Pith. sign in

REVIEW 4 major objections 5 minor 171 references

A five-parameter description of hair texture, disentangled from style via centerlines, lets generative models create and transfer feature-rich strands.

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

2026-07-14 15:22 UTC pith:SHDOEPMQ

load-bearing objection Useful CG pipeline for controllable curly/afro hair texture transfer; the 5D “bijection” and labeling fidelity are overstated but the engineering is real. the 4 major comments →

arxiv 2607.09811 v1 pith:SHDOEPMQ submitted 2026-07-10 cs.CV

Detangled: A Framework for Creating, Editing, and Inferencing Feature Rich Hair Strands

Sarah Jobalia , Yitong Deng , Carolyn Smith , Ronald Fedkiw This is my paper
classification cs.CV
keywords hair strandsstrand texturecenterlinecanonical spacetexture transferdiffusion modelstyle guidesdisentanglement
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 paper argues that the persistent, internal shape of a hair strand—curling, twisting, and related patterns—can be captured by five accessible parameters (thickness, curl radius, curl wavelength, twist prevalence, and porosity) that form a practical bijection with real hair textures. Because those parameters should not depend on how the strand is oriented on the head, the authors extract a centerline (style guide) for each strand and map the strand into a canonical space where the centerline is straight. A novel optimization improves the centerline so that low-frequency style is cleanly separated from high-frequency texture. With strands in that space, a diffusion model generates new strands that match any desired texture parameters; a second network continuously labels candidate strands so the diffusion process can be supervised. The same machinery lets an artist replace the texture of an existing groom while preserving its overall style, either from numeric parameters or from a few example strands. The result is a practical route to modeling curly and afro-textured hair that conventional tools and data-driven methods have long treated as failure cases.

Core claim

A five-dimensional parameter space (thickness, curl radius, curl wavelength, twist prevalence, porosity) is intended as a bijection with naturally occurring strand textures; once strands are mapped into a centerline-based canonical space that removes overall direction, a diffusion model supervised by a texture-labeling network can generate new strands of any desired texture and transfer that texture onto existing grooms without changing their style guides.

What carries the argument

The centerline-to-canonical-space map: each strand is straightened so its style guide lies on the negative z-axis, after which residual geometry is pure texture; generation and labeling both operate in this space, and the inverse map restores world-space placement.

Load-bearing premise

That five scalar parameters plus one particular centerline optimization fully capture the visually and mechanically relevant texture of real hair after simulation and drying, so a network trained mostly on synthetic data plus roughly a thousand hand labels can supervise generation across the natural range.

What would settle it

Take a set of real, high-resolution strand scans spanning the claimed texture range, extract their five parameters with the labeling network, regenerate strands from those parameters alone, and measure whether the regenerated strands are statistically indistinguishable from the originals under the same geometric and visual metrics the paper uses for its synthetic results.

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

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

4 major / 5 minor

Summary. The paper proposes a framework for describing, generating, and editing feature-rich hair strands by defining strand texture via a five-dimensional parameter space (thickness, curl radius, curl wavelength, twist prevalence, porosity) intended as a bijection with natural textures. It disentangles texture from overall style by computing refined centerlines (bootstrapped from DCT then optimized for radial-distance variance plus smoothness regularizers) that map strands into a canonical texture space with straight style guides. A diffusion model (1D U-Net on curvature-torsion-speed signals) generates new strands in this space, supervised by a separate labeling network (bi-LSTM+attention for twist prevalence; transformer for characteristic torsion, with procedural low-frequency curvature) that recovers the parameters; this enables texture transfer or parameter-driven editing of existing grooms while preserving style guides, with qualitative results on diverse hair types.

Significance. If the parameterization and supervision hold, the work would be a meaningful advance for computer graphics and vision: it directly targets the chronic under-representation of curly and afro-textured hair in digital media and generative models, supplies an artist- and stylist-accessible control space grounded in scientific and cultural sources, and demonstrates a practical disentanglement-plus-diffusion pipeline that can retrofit existing grooms. The centerline canonicalization, Frenet-Serret reconstruction, and measurement-guided diffusion are concrete technical contributions that could transfer to other strand- or curve-based assets. The absence of quantitative validation currently limits the strength of the claim, but the direction is timely and socially relevant.

