Pith. sign in

REVIEW 1 major objections 5 minor 300 references

Controllable 3D digital humans are built in three stages—prior learning, personalization, and animation—and the literature sorts cleanly by body region and prior type.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-11 19:55 UTC pith:DKY47HRT

load-bearing objection Solid STAR that actually organizes the field; taxonomy is useful even if selection-dependent. the 1 major comments →

arxiv 2607.04341 v1 pith:DKY47HRT submitted 2026-07-05 cs.GR cs.CLcs.CVeess.IV

How to Build Digital Humans? From Priors to Photorealistic Avatars

Wojciech Zielonka , Tobias Kirschstein , Timo Bolkart , Simon Giebenhain , Vanessa Sklyarova , Xiang Deng , Donglai Xiang , Shunsuke Saito
show 3 more authors
Yebin Liu Matthias Niessner Justus Thies
This is my paper
classification cs.GR cs.CLcs.CVeess.IV
keywords digital humans3D avatarshuman priorsphotorealistic avatarsavatar taxonomyhead avatarsfull-body avatarslayered representations
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This state-of-the-art report claims that modern systems for creating controllable photorealistic 3D human avatars share a common three-stage pipeline: first learn general priors of human appearance and motion from data, then create a personalized avatar from identity inputs, and finally animate it with control signals. The authors focus on the first two stages and show that existing methods can be organized by a multi-axis taxonomy that includes body region (head, hair, hands, garments, full body) and the kind of prior used (3D morphable models, 3D GANs, 2D image or video diffusion, and others). They review representations, layered decompositions, and the shared principles that recur across regions, while pointing newcomers to the key literature and open problems such as efficiency, agency, hybrid 2D–3D solutions, and spatial awareness. A sympathetic reader cares because the taxonomy turns a sprawling literature into a practical map for building avatars that work from studio captures down to single images or text.

Core claim

Current controllable 3D avatar systems consist of three stages—learning priors of human appearance and motion, creating a personalized avatar from identity specifications, and animating the result—and existing work can be systematically categorized along body regions and the priors employed, as formalized by the report’s definitions, pipeline framework, and taxonomy tables.

What carries the argument

The three-stage avatar-system framework (prior learning → avatar creation → animation) together with the multi-axis taxonomy that places methods by body region (head, hair, hands, garments, full body) and prior type (3DMMs, 3D GANs, 2D/video diffusion, etc.); it organizes the literature and makes design choices comparable.

Load-bearing premise

The authors’ curated selection of papers from major computer-vision and graphics venues 2021–2025 plus arXiv preprints is representative enough that the resulting multi-axis taxonomy will stay stable and useful without major reordering from omitted work.

What would settle it

A large cluster of recent high-impact methods that cannot be placed into any of the taxonomy’s body-region or prior-type cells without forcing new top-level categories would show the claimed organization is incomplete.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • New avatar methods can be classified and compared by where they sit on the prior-type and body-region axes rather than by ad-hoc labels.
  • Layered representations that separate hair, hands, and garments become the natural route to editable, simulation-ready full-body avatars.
  • Under-constrained creation (single image or text) will continue to rely on stronger generative priors distilled from large 2D/3D datasets.
  • Open problems such as avatar agency, spatial awareness, hybrid 2D–3D pipelines, and standardized benchmarks are now framed as the next concrete research targets.
  • Practitioners gain a shared vocabulary and reference map that lowers the barrier for newcomers entering the field.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the three-stage framing holds, industrial pipelines for telepresence and virtual try-on will converge on modular prior libraries that can be swapped without redesigning the animation stage.
  • The same taxonomy suggests that progress on hair and garment layers will bottleneck full-body realism more than facial expression modeling in the near term.
  • Hybrid systems that use 3D avatars only as structural guides for video-diffusion models are a direct corollary of the report’s discussion of 2D versus 3D trade-offs.
  • Standardized public leaderboards for each taxonomy cell would make the claimed organization falsifiable and accelerate method comparison.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

1 major / 5 minor

Summary. This STAR surveys controllable 3D human avatar creation, organizing the literature around a three-stage pipeline (prior learning, personalized avatar creation, animation) while focusing on the first two stages. It defines 2D vs. 3D avatars, reviews core representations (meshes, NeRF, 3DGS, strands, sewing patterns), and introduces a multi-axis taxonomy (body region × prior type × representation × input modality) formalized in Tables 1–3 and applied to heads, hair, hands, garments, and full-body systems. The report covers studio-to-few-shot regimes, layered assets, material/relighting fundamentals, ethical risks, and open problems (agency, hybrid 2D/3D, benchmarks).

