Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

MVPainter: Accurate and Detailed 3D Texture Generation via Multi-View Diffusion with Geometric Control

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read MVPainter claims that feeding a multi-view diffusion model the mesh's normal and depth maps, plus training on filtered, lighting-augmented data, yields textures that beat existing open methods on reference fidelity, geometry consistency…

desk verdict A well-engineered 3D texture generation system with clean ablations, but the SOTA claim is built on a VLM judge that the paper itself admits is weak on fine detail, so the headline ranking is not yet established. read the letter →

arxiv 2505.12635 v1 pith:ZVEU4Y7V submitted 2025-05-19 cs.CV

Mingqi Shao , Feng Xiong , Zhaoxu Sun , Mu Xu This is my paper · ORCID
classification cs.CV
keywords 3Dtexturegenerationmulti-viewdiffusiongeometricconditioningControlNetPBRmaterialextractionvision-languageevaluationbakingsingle-image
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tackles the bottleneck left by modern single-image-to-3D pipelines: after geometry is generated, the texture is often misaligned, wrong-colored, or blurry. MVPainter's claim is that these failures can be split into three addressable axes—reference-texture alignment, geometry-texture consistency, and local texture quality—and that a multi-view diffusion model conditioned on the mesh's normal and depth maps, trained on filtered and lighting-augmented data, improves all three at once. On 210 reference images evaluated with a vision-language-model Elo protocol, MVPainter scores above Hunyuan3D-2.0 and MV-Adapter on every axis, across geometries produced by four different generators. The paper also extracts basecolor, metallic, and roughness maps from the generated views, so the output is a PBR-ready textured mesh rather than a baked RGB texture. A sympathetic reader would take the contribution as a reproducible system-level recipe: better data, geometric control, and a perception-aligned evaluation protocol.

What carries the argument

The central object is the union ControlNet: a ControlNet whose condition encoder and transformer merge several geometric control images before the fused representation enters the diffusion backbone. MVPainter feeds it two complementary controls—a normal map for fine local surface orientation and a depth map for global shape and contours—so the generated six views are anchored to the actual mesh. The three-stage training plan (UNet multi-view pretraining, then ControlNet training with frozen UNet, then joint fine-tuning on a 100,000-object high-quality subset) is what carries the geometry-alignment and detail claims. A secondary mechanism is the PBR attribute extractor, which takes the six generated views and outputs basecolor, metallic, and roughness maps through parallel view/component/image attention at 512×512 resolution.

What would settle it

Run a larger human pairwise study—say 50 or more raters on the same 210 references and meshes—and compare the Elo ordering with QWen2.5-VL-32B's: if humans rank MVPainter below Hunyuan3D-2.0 or MV-Adapter on geometry-texture consistency or local texture quality, the paper's central state-of-the-art claim is refuted.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that a fixed multi-view RGB diffusion backbone becomes a state-of-the-art texturing engine once it is steered by explicit geometric conditioning and trained on deliberately chosen data. The steering is done by a union ControlNet that fuses normal maps and depth maps of the target mesh and feeds them into a frozen diffusion UNet; the data selection keeps samples with high color entropy and high Sobel-gradient texture complexity, and augments references with varied lighting and viewpoint perturbations so the model does not bake highlights or rely on the frontal-view shortcut. The result, measured by pairwise Elo ratings from QWen2.5-VL-32B, is that MVPainter ranks highest on reference alignment, geometry-texture consistency, and local texture quality for all four test geometry generators, and its PBR extractor—a parallelized, high-resolution variant of IDArb—produces basecolor/metallic/roughness maps judged closer to ground truth than IDArb's.

Load-bearing premise

The ranking depends on the assumption that QWen2.5-VL-32B's pairwise Elo ratings match human perception of fine texture quality, an agreement checked with only five human raters whose margins differed noticeably on the detail-sensitive axes.