major comments (4)
  1. [Abstract, §3, §5.2] Abstract and §3 claim the five-dimensional space is 'intended to be a bijection with naturally occurring hair strand textures' and is both qualitatively accessible and quantitatively precise for generation. Thickness and porosity never enter the labeling network (§5.2) or the diffusion conditioning/guidance (§6–7); they affect only mass/stiffness during offline data simulation. For the remaining three parameters the labeling network is trained almost entirely on unsimulated procedural helices (known ground-truth by construction) and fine-tuned on only ~1k hand-labeled simulated strands. No recovery metrics (MAE, correlation, confusion matrices, or even qualitative error analysis) are reported on held-out simulated or real strands after gravity, period-skipping and drying. Without demonstrated invertibility or labeling fidelity the bijection and the controllable generation remain aspirati
  2. [§5.2, §6.4, §7.2] The generative model is supervised by the labeling network of §5.2 (and, for parameter-driven editing, by the same network at inference time via Diffusion Posterior Sampling). Because no quantitative evaluation of that network is supplied, it is impossible to assess whether the guidance signal is accurate enough to enforce the claimed texture parameters. A minimal ablation—recovery error on a held-out simulated set, or correlation between target and re-labeled generated strands—would be required to support the central claim that the pipeline produces strands 'conforming to any desired texture'.
  3. [§7.5, Figs. 12–16] All results (§7.5, Figs. 1, 8–16) are purely qualitative renderings of texture transfer and failure-case repairs. There are no quantitative metrics of generation fidelity, style preservation, interpenetration rates, or comparison against prior strand generators or artist baselines. Given that the paper positions itself as enabling robust creation and editing of underrepresented hair types, the lack of any measurable evaluation leaves the practical efficacy of the pipeline unproven.
  4. [§4.3, Eq. (1)] The centerline refinement (§4.3, Eq. 1 plus the three additional regularizers on adjacent radial distances, displacement standard deviation and curvature) is presented as a novel, clean disentanglement. The relative weights of those regularizers, the precise extension length used near root/tip, and any sensitivity analysis are omitted. Because the subsequent canonical-space generation and style-guide resampling both depend on the quality of these centerlines, the free parameters of the optimization are load-bearing and should be stated and justified.
minor comments (5)
  1. [§3.1, §5.2] Porosity and thickness are listed among the five texture parameters yet never appear in the labeling equations or the diffusion conditioning; a short clarifying sentence in §3 or §5 would avoid reader confusion.
  2. [Fig. 2, §3.1.3] Figure 2’s qualitative twist-prevalence scale is useful, but the mapping from visual category to the continuous Tf value used by the network is not stated; a brief interpolation rule would help reproducibility.
  3. [§5.2, §6.2] Hyper-parameters of both labeling networks (LSTM/attention dimensions, transformer heads, learning rates, dropout) and of the diffusion U-Net (exact channel schedule, FiLM injection details) are only partially listed; a short table or appendix entry would aid re-implementation.
  4. [Appendix A, Figs. 12–16] Several figures (e.g., Figs. 12–13) rely on Houdini post-processing for render hairs; the appendix description is helpful but the exact clump/frizz settings used for each result row would improve visual reproducibility.
  5. [passim] Minor typographical issues: 'prevalance' vs. 'prevalence', 'oscilations', 'repitition', 'indivuduals', 'horizantally', 'neccesary', etc., appear throughout; a careful proof-read would clean them.

Circularity Check

0 steps flagged

No load-bearing circularity: the 5D texture space is proposed (not derived as a forced bijection), centerlines are optimized independently, and diffusion is supervised by a separately trained labeler rather than equating outputs to inputs by construction.

full rationale

The paper's central pipeline (parameter space + canonical centerlines + diffusion supervised by a texture labeler) does not reduce any claimed prediction or first-principles result to its own inputs by definition or by a self-citation uniqueness theorem. The five parameters are introduced as an expert-informed proposal that is 'intended to be a bijection' (Abstract, §3); they are not obtained by solving an equation whose solution is forced to equal the data. Centerline extraction bootstraps a prior DCT method then minimizes a variance-of-radii objective plus regularizers (§4.3, Eq. 1); this is a computational procedure, not a tautology. Procedural helices supply known ground-truth labels for the bulk of the labeling-network training (§5.2), after which the network is fine-tuned on a modest hand-labeled simulated set; the diffusion model is then guided by that network (or over-fit to a few target strands) via posterior sampling (§6.4). Because the labeler is an approximate neural map rather than an algebraic identity, generated strands are not forced to recover the target parameters by construction. Self-citations (e.g., simulation [53], initial centerlines [25]) are ordinary methodological references and do not underwrite a uniqueness claim that forbids alternatives. The absence of quantitative recovery metrics for the labeler after simulation is a validation gap, not circularity. Score 1 reflects only the minor, non-load-bearing self-reference inherent in any multi-stage neural pipeline that re-uses its own trained components.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 3 invented entities

The central claim rests on a new parameterization treated as a bijection, a geometric disentanglement procedure, and standard ML training assumptions. Free parameters are the usual network hyper-parameters and the hand-chosen regularization weights in the centerline objective; axioms include classical differential geometry and the physical premise that internal protein bonds produce the observed texture; invented entities are the five-parameter texture space itself and the refined style-guide/centerline objects.