Significance. If the taxonomy holds, the paper supplies a durable organizing framework and entry point for a fast-moving, multi-subfield area. Strengths include the explicit three-stage system model (Fig. 2), the consistent legend and tables that make methods comparable, the layered treatment of hair/hands/garments, and the forward-looking discussion of agency, spatial awareness, and evaluation gaps. As a pure survey it does not claim new experiments; its value is synthesis and standardization for newcomers and practitioners.

major comments (1)
  1. The taxonomy’s claim to be systematic (Abstract; §1 Selection scheme; Tables 1–3) rests on the untested premise that the 2021–2025 major-venue + arXiv filter is representative. Related surveys (Wang et al., Gu et al., §2) are cited but coverage overlap is not quantified, nor is there any sensitivity check on the year/venue cutoff. A short appendix table of inclusion criteria, approximate paper counts per axis, and known near-misses would make the representativeness assumption falsifiable and would strengthen the central claim that the axes are stable rather than corpus-convenient.
minor comments (5)
  1. §4.1: the hybrid nature of 3DGS (explicit points + continuous kernels) is stated clearly; a one-sentence contrast with pure RBF literature would help readers new to the representation.
  2. Table 2 vs. Table 3: a few methods appear under both face and full-body or hair/garment columns without cross-reference; adding a note on multi-region methods would reduce ambiguity.
  3. §11.1 Avatar benchmarks: the NeRSemble leaderboard is a useful exception; listing one or two other public leaderboards (or their absence) for full-body and hands would make the evaluation gap more concrete.
  4. Scattered typos and line-break artifacts (e.g., “V olumetric”, “Nießner” variants) should be cleaned in production.
  5. Figure 1 caption lists many citations; ensuring every cited work appears in the main taxonomy tables would improve navigability.

Circularity Check

0 steps flagged

Survey taxonomy of external literature; no derivation chain that reduces claimed results to the authors' own fitted inputs or self-definitional premises.

full rationale

This is a state-of-the-art report that organizes existing controllable 3D avatar methods into a three-stage pipeline (prior learning, avatar creation, animation) and a multi-axis taxonomy (body region, prior type, representation, etc.), as formalized in Sections 3–4 and Tables 1–3. The paper does not claim first-principles derivations, uniqueness theorems, or quantitative predictions obtained by fitting parameters to data and then re-predicting related quantities. Self-citations (e.g., GaussianAvatars, NeRSemble, NPHM, INSTA) appear only as ordinary examples among many external works; they are not load-bearing for any uniqueness claim or forced ansatz. The selection scheme (major CV/CG venues 2021–2025 + arXiv, filtered by scope) is an explicit methodological choice for a survey, not a circular reduction of a result to its own inputs. No equation or claimed prediction collapses by construction to a quantity defined by the authors. Therefore the circularity score is 0.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

As a survey, the paper rests on domain conventions of computer graphics/vision rather than free parameters or newly postulated physical entities. The main load-bearing assumptions are definitional (what counts as a 2D/3D avatar and a prior) and selectional (which papers belong in the taxonomy).

axioms (3)
  • domain assumption An avatar system factors into (optional) prior learning, avatar creation from identity specs, and animation under control signals C = {M[,L],π}.
    Introduced in Section 3 and Figure 2 as the unifying framework; all subsequent categorization assumes this three-stage decomposition.
  • domain assumption 3D avatars are defined by an explicit 3D representation f_3D that is rendered by R, while 2D avatars synthesize images directly; both remain controllable by viewpoint π.
    Section 3 formal definition; used to separate literature throughout.
  • ad hoc to paper Papers published at major CV/CG venues 2021–2025 plus selected arXiv preprints, filtered by the authors’ scope, adequately represent the state of the art for taxonomy construction.
    Stated in Section 1 Selection scheme; the taxonomy’s completeness rests on this curation choice.

pith-pipeline@v1.1.0-grok45 · 58203 in / 2238 out tokens · 28356 ms · 2026-07-11T19:55:48.746621+00:00 · methodology

0 comments
read the original abstract

This state-of-the-art report provides an overview of controllable 3D human avatar creation. We describe current 3D avatar systems, which typically consist of three stages: (i) learning priors of human appearance and motion, (ii) creating a personalized avatar, and (iii) animating the avatar. To limit the scope, we focus on the prior learning and avatar creation stages. We define current avatar representations and introduce a taxonomy that categorizes existing work along multiple axes, including body regions and employed priors. We review methods for full-body and head avatars, as well as layered representations that decompose the body into components such as hands, hair, and garments. Finally, we outline common underlying principles, reference key literature for newcomers, and discuss open challenges and future research directions.

Figures

Figures reproduced from arXiv: 2607.04341 by Donglai Xiang, Justus Thies, Matthias Niessner, Shunsuke Saito, Simon Giebenhain, Timo Bolkart, Tobias Kirschstein, Vanessa Sklyarova, Wojciech Zielonka, Xiang Deng, Yebin Liu.