Editorial extensions

If this is right

  • Texture quality can be upgraded for an already-generated mesh without retraining a 3D-native generator; normal-and-depth conditioning plus curated data is enough to beat RGB-only baselines.
  • Because the conditioning accepts multiple control modalities, the same architecture can absorb other geometric signals such as position maps or view-direction encodings.
  • Lighting and rotation augmentation prevent the model from copying the reference image too literally, so the generated texture does not carry baked-in speculars and shadows from the source photo.
  • Extracting PBR maps from the multi-view RGB output makes the textured mesh usable in standard physically based renderers, a step beyond baked textures.
  • Releasing the data-construction, training, and evaluation code makes the claimed gains independently checkable on other geometry generators.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the VLM pairwise Elo protocol is as human-aligned as the five-rater study suggests, the same protocol could cheaply benchmark related tasks—texture editing, relighting, or material prediction—without collecting large human preference sets.
  • The fixed six-view setup is the paper's acknowledged boundary; an adaptive viewpoint sampler that adds views where visibility or self-occlusion is high is a natural next step that the architecture's geometry conditioning already supports.
  • The data-selection recipe (HSV color entropy plus Sobel gradient magnitude, thresholded to 100k objects) is transferable: it could be used to curate training sets for other detail-sensitive generative models.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes MVPainter, a system for generating textures on a given 3D mesh from a single reference image. The method has three main components: a data pipeline that filters 3D assets by color entropy and texture complexity and augments reference views with varied lighting and rotations; a multi-view diffusion model with a union ControlNet conditioned on normal and depth maps, trained in three stages; and a PBR attribute extractor adapted from IDArb with parallel attention and higher resolution. The authors evaluate RGB texture generation against Hunyuan3D-2.0 and MV-Adapter on 210 reference images combined with geometries from TripoSG, Hunyuan3D-2.0, TRELLIS, and Hi3DGen, using pairwise QWen2.5-VL judgments converted to Elo scores, and report ablations showing the benefit of lighting augmentation, rotation augmentation, and high-quality fine-tuning. They also compare their PBR extractor qualitatively with IDArb and with commercial systems.

Significance. If the empirical claims hold, MVPainter is a practical and reproducible improvement: the manuscript gives enough training details to follow the pipeline, releases data construction tools and evaluation scripts, and the ablations isolate the contribution of each design choice. However, the headline claim of state-of-the-art performance is currently supported by a VLM-based Elo evaluation whose reliability is explicitly qualified by the authors for two of the three claimed dimensions, and the only human validation uses five raters without statistical analysis. The technical contribution is credible, but the evaluation needs strengthening before the paper's central claim is established.

major comments (3)
  1. [§3.1, Tables 1–2] The central state-of-the-art claim is not established for two of the three headline dimensions. The paper states in §3.1 that 'current VLMs still fall short of capturing fine-grained details as reliably as humans' for Geometry-Texture Consistency and Local Texture Quality, yet Table 1 reports MVPainter's superiority on exactly those dimensions using QWen2.5-VL Elo scores. The only human validation, Table 2, is based on five raters and shows a 12-point human margin (1192 vs 1180) where the VLM margin is 110 points (1136 vs 1026) on geometry-texture consistency for TripoSG geometries. No confidence intervals, inter-rater agreement, or significance tests are reported for either table. I request a substantially larger human study or an external perceptual metric before the 'human-aligned state-of-the-art' wording is used.
  2. [§2.4 and §4] The fixed six-viewpoint generation scheme is acknowledged in the conclusion to leave self-occluded objects incompletely covered, but the paper claims robust generalization 'across geometries generated by various methods' without restricting this scope. Since the evaluation set consists of common object types generated by SDXL/GPT-4o and may not include strongly self-occluded shapes, the generalization claim is broader than the tested regime. Please either evaluate on self-occluded shapes or explicitly scope the claim to objects adequately covered by the six viewpoints.
  3. [§3.2, Figs. 7–8] The PBR extractor contribution is supported only by qualitative side-by-side images. The claim that the improved extractor is 'comparable to commercial applications' is not backed by any quantitative comparison or user study on PBR maps, and the text at the end of §3.2 says 'Fig.7 shows that our PBR model can achieve comparable performance to commercial applications' although Fig.7 compares with IDArb while Fig.8 shows the commercial comparison. Please add a quantitative evaluation (e.g., basecolor/roughness error on a labeled subset of ARB-OBJAVERSE) or soften the claim.
minor comments (4)
  1. [§2.2] The phrase 'union contronet architecture' contains a typo; it should read 'union ControlNet architecture'.
  2. [§3.2] The final paragraph contains typos: 'futher' should be 'further' and 'out PBR extractor' should be 'our PBR extractor'.
  3. [§2.1, Eq. (7)] The fixed coefficient λ = 35 is stated without motivation or sensitivity analysis; a one-line statement of how this value was chosen would help reproducibility.
  4. [§3.1, after Eq. (12)] The Elo computation is averaged over 100 random shuffles, but the spread across shuffles is not reported; giving a standard deviation or range would make the stability of the ratings transparent.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: MVPainter's state-of-the-art claim rests on external comparisons and a held-out evaluation set, not on a fitted parameter or a self-citation chain.

