/textured-3d-gan/models/gan.py
code served (permissive licence) · get_code("b7d3b52c37568cae")
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, 16, 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

# The function positional_encoding takes two integer parameters: Ny and Nx
# It uses them to create positional encoding arrays
# Ny is the height dimension, Nx is the width dimension
# The function performs linspace operations and meshgrid, then stacks results

ARGS = {
    "Ny": 16,
    "Nx": 8,
}
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, 16, 8) — shape (4, 16, 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
load_cmr_paths Not yet run
unverified (harvested, no recorded run)
dariopavllo/textured-3d-gan/detection_tool.py
code served (permissive licence) · get_code("e1f6c3cba13f3ab2")
How it ran

no invocation recorded.

loss_flat Not yet run
unverified (harvested, no recorded run)
dariopavllo/textured-3d-gan/utils/losses.py
code served (permissive licence) · get_code("5ce26bd44f33f167")
How it ran

no invocation recorded.

to_grid Not yet run
unverified (harvested, no recorded run)
dariopavllo/textured-3d-gan/pose_optimization_step1.py
code served (permissive licence) · get_code("bd0faab0cfcf45a0")
How it ran

no invocation recorded.

to_grid Not yet run
unverified (harvested, no recorded run)
dariopavllo/textured-3d-gan/run_reconstruction.py
code served (permissive licence) · get_code("e23deeeb06942c6d")
How it ran

no invocation recorded.

to_grid_tex Not yet run
unverified (harvested, no recorded run)
dariopavllo/textured-3d-gan/run_generation.py
code served (permissive licence) · get_code("7ab31145bc29a65d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 0 later) and 7 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
divide_pred 1 other paper
1 dated earlier
2006.07660 (2020-06)
Convolutional Generation of Textured 3D Meshes
mean_iou 1 other paper
1 dated earlier
2006.07660 (2020-06)
Convolutional Generation of Textured 3D Meshes
to_grid 1 other paper
1 dated earlier
2006.07660 (2020-06)
Convolutional Generation of Textured 3D Meshes
to_grid_tex 1 other paper
1 dated earlier
2006.07660 (2020-06)
Convolutional Generation of Textured 3D Meshes

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

Latent dimension for "light intensity" Figure 1. Left: we focus on GANs, where our generator outputs a triangle mesh and a UV-mapped texture. Middle: our method learns to synthesize textured 3D meshes given a real-world collection of 2D images. Top-right: we showcase a setting where we train a single model to generate all classes. This model successfully disentangles some factors of the 3D environment (e.g. lighting/shadows) without explicit supervision. Bottom-right: we also demonstrate a conditional model that generates meshes from 3D semantic layouts.

For agents

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

get_harvested_code_for_paper("2103.15627")
get_code_for_paper("2103.15627")
have("2103.15627")

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

curl https://syntology.ai/api/ran/2103.15627.json

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

[![Syntology run record](https://syntology.ai/api/ran/2103.15627.svg)](https://syntology.ai/paper/2103.15627)

Connect an agent — have() is free.