Figure 1
Figure 1. Figure 1: Building digital avatars requires considering many components, such as the human head, hair, hands, and garments, and ulti￾mately full-body avatars. Much like da Vinci’s Vitruvian Man symbolizes the ideal proportions and unity of the human form, the construction of digital avatars demands a holistic approach in which each part integrates into a coherent whole. Each of these elements presents distinct chall… view at source ↗
Figure 2
Figure 2. Figure 2: A Common Framework of Avatar Systems. Images adapted from FFHQ [KAL∗ 21], NPHM [GKG∗ 23], THuman [YZG∗ 21], and TaoAvatar [CHW∗ 25]. identity, such as editing instructions or alternative reference im￾ages, are incorporated during the avatar creation process which we will describe below. The animation controls M may include low￾level specifications, such as per-frame body joint angles, as well as high-level… view at source ↗
Figure 3
Figure 3. Figure 3: Types of 2D and 3D priors commonly used for digital humans. Images adapted from Controlnet [ZRA23], EG3D [CLC∗ 22], SMPL [LMR∗ 15], and Perm [HSS∗ 25]. model encoding generic beliefs about humans has to be employed to hallucinate or fill in unseen regions. Such human prior models can be quite diverse in their design (see [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Overview of different research fields in 3D Head Avatars. Representative images are taken from GaussianAvatars [QKS∗ 24], RGCA [SSS∗ 24], INSTA [ZBT23], Cao et al. [CSK∗ 22], GPHM [XWZ∗ 24], ROME [KSLZ22], HeadStudio [ZMF∗ 24], Next3D [SWW∗ 23], and GaussianSpeech [ASK∗ 24]. the dynamics by proposing MLP-based neural deformations con￾ditioned on M. NPGA [GKR∗ 24] focus on higher-fidelity motion control, by… view at source ↗
Figure 5
Figure 5. Figure 5: Hair reconstruction methods. Representative images are taken from HairCUP [KSN∗ 25], Difflocks [RWF∗ 25], Neural Hair￾cut [SCD∗ 23], HAAR [SZH∗ 23]. PCA space. Building on this idea, Im2Haircut [SZP∗ 25] trains a transformer-based prior to predict the PCA map from an input im￾age, followed by inversion in the prior space, improving efficiency and accuracy. The methods above aim to develop general solutions… view at source ↗
Figure 6
Figure 6. Figure 6: Hand reconstruction methods. Representative im￾ages are taken from NIMBLE [LZQ∗ 22], Handy [PPM∗ 23], UR￾Hand [CMG∗ 24], and HOLD [FPK∗ 24], respectively. due to its efficiency and fidelity. Similar to face 3DMMs [BV99], Khamis et al. [KTS∗ 15] apply PCA to the scans of 50 peo￾ple using a depth sensor to obtain a linear identity shape space. However, pose-dependent deformation is limited to linear blend sk… view at source ↗
Figure 7
Figure 7. Figure 7: Garment reconstruction methods. Representative images are taken from DressRecon [TXT∗ 25], ChatGarment [BXX∗ 25], HOOD [GBH23] and PhysAvatar [ZZY∗ 24], respectively. Compared with work that directly measures forces and deformation in a laboratory setting [WOR11, MBT∗ 12, CTT17, ZLB∗ 24], it is inherently more challenging to estimate these parameters from just visual input, yet more relevant to our discuss… view at source ↗
Figure 8
Figure 8. Figure 8: Overview of 3D full-body avatar research. Representative images are taken from SNARF [CZB∗ 21], TaoAvatar [CHW∗ 25], Vid2Avatar-Pro [GLK∗ 25], PersonNeRF [WSCKS23], and AniGS [QZZ∗ 25]. surface-aligned radiance volumes [SZZ∗ 22], mesh-anchored light fields [ZHY∗ 22,KLF∗ 23], real-time deep dynamic character mod￾els [HLX∗ 21, HLX∗ 23], tri-plane representations for controllable synthesis [ZZTH24], and signe… view at source ↗
Figure 9
Figure 9. Figure 9: Relighting requires proper disentangling of material properties such as normals, diffuse, and specular from illumi￾nation, often by using OLAT [DHT∗ 00] captures. Image from RFGCA [WSS∗ 25]. Illumination [RH01,SKS02] to relightable 3DGS avatars [SSS∗ 24, WSS∗ 25,SGN25]. Lambertian surfaces. When assuming a pure Lambertian surface, for which the BRDF is fr(x,l,v) = ρ(x)/π, the rendering equation is view-ind… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

300 extracted references · 8 linked inside Pith

  1. [1]

    M., Chen J., Chen J., Chen Z., Cheng S., et al

    Agrawal V., Akinyemi A., Alvero K., Behrooz M., Buffalini J., Carlucci F. M., Chen J., Chen J., Chen Z., Cheng S., et al. : Seamless interaction: Dyadic audiovisual motion modeling and large-scale dataset. arXiv preprint arXiv:2506.22554 (2025)

    Pith/arXiv arXiv 2025

  2. [2]