full rationale

MVPainter's derivation chain is self-contained. The central SOTA claim is an empirical comparison: Table 1 reports Elo scores from pairwise VLM judgments of MVPainter against Hunyuan3D-2.0 and MV-Adapter on 210 GPT-4o/SDXL reference images with geometries from four external generators, and Table 2 compares the VLM against five human raters. No equation in Sec. 2 defines the generated textures in terms of the evaluation scores; the data-filtering metrics in Eqs. (2)-(7) are preprocessing criteria, not fitted parameters later renamed as predictions. The only self-citations, [24,25], appear in Sec. 1 as examples of geometry-generation methods and are not used to justify the texture-generation architecture or the evaluation. The paper's explicit admission that VLMs 'still fall short of capturing fine-grained details as reliably as humans' (Sec. 3.1) is a limitation on measurement validity, not circular derivation, because the ranking is still obtained from an external judge over held-out comparisons. No self-definitional, fitted-input, or imported-uniqueness step exists.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim is empirical; no new entities or formal axioms are introduced. The free parameters are data-processing choices that shape the training distribution and are not swept for sensitivity, so they weaken the generality of the reported SOTA.

free parameters (3)
  • lambda (quality score weight) = 35
    In Eq. 7 the combined quality score C_total = lambda * C_color + C_texture sets lambda = 35 with no sensitivity analysis or justification for the balance.
  • High-quality dataset cutoff = top 100,000 objects
    The top 100,000 of about 1.2M models are selected; the threshold is chosen by hand and no ablation varies it.
  • Reference view augmentation budget = 15 reference images per object; azimuth (-30,30), elevation (-10,30)
    The number of augmented lighting/rotation references and perturbation ranges are hand-chosen; no study shows how the model degrades with smaller ranges.
assumptions (4)
  • domain assumption Objaverse and web-collected models are representative of objects for texture generation
    Central training data source; if the distribution is unrepresentative, results may not generalize to in-the-wild inputs.
  • domain assumption QWen2.5-VL pairwise comparisons align with human quality judgments on the three dimensions
    Load-bearing evaluation premise; the user study has only five raters and shows discrepancies on fine-detail dimensions.
  • domain assumption Normal and depth maps are sufficient geometric controls for texture-geometry alignment
    The architecture relies on these two modalities; the paper notes extensibility but does not test other controls.
  • domain assumption IDArb-style intrinsic decomposition is a valid basis for PBR extraction
    The PBR module inherits IDArb's model and the ARB-OBJAVERSE training data; the paper modifies attention and resolution but keeps the underlying decomposition task.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MVPainter: Accurate and Detailed 3D Texture Generation via Multi-View Diffusion with Geometric Control." pith.science (2026). https://pith.science/paper/ZVEU4Y7V

@misc{pith2026250512635,
  author       = {Pith},
  title        = {Pith review of: MVPainter: Accurate and Detailed 3D Texture Generation via Multi-View Diffusion with Geometric Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZVEU4Y7V}},
  note         = {Machine review of arXiv:2505.12635}
}
read the original abstract

Recently, significant advances have been made in 3D object generation. Building upon the generated geometry, current pipelines typically employ image diffusion models to generate multi-view RGB images, followed by UV texture reconstruction through texture baking. While 3D geometry generation has improved significantly, supported by multiple open-source frameworks, 3D texture generation remains underexplored. In this work, we systematically investigate 3D texture generation through the lens of three core dimensions: reference-texture alignment, geometry-texture consistency, and local texture quality. To tackle these issues, we propose MVPainter, which employs data filtering and augmentation strategies to enhance texture fidelity and detail, and introduces ControlNet-based geometric conditioning to improve texture-geometry alignment. Furthermore, we extract physically-based rendering (PBR) attributes from the generated views to produce PBR meshes suitable for real-world rendering applications. MVPainter achieves state-of-the-art results across all three dimensions, as demonstrated by human-aligned evaluations. To facilitate further research and reproducibility, we also release our full pipeline as an open-source system, including data construction, model architecture, and evaluation tools.