    G., Chaudhuri S., Saito J., Groueix T

    Aigerman N., Gupta K., Kim V. G., Chaudhuri S., Saito J., Groueix T. : Neural jacobian fields: learning intrinsic mappings of arbitrary meshes. Transactions on Graphics, (Proc. SIGGRAPH) 41, 4 (2022), 1--17

    2022

  3. [3]

    : Pose with style: Detail-preserving pose-guided image synthesis with conditional stylegan

    AlBahar B., Lu J., Yang J., Shu Z., Shechtman E., Huang J.-B. : Pose with style: Detail-preserving pose-guided image synthesis with conditional stylegan. Transactions on Graphics (TOG) 40, 6 (2021), 1--11

    2021

  4. [4]

    L., Theobalt C., Pons-Moll G

    Alldieck T., Magnor M., Bhatnagar B. L., Theobalt C., Pons-Moll G. : Learning to reconstruct people in clothing from a single rgb camera. In Conference on Computer Vision and Pattern Recognition (CVPR) (2019), pp. 1175--1186

    2019

  5. [5]

    : Detailed human avatars from monocular video

    Alldieck T., Magnor M., Xu W., Theobalt C., Pons-Moll G. : Detailed human avatars from monocular video. In 3DV (2018), pp. 98--109

    2018

  6. [6]

    : Video based reconstruction of 3d people models

    Alldieck T., Magnor M., Xu W., Theobalt C., Pons-Moll G. : Video based reconstruction of 3d people models. In Conference on Computer Vision and Pattern Recognition (CVPR) (2018), pp. 8387--8397

    2018

  7. [7]

    : Tex2shape: Detailed full human body geometry from a single image

    Alldieck T., Pons-Moll G., Theobalt C., Magnor M. : Tex2shape: Detailed full human body geometry from a single image. In International Conference on Computer Vision (ICCV) (2019), pp. 2293--2303

    2019

  8. [8]

    : An anisotropic phong brdf model

    Ashikhmin M., Shirley P. : An anisotropic phong brdf model. Journal of graphics tools 5, 2 (2000), 25--32

    2000

  9. [9]

    : Gaussianspeech: Audio-driven gaussian avatars, 2024

    Aneja S., Sevastopolsky A., Kirschstein T., Thies J., Dai A., Nießner M. : Gaussianspeech: Audio-driven gaussian avatars, 2024

    2024

  10. [10]

    : Clipface: Text-guided editing of textured 3d morphable models

    Aneja S., Thies J., Dai A., Nie ner M. : Clipface: Text-guided editing of textured 3d morphable models. In SIGGRAPH Conference Papers (SA) (2023)

    2023

  11. [11]

    Abdi H., Williams L. J. : Principal component analysis. Wiley interdisciplinary reviews: computational statistics 2, 4 (2010), 433--459

    2010

  12. [12]

    : Gaussian shell maps for efficient 3d human generation

    Abdal R., Yifan W., Shi Z., Xu Y., Po R., Kuang Z., Chen Q., Yeung D.-Y., Wetzstein G. : Gaussian shell maps for efficient 3d human generation. In Conference on Computer Vision and Pattern Recognition (CVPR) (2024), pp. 9441--9451

    2024

  13. [13]

    : Super-helices for predicting the dynamics of natural hair

    Bertails F., Audoly B., Cani M.-P., Querleux B., Leroy F., L\' e v\^ e que J.-L. : Super-helices for predicting the dynamics of natural hair. Transactions on Graphics, (Proc. SIGGRAPH) 25, 3 (2006), 1180--1187

    2006

  14. [14]

    W., Gross M

    Beeler T., Bickel B., Noris G., Beardsley P., Marschner S., Sumner R. W., Gross M. : Coupled 3d reconstruction of sparse facial hair and skin. Transactions on Graphics, (Proc. SIGGRAPH) 31, 4 (2012)

    2012

  15. [15]

    : Autoavatar: Autoregressive neural fields for dynamic avatar modeling

    Bai Z., Bagautdinov T., Romero J., Zollh \"o fer M., Tan P., Saito S. : Autoavatar: Autoregressive neural fields for dynamic avatar modeling. In European Conference on Computer Vision (ECCV) (2022), Springer, pp. 222--239

    2022

  16. [16]

    C., Dale D., Dong N., Duppenthaler M., Duquenne P.-A., Ellis B., Elsahar H., Haaheim J., et al

    Barrault L., Chung Y.-A., Meglioli M. C., Dale D., Dong N., Duppenthaler M., Duquenne P.-A., Ellis B., Elsahar H., Haaheim J., et al. : Seamless: Multilingual expressive and streaming speech translation. arXiv (2023)

    2023

  17. [17]

    : Stable video diffusion: Scaling latent video diffusion models to large datasets

    Blattmann A., Dockhorn T., Kulal S., Mendelevitch D., Kilian M., Lorenz D., Levi Y., English Z., Voleti V., Letts A., et al. : Stable video diffusion: Scaling latent video diffusion models to large datasets. arXiv (2023)

    2023

  18. [18]