Figures

Figures reproduced from arXiv: 2505.12635 by the authors.

Figure 1
Figure 1. MVPainter generates textures with excellent geometry alignment, reference fidelity, and rich detail, demonstrating strong capa [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Three key challenges in existing 3D texture generation [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of our framework. First, we apply data filtering and augmentation strategies to ensure that the training data contains [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Architecture of union controlnet. It performs multi [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Example of the concatenated image which then is evalu [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Qualitative comparison of baked RGB texture generation methods. Here, we only select the front and side views for illustration. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Qualitative comparison of our PBR extractor and its baseline IDArb. Our model can capture more accurate details than IDArb. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Our PBR extractor can achieve comparable performance to commercial applications. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 11
Figure 11. Figure 11: Importance of fine-tuning with high-quality dataset. [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]
Figure 10
Figure 10. Figure 10: Importance of the rotation augmentation. The model [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]
Figure 12
Figure 12. Figure 12: MVPainter’s performance on the geometries generated by TripoSG. The first column is the reference image, and the first row [PITH_FULL_IMAGE:figures/full_fig_p011_12.png]
Figure 13
Figure 13. Figure 13: MVPainter’s performance on the geometries generated by Hunyuan3D-2.0. [PITH_FULL_IMAGE:figures/full_fig_p012_13.png]
Figure 14
Figure 14. Figure 14: MVPainter’s performance on the geometries generated by TRELLIS. [PITH_FULL_IMAGE:figures/full_fig_p013_14.png]
Figure 15
Figure 15. Figure 15: MVPainter’s performance on the geometries generated by Hi3DGen. [PITH_FULL_IMAGE:figures/full_fig_p014_15.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. GeoCache: Training-Free Acceleration of Multi-View Texture Diffusion via Geometric Delta Transport

    cs.CV 2026-08 conditional novelty 7.0 of 10

    A training-free plugin that speeds up multi-view texture diffusion by transporting anchor views' per-step clean-signal updates to non-anchor views via geometry, preserving fidelity better than temporal caches at over ...

Reference graph

Works this paper leans on

28 extracted references · 10 canonical work pages · cited by 1 Pith paper

  1. [1]

    Meta 3d gen

    Raphael Bensadoun, Tom Monnier, Yanir Kleiman, Filippos Kokkinos, Yawar Siddiqui, Mahendra Kariya, Omri Harosh, Roman Shapovalov, Benjamin Graham, Emilien Garreau, et al. Meta 3d gen. arXiv preprint arXiv:2407.02599, 2024. 2

    arXiv 2024

  2. [2]

    Statistical analysis of the elo rating system in chess

    Arthur Berg. Statistical analysis of the elo rating system in chess. Chance, 33(3):31–38, 2020. 6

    work page 2020

  3. [3]

    Dreamavatar: Text-and-shape guided 3d hu- man avatar generation via diffusion models

    Yukang Cao, Yan-Pei Cao, Kai Han, Ying Shan, and Kwan- Yee K Wong. Dreamavatar: Text-and-shape guided 3d hu- man avatar generation via diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 958–968, 2024. 2

    work page 2024

  4. [4]

    Objaverse: A universe of annotated 3d objects

    Matt Deitke, Dustin Schwenk, Jordi Salvador, Luca Weihs, Oscar Michel, Eli VanderBilt, Ludwig Schmidt, Kiana Ehsani, Aniruddha Kembhavi, and Ali Farhadi. Objaverse: A universe of annotated 3d objects. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 13142–13153, 2023. 5

    work page 2023

  5. [5]

    Ac- tion2motion: Conditioned generation of 3d human motions

    Chuan Guo, Xinxin Zuo, Sen Wang, Shihao Zou, Qingyao Sun, Annan Deng, Minglun Gong, and Li Cheng. Ac- tion2motion: Conditioned generation of 3d human motions. In Proceedings of the 28th ACM International Conference on Multimedia, pages 2021–2029, 2020. 2

    2021

  6. [6]

    Game gener- ation via large language models

    Chengpeng Hu, Yunlong Zhao, and Jialin Liu. Game gener- ation via large language models. In 2024 IEEE Conference on Games (CoG), pages 1–4. IEEE, 2024. 2

    work page 2024

  7. [7]

    Humannorm: Learning normal diffusion model for high-quality and realistic 3d hu- man generation

    Xin Huang, Ruizhi Shao, Qi Zhang, Hongwen Zhang, Ying Feng, Yebin Liu, and Qing Wang. Humannorm: Learning normal diffusion model for high-quality and realistic 3d hu- man generation. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition , pages 4568–4577, 2024. 2

    2024

  8. [8]

    Mv-adapter: Multi-view consistent image generation made easy

    Zehuan Huang, Yuan-Chen Guo, Haoran Wang, Ran Yi, Lizhuang Ma, Yan-Pei Cao, and Lu Sheng. Mv-adapter: Multi-view consistent image generation made easy. arXiv preprint arXiv:2412.03632, 2024. 2, 6, 7

    arXiv 2024

Show all 28 references
  1. [9]

    Gpt-4o system card

    Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perel- man, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Weli- hinda, Alan Hayes, Alec Radford, et al. Gpt-4o system card. arXiv preprint arXiv:2410.21276, 2024. 6

    2024 arXiv

  2. [10]

    Vr-gs: A physical dynamics-aware interactive gaussian splatting system in virtual reality

    Ying Jiang, Chang Yu, Tianyi Xie, Xuan Li, Yutao Feng, Huamin Wang, Minchen Li, Henry Lau, Feng Gao, Yin Yang, et al. Vr-gs: A physical dynamics-aware interactive gaussian splatting system in virtual reality. In ACM SIG- GRAPH 2024 Conference Papers, pages 1–1, 2024. 2

    2024

  3. [11]

    Craftsman: High-fidelity mesh generation with 3d native generation and interactive geometry refiner

    Weiyu Li, Jiarui Liu, Rui Chen, Yixun Liang, Xuelin Chen, Ping Tan, and Xiaoxiao Long. Craftsman: High-fidelity mesh generation with 3d native generation and interactive geometry refiner. arXiv preprint arXiv:2405.14979, 2024. 2

    2024 arXiv

  4. [12]

    Advances in 3d generation: A survey

    Xiaoyu Li, Qi Zhang, Di Kang, Weihao Cheng, Yiming Gao, Jingbo Zhang, Zhihao Liang, Jing Liao, Yan-Pei Cao, and Ying Shan. Advances in 3d generation: A survey. arXiv preprint arXiv:2401.17807, 2024. 2

    2024 arXiv

  5. [13]

    Triposg: High-fidelity 3d shape synthesis using large-scale rectified flow models

    Yangguang Li, Zi-Xin Zou, Zexiang Liu, Dehu Wang, Yuan Liang, Zhipeng Yu, Xingchao Liu, Yuan-Chen Guo, Ding Liang, Wanli Ouyang, et al. Triposg: High-fidelity 3d shape synthesis using large-scale rectified flow models. arXiv preprint arXiv:2502.06608, 2025. 2, 6

    2025 arXiv

  6. [14]

    Idarb: Intrinsic decomposition for arbitrary number of input views and illuminations

    Zhibing Li, Tong Wu, Jing Tan, Mengchen Zhang, Jiaqi Wang, and Dahua Lin. Idarb: Intrinsic decomposition for arbitrary number of input views and illuminations. arXiv preprint arXiv:2412.12083, 2024. 5, 6, 7

    2024 arXiv

  7. [15]

    Sdxl: Improving latent diffusion mod- els for high-resolution image synthesis

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas M ¨uller, Joe Penna, and Robin Rombach. Sdxl: Improving latent diffusion mod- els for high-resolution image synthesis. arXiv preprint arXiv:2307.01952, 2023. 6

    2023 arXiv

  8. [16]

    Zero123++: a single image to consistent multi-view dif- fusion base model

    Ruoxi Shi, Hansheng Chen, Zhuoyang Zhang, Minghua Liu, Chao Xu, Xinyue Wei, Linghao Chen, Chong Zeng, and Hao Su. Zero123++: a single image to consistent multi-view dif- fusion base model. arXiv preprint arXiv:2310.15110, 2023. 4

    arXiv

  9. [17]