    Baron Fourier J. B. J., et al. : The analytical theory of heat. Courier Corporation, 2003

    2003

  19. [19]

    : Secure digital camera

    Blythe P., Fridrich J. : Secure digital camera. Digital Investigation (2004)

    2004

  20. [20]

    : Generative neural articulated radiance fields

    Bergman A., Kellnhofer P., Yifan W., Chan E., Lindell D., Wetzstein G. : Generative neural articulated radiance fields. Advances in Neural Information Processing Systems (NeurIPS) 35 (2022), 19900--19916

    2022

  21. [21]

    : Cloth3d: clothed 3d humans

    Bertiche H., Madadi M., Escalera S. : Cloth3d: clothed 3d humans. In European Conference on Computer Vision (ECCV) (2020), Springer, pp. 344--359

    2020

  22. [22]

    : Pbns: physically based neural simulation for unsupervised garment pose space deformation

    Bertiche H., Madadi M., Escalera S. : Pbns: physically based neural simulation for unsupervised garment pose space deformation. Transactions on Graphics, (Proc. SIGGRAPH Asia) 40, 6 (2021), 1--14

    2021

  23. [23]

    : Neural cloth simulation

    Bertiche H., Madadi M., Escalera S. : Neural cloth simulation. Transactions on Graphics, (Proc. SIGGRAPH) 41, 6 (2022), 1--14

    2022

  24. [24]

    : Cgs-gan: 3d consistent gaussian splatting gans for high resolution human head synthesis, 2025

    Barthel F., Morgenstern W., Hinzer P., Hilsmann A., Eisert P. : Cgs-gan: 3d consistent gaussian splatting gans for high resolution human head synthesis, 2025

    2025

  25. [25]

    : Projective dynamics: Fusing constraint projections for fast simulation

    Bouaziz S., Martin S., Liu T., Kavan L., Pauly M. : Projective dynamics: Fusing constraint projections for fast simulation. Transactions on Graphics, (Proc. SIGGRAPH) 33, 4 (2014), 154:1--154:11

    2014

  26. [26]

    : Deepsd: Automatic deep skinning and pose space deformation for 3d garment animation

    Bertiche H., Madadi M., Tylson E., Escalera S. : Deepsd: Automatic deep skinning and pose space deformation for 3d garment animation. In Conference on Computer Vision and Pattern Recognition (CVPR) (2021), pp. 5471--5480

    2021

  27. [27]

    : Dynamic surface function networks for clothed human bodies

    Burov A., Nie ner M., Thies J. : Dynamic surface function networks for clothed human bodies. In International Conference on Computer Vision (ICCV) (2021), pp. 10754--10764

    2021

  28. [28]

    : Markerless garment capture

    Bradley D., Popa T., Sheffer A., Heidrich W., Boubekeur T. : Markerless garment capture. Transactions on Graphics, (Proc. SIGGRAPH) (2008), 1--9

    2008

  29. [29]

    W., Fidler S., Kreis K

    Blattmann A., Rombach R., Ling H., Dockhorn T., Kim S. W., Fidler S., Kreis K. : Align your latents: High-resolution video synthesis with latent diffusion models. In Conference on Computer Vision and Pattern Recognition (CVPR) (2023), pp. 22563--22575

    2023

  30. [30]

    : Gaussian shadow casting for neural characters

    Bolanos L., Su S.-Y., Rhodin H. : Gaussian shadow casting for neural characters. In Conference on Computer Vision and Pattern Recognition (CVPR) (2024), pp. 20997--21006

    2024

  31. [31]

    L., Tiwari G., Theobalt C., Pons-Moll G

    Bhatnagar B. L., Tiwari G., Theobalt C., Pons-Moll G. : Multi-garment net: Learning to dress 3d people from images. In International Conference on Computer Vision (ICCV) (2019), pp. 5420--5430

    2019

  32. [32]

    : A morphable model for the synthesis of 3d faces

    Blanz V., Vetter T. : A morphable model for the synthesis of 3d faces. In Proc. SIGGRAPH (1999), ACM Press/Addison-Wesley Publishing Co., pp. 187--194

    1999

  33. [33]

    : Large steps in cloth simulation

    Baraff D., Witkin A. : Large steps in cloth simulation. In SIGGRAPH Conference Papers (SA). 1998, pp. 43--54

    1998

  34. [34]

    : Driving-signal aware full-body avatars

    Bagautdinov T., Wu C., Simon T., Prada F., Shiratori T., Wei S.-E., Xu W., Sheikh Y., Saragih J. : Driving-signal aware full-body avatars. Transactions on Graphics, (Proc. SIGGRAPH) 40, 4 (2021), 1--17

    2021

  35. [35]

    J., Feng Y

    Bian S., Xu C., Xiu Y., Grigorev A., Liu Z., Lu C., Black M. J., Feng Y. : Chatgarment: Garment estimation, generation and editing via large language models. In Conference on Computer Vision and Pattern Recognition (CVPR) (2025), pp. 2924--2934