    Octgpt: Octree-based multi- scale autoregressive models for 3d shape generation

    Si-Tong Wei, Rui-Huan Wang, Chuan-Zhi Zhou, Baoquan Chen, and Peng-Shuai Wang. Octgpt: Octree-based multi- scale autoregressive models for 3d shape generation. arXiv preprint arXiv:2504.09975, 2025. 2

    2025 arXiv

  10. [18]

    Generative ai and the technological imagi- nary of game design

    Stefan Werning. Generative ai and the technological imagi- nary of game design. In Creative Tools and the Softwariza- tion of Cultural Production, pages 67–90. Springer, 2024. 2

    2024

  11. [19]

    Gpt-4v (ision) is a human-aligned evaluator for text-to-3d genera- tion

    Tong Wu, Guandao Yang, Zhibing Li, Kai Zhang, Ziwei Liu, Leonidas Guibas, Dahua Lin, and Gordon Wetzstein. Gpt-4v (ision) is a human-aligned evaluator for text-to-3d genera- tion. In Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition , pages 22...

  12. [20]

    Structured 3d latents for scalable and versatile 3d gen- eration

    Jianfeng Xiang, Zelong Lv, Sicheng Xu, Yu Deng, Ruicheng Wang, Bowen Zhang, Dong Chen, Xin Tong, and Jiaolong Yang. Structured 3d latents for scalable and versatile 3d gen- eration. arXiv preprint arXiv:2412.01506, 2024. 2, 6

    2024 arXiv

  13. [21]

    Controlnet++: All-in-one controlnet for image generations and editing! https : / / github

    xinsir6. Controlnet++: All-in-one controlnet for image generations and editing! https : / / github . com / xinsir6/ControlNetPlus, 2024. 4

    2024

  14. [22]

    An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, et al. Qwen2. 5 technical report. arXiv preprint arXiv:2412.15115, 2024. 7

    2024 arXiv

  15. [23]

    Hi3dgen: High-fidelity 3d geometry generation from images via normal bridging

    Chongjie Ye, Yushuang Wu, Ziteng Lu, Jiahao Chang, Xi- aoyang Guo, Jiaqing Zhou, Hao Zhao, and Xiaoguang Han. Hi3dgen: High-fidelity 3d geometry generation from images via normal bridging. arXiv preprint arXiv:2503.22236 , 3,

    arXiv

  16. [24]

    3d representation in 512-byte: Variational tokenizer is the key for autoregressive 3d generation

    Jinzhi Zhang, Feng Xiong, and Mu Xu. 3d representation in 512-byte: Variational tokenizer is the key for autoregressive 3d generation. arXiv preprint arXiv:2412.02202, 2024. 2

    2024 arXiv

  17. [25]

    G3pt: Un- leash the power of autoregressive modeling in 3d genera- tion via cross-scale querying transformer

    Jinzhi Zhang, Feng Xiong, and Mu Xu. G3pt: Un- leash the power of autoregressive modeling in 3d genera- tion via cross-scale querying transformer. arXiv preprint arXiv:2409.06322, 2024. 2

    2024 arXiv

  18. [26]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. In Proceedings of the IEEE/CVF international conference on computer vision, pages 3836–3847, 2023. 4

    2023

  19. [27]

    Clay: A controllable large-scale generative model for creat- ing high-quality 3d assets

    Longwen Zhang, Ziyu Wang, Qixuan Zhang, Qiwei Qiu, Anqi Pang, Haoran Jiang, Wei Yang, Lan Xu, and Jingyi Yu. Clay: A controllable large-scale generative model for creat- ing high-quality 3d assets. ACM Transactions on Graphics (TOG), 43(4):1–20, 2024. 2

    2024

  20. [28]

    Hunyuan3d 2.0: Scaling diffu- sion models for high resolution textured 3d assets generation

    Zibo Zhao, Zeqiang Lai, Qingxiang Lin, Yunfei Zhao, Haolin Liu, Shuhui Yang, Yifei Feng, Mingxin Yang, Sheng Zhang, Xianghui Yang, et al. Hunyuan3d 2.0: Scaling diffu- sion models for high resolution textured 3d assets generation. arXiv preprint arXiv:2501.12202, 2025. 2, 3, 6, 7

    2025 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.