    2025

  36. [36]

    C., Beatson R

    Carr J. C., Beatson R. K., Cherrie J. B., Mitchell T. J., Fright W. R., McCallum B. C., Evans T. R. : Reconstruction and representation of 3d objects with radial basis functions. In Proceedings of the 28th annual conference on Computer graphics and interactive techniques (2001), pp. 67--76

    2001

  37. [37]

    J.-Y., Bitterli B., Tappan C., Burley B

    Chiang M. J.-Y., Bitterli B., Tappan C., Burley B. : A practical and controllable hair and fur model for production path tracing. In SIGGRAPH Conference Papers (SA) (2015)

    2015

  38. [38]

    P., Wang Y.-X

    Chen J.-K., Bansal A., Vo M. P., Wang Y.-X. : Virtual fitting room: Generating arbitrarily long videos of virtual try-on from a single image. In Advances in Neural Information Processing Systems (NeurIPS) (2025)

    2025

  39. [39]

    : Dna-rendering: A diverse neural actor repository for high-fidelity human-centric rendering

    Cheng W., Chen R., Fan S., Yin W., Chen K., Cai Z., Wang J., Gao Y., Yu Z., Lin Z., et al. : Dna-rendering: A diverse neural actor repository for high-fidelity human-centric rendering. In International Conference on Computer Vision (ICCV) (2023), pp. 19982--19993

    2023

  40. [40]

    V., H\" a ne C., Nam G., Bazin J.-C., De la Torre F

    Chen Y., Carrasco F. V., H\" a ne C., Nam G., Bazin J.-C., De la Torre F. : Doubly hierarchical geometric representations for strand-based human hairstyle generation. In Advances in Neural Information Processing Systems (NeurIPS) (2025)

    2025

  41. [41]

    J., Bolkart T

    Chhatre K., Daněček R., Athanasiou N., Becherini G., Peters C., Black M. J., Bolkart T. : AMUSE : Emotional speech-driven 3D body animation via disentangled latent diffusion. In Conference on Computer Vision and Pattern Recognition (CVPR) (2024), pp. 1942--1953

    2024

  42. [42]

    : Renderman: An advanced path-tracing architecture for movie rendering

    Christensen P., Fong J., Shade J., Wooten W., Schubert B., Kensler A., Friedman S., Kilpatrick C., Ramshaw C., Bannister M., Rayner B., Brouillat J., Liani M. : Renderman: An advanced path-tracing architecture for movie rendering. Transactions on Graphics (TOG) 37, 3 (2018)

    2018

  43. [43]

    : Wan-animate: Unified character animation and replacement with holistic replication, 2025

    Cheng G., Gao X., Hu L., Hu S., Huang M., Ji C., Li J., Meng D., Qi J., Qiao P., Shen Z., Song Y., Sun K., Tian L., Wang F., Wang G., Wang Q., Wang Z., Xiao J., Xu S., Zhang B., Zhang P., Zhang X., Zhang Z., Zhou J., Zhuo L. : Wan-animate: Unified character animation and replacement with holistic replication, 2025

    2025

  44. [44]

    Chan C., Ginosar S., Zhou T., Efros A. A. : Everybody dance now. In International Conference on Computer Vision (ICCV) (2019), pp. 5933--5942

    2019

  45. [45]

    : Generalizable and animatable gaussian head avatar

    Chu X., Harada T. : Generalizable and animatable gaussian head avatar. Advances in Neural Information Processing Systems (NeurIPS) 37 (2024), 57642--57670

    2024

  46. [46]

    : Primdiffusion: Volumetric primitives diffusion for 3d human generation

    Chen Z., Hong F., Mei H., Wang G., Yang L., Liu Z. : Primdiffusion: Volumetric primitives diffusion for 3d human generation. Advances in Neural Information Processing Systems (NeurIPS) 36 (2023), 13664--13677

    2023

  47. [47]

    : Lisa: Learning implicit shape and appearance of hands

    Corona E., Hodan T., Vo M., Moreno-Noguer F., Sweeney C., Newcombe R., Ma L. : Lisa: Learning implicit shape and appearance of hands. In Conference on Computer Vision and Pattern Recognition (CVPR) (2022), pp. 20533--20543

    2022

  48. [48]

    : Taoavatar: Real-time lifelike full-body talking avatars for augmented reality via 3d gaussian splatting

    Chen J., Hu J., Wang G., Jiang Z., Zhou T., Chen Z., Lv C. : Taoavatar: Real-time lifelike full-body talking avatars for augmented reality via 3d gaussian splatting. In Conference on Computer Vision and Pattern Recognition (CVPR) (2025)

    2025

  49. [49]

    J., Geiger A., Hilliges O

    Chen X., Jiang T., Song J., Yang J., Black M. J., Geiger A., Hilliges O. : gdna: Towards generative detailed neural avatars. In Conference on Computer Vision and Pattern Recognition (CVPR) (2022), pp. 20427--20437

    2022

  50. [50]

    : Does 3d gaussian splatting need accurate volumetric rendering? In Computer Graphics Forum (2025), Wiley Online Library, p

    Celarek A., Kopanas G., Drettakis G., Wimmer M., Kerbl B. : Does 3d gaussian splatting need accurate volumetric rendering? In Computer Graphics Forum (2025), Wiley Online Library, p. e70032

    2025

  51. [51]

    L., Kress R., Kress R

    Colton D. L., Kress R., Kress R. : Inverse acoustic and electromagnetic scattering theory, vol. 93. Springer, 1998

    1998

  52. [52]

    : Relighting4d: Neural relightable human from videos

    Chen Z., Liu Z. : Relighting4d: Neural relightable human from videos. In European Conference on Computer Vision (ECCV) (2022), Springer, pp. 606--623

    2022

  53. [53]

    R., Lin C

    Chan E. R., Lin C. Z., Chan M. A., Nagano K., Pan B., De Mello S., Gallo O., Guibas L. J., Tremblay J., Khamis S., et al. : Efficient geometry-aware 3d generative adversarial networks. In Conference on Computer Vision and Pattern Recognition (CVPR) (2022), Computer Vision Foundation / IEEE , pp. 16123--16133

    2022

  54. [54]

    : Relightable neural actor with intrinsic decomposition and pose control

    Carbonera Luvizon D., Golyanik V., Kortylewski A., Habermann M., Theobalt C. : Relightable neural actor with intrinsic decomposition and pose control. In European Conference on Computer Vision (ECCV) (2024), pp. 465--483

    2024

  55. [55]

    : High-quality hair modeling from a single portrait photo

    Chai M., Luo L., Sunkavalli K., Carr N., Hadap S., Zhou K. : High-quality hair modeling from a single portrait photo. Transactions on Graphics, (Proc. SIGGRAPH) 34, 6 (2015)

    2015

  56. [56]

    H., Liu Z., Yang Y., Yuksel C

    Chen A. H., Liu Z., Yang Y., Yuksel C. : Vertex block descent. Transactions on Graphics, (Proc. SIGGRAPH) 43, 4 (2024), 1--16

    2024

  57. [57]

    : Hunyuanvideo-avatar: High-fidelity audio-driven human animation for multiple characters, 2025

    Chen Y., Liang S., Zhou Z., Huang Z., Ma Y., Tang J., Lin Q., Zhou Y., Lu Q. : Hunyuanvideo-avatar: High-fidelity audio-driven human animation for multiple characters, 2025

    2025

  58. [58]

    : Urhand: Universal relightable hands

    Chen Z., Moon G., Guo K., Cao C., Pidhorskyi S., Simon T., Joshi R., Dong Y., Xu Y., Pires B., et al. : Urhand: Universal relightable hands. In Conference on Computer Vision and Pattern Recognition (CVPR) (2024), pp. 119--129

    2024

  59. [59]

    P., Tang S., Tzionas D., Black M

    Choutas V., M \"u ller L., Huang C.-H. P., Tang S., Tzionas D., Black M. J. : Accurate 3d body shape regression using metric and semantic attributes. In Conference on Computer Vision and Pattern Recognition (CVPR) (2022), pp. 2718--2728

    2022

  60. [60]

    : Cloth and skin deformation with a triangle mesh based convolutional neural network

    Chentanez N., Macklin M., M "u ller M., Jeschke S., Kim T.-Y. : Cloth and skin deformation with a triangle mesh based convolutional neural network. Computer Graphics Forum (CGF) 39, 8 (2020), 123--134

    2020

  61. [61]

    : Interactavatar: Modeling hand-face interaction in photorealistic avatars with deformable gaussians

    Chen K., Mohan S., Theiss J., Oprea S., Sridhar S., Prakash A. : Interactavatar: Modeling hand-face interaction in photorealistic avatars with deformable gaussians. In International Conference on Computer Vision (ICCV) (2025), pp. 10410--10420

    2025

  62. [62]

    : Foundhand: Large-scale domain-specific learning for controllable hand image generation

    Chen K., Min C., Zhang L., Hampali S., Keskin C., Sridhar S. : Foundhand: Large-scale domain-specific learning for controllable hand image generation. In Conference on Computer Vision and Pattern Recognition (CVPR) (2025), pp. 17448--17460

    2025

  63. [63]

    Community B. O. : Blender - a 3d modelling and rendering package, 2018

    2018

  64. [64]

    : Smplicit: Topology-aware generative model for clothed people

    Corona E., Pumarola A., Alenya G., Pons-Moll G., Moreno-Noguer F. : Smplicit: Topology-aware generative model for clothed people. In Conference on Computer Vision and Pattern Recognition (CVPR) (2021), pp. 11875--11885

    2021

  65. [65]

    : Viton-hd: High-resolution virtual try-on via misalignment-aware normalization

    Choi S., Park S., Lee M., Choo J. : Viton-hd: High-resolution virtual try-on via misalignment-aware normalization. In Conference on Computer Vision and Pattern Recognition (CVPR) (2021), pp. 14131--14140

    2021

  66. [66]

    : Zero-shot detection of ai-generated images, 2024

    Cozzolino D., Poggi G., Nießner M., Verdoliva L. : Zero-shot detection of ai-generated images, 2024

    2024

  67. [67]

    : D\^ 3-human: Dynamic disentangled digital human from monocular video

    Chen H., Peng B., Tao Y., Zhang J. : D\^ 3-human: Dynamic disentangled digital human from monocular video. In Conference on Computer Vision and Pattern Recognition (CVPR) (2025), pp. 10836--10846

    2025

  68. [68]

    : Id-reveal: Identity-aware deepfake video detection

    Cozzolino D., R \"o ssler A., Thies J., Nie ner M., Verdoliva L. : Id-reveal: Identity-aware deepfake video detection. arXiv (2020)

    2020

  69. [69]

    K., Schwartz G., Zollhoefer M., Saito S., Lombardi S., Wei S.-E., Belko D., Yu S.-I., et al

    Cao C., Simon T., Kim J. K., Schwartz G., Zollhoefer M., Saito S., Lombardi S., Wei S.-E., Belko D., Yu S.-I., et al. : Authentic volumetric avatars from a phone scan. Transactions on Graphics, (Proc. SIGGRAPH) 41, 4 (2022), 1--19

    2022

  70. [70]

    : Autohair: Fully automatic hair modeling from a single image

    Chai M., Shao T., Wu H., Weng Y., Zhou K. : Autohair: Fully automatic hair modeling from a single image. Transactions on Graphics, (Proc. SIGGRAPH) 35 (2016), 1--12

    2016

  71. [71]

    L., Torrance K

    Cook R. L., Torrance K. E. : A reflectance model for computer graphics. Transactions on Graphics, (Proc. SIGGRAPH) 1, 1 (1982), 7--24

    1982

  72. [72]

    : Forensictransfer: Weakly-supervised domain adaptation for forgery detection

    Cozzolino D., Thies J., R \"o ssler A., Riess C., Nie ner M., Verdoliva L. : Forensictransfer: Weakly-supervised domain adaptation for forgery detection. arXiv (2018)

    2018

  73. [73]

    : Modeling and data-driven parameter estimation for woven fabrics

    Clyde D., Teran J., Tamstorf R. : Modeling and data-driven parameter estimation for woven fabrics. In Symposium on Computer Animation (SCA) (2017), pp. 1--11

    2017

  74. [74]

    : Hand avatar: Free-pose hand animation and rendering from monocular video

    Chen X., Wang B., Shum H.-Y. : Hand avatar: Free-pose hand animation and rendering from monocular video. In Conference on Computer Vision and Pattern Recognition (CVPR) (2023), pp. 8683--8693

    2023

  75. [75]

    : Single-view hair modeling for portrait manipulation

    Chai M., Wang L., Weng Y., Yu Y., Guo B., Zhou K. : Single-view hair modeling for portrait manipulation. Transactions on Graphics, (Proc. SIGGRAPH) 31, 4 (2012)

    2012

  76. [76]

    : Dynamic hair manipulation in images and videos

    Chai M., Wang L., Weng Y., Jin X., Zhou K. : Dynamic hair manipulation in images and videos. Transactions on Graphics, (Proc. SIGGRAPH) 32, 4 (2013)

    2013

  77. [77]

    : Structure-preserving 3d garment modeling with neural sewing machines

    Chen X., Wang G., Zhu D., Liang X., Torr P., Lin L. : Structure-preserving 3d garment modeling with neural sewing machines. Advances in Neural Information Processing Systems (NeurIPS) 35 (2022), 15147--15159

    2022

  78. [78]

    : EgoAvatar : Egocentric view-driven and photorealistic full-body avatars

    Chen J., Wang J., Zhang Y., Pandey R., Beeler T., Habermann M., Theobalt C. : EgoAvatar : Egocentric view-driven and photorealistic full-body avatars. In SIGGRAPH Asia Conference Papers (SA) (2024), pp. 1--11

    2024

  79. [79]

    : Diffusionfake: Enhancing generalization in deepfake detection via guided stable diffusion

    Chen S., Yao T., Liu H., Sun X., Ding S., Ji R., et al. : Diffusionfake: Enhancing generalization in deepfake detection via guided stable diffusion. Advances in Neural Information Processing Systems 37 (2024), 101474--101497

    2024

  80. [80]

    J., Hilliges O., Geiger A

    Chen X., Zheng Y., Black M. J., Hilliges O., Geiger A. : Snarf: Differentiable forward skinning for animating non-rigid neural implicit shapes. In International Conference on Computer Vision (ICCV) (2021), pp. 11594--11604

    2021

Showing first 80 references.