free parameters (4)
  • centerline optimization regularizer weights (radial variance, adjacent-radius difference, curvature, displacement std)
    Equation (1) plus the three regularizers are minimized; relative weights are not derived from first principles and affect how cleanly texture is separated from geometry.
  • network architecture sizes and training hyper-parameters (LSTM/attention dims, U-Net channels 64-256, learning rates, dr
    Chosen by experimentation; performance of both the labeler and the diffusion model depends on them.
  • number and selection of hand-labeled simulated strands (~1k) and artist-created grooms
    Used for fine-tuning the labeler and for data curation; selection criteria are qualitative.
  • style-guide density resampling factor (square of radius ratio)
    Heuristic circle-packing argument; not derived from collision or volume constraints.
axioms (4)
  • standard math Frenet-Serret frame and curvature-torsion-speed representation uniquely determine a space curve up to rigid motion (classical differential geometry).
    Used throughout Sections 5.1 and 6.5 for both labeling and reconstruction.
  • domain assumption Internal biochemical forces (protein bonds, lipid formations) produce persistent strand texture that is independent of external mechanical style and can be summarized by five scalar parameters.
    Stated in the introduction and Section 3; underpins the entire disentanglement claim.
  • domain assumption A Cosserat-rod simulation of wet hair followed by rest-state reset produces post-drying geometry whose texture parameters remain meaningful and labelable.
    Section 4.2; justified by references to hair science but not re-validated here.
  • ad hoc to paper The five-parameter space is (approximately) bijective with naturally occurring textures.
    Explicitly claimed in the abstract and Section 3; no formal proof or exhaustive sampling is supplied.
invented entities (3)
  • five-dimensional strand-texture parameter space (thickness, curl radius, curl wavelength, twist prevalence, porosity) no independent evidence
    purpose: Provide a user-accessible and machine-usable description that is claimed to cover natural hair textures.
    Synthesized from scientific and cultural sources; treated as the fundamental latent variables for generation and editing.
  • refined style-guide / centerline obtained by post-DCT radial-variance optimization in canonical space no independent evidence
    purpose: Disentangle overall strand direction from intrinsic texture so that generation can occur in a canonical frame.
    Novel algorithmic object; success is judged visually by absence of low-frequency oscillations.
  • texture-labeling neural network (LSTM+attention for twist prevalence + transformer for characteristic torsion) no independent evidence
    purpose: Supply differentiable supervision signals to the diffusion model without manual labeling at every step.
    Trained first on unsimulated data then fine-tuned; accuracy is not quantified beyond qualitative examples.

pith-pipeline@v1.1.0-grok45 · 33356 in / 3478 out tokens · 38150 ms · 2026-07-14T15:22:41.038581+00:00 · methodology

0 comments
read the original abstract

We present a framework for understanding and generating feature rich hair strands. Drawing upon both scientific and cultural expertise, we define strand texture as the various distinctive patterns (curling, switchbacks, twist, etc.) that are formed by forces internal to a hair strand. We begin by proposing a novel five-dimensional parameter space, intended to be a bijection with naturally occurring hair strand textures. This encoding is both qualitatively accessible, allowing users to readily locate their own hair in the parameter space, and quantitatively precise, allowing the generation of individual strands from texture inputs. Importantly, strand texture should be independent from the overall strand direction. In order to disentangle strand texture from the overall strand direction, we identify centerline geometry and use it to map strands into a canonical space (a strand texture space). We construct centerlines using a novel method that cleanly distills complex hair grooms, separating the strand texture from the overall style (parameterized by style guides). We enable the creation of new strands conforming to our parametric description of texture via a generative artificial intelligence approach supervised by a separate neural network trained to label candidate strands according to our five-parameter description. The ability to create new strands conforming to any desired texture enables groom editing using either texture transfer or user-provided inputs. We demonstrate results on a variety of hair types.

Figures

Figures reproduced from arXiv: 2607.09811 by Carolyn Smith, Ronald Fedkiw, Sarah Jobalia, Yitong Deng.

Figure 1
Figure 1. Figure 1: Using our pipeline, we transform an input groom (center) to reflect various target hair strand textures while maintaining the overall style. The four [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: We offer a series of labeled visualizations to help users identify twist prevalence. After identifying hair type from these visualizations and accompanying descriptions, 𝑇𝑓 can be ascertained. A twist prevalence of 0 indicates that strands are helical with smoothly varying curvature, differing from each other only slightly. Low-medium twist prevalence strands (𝑇𝑓 = .25) exhibit rare deviations from the hel… view at source ↗
Figure 4
Figure 4. Figure 4: Three hair strands depicted both before (green) and after (red) physics-based simulation. The top end of the strand is fixed/constrained, while the rest of the rest of the strand responds to a downward gravitational force. (a) The increased mass acting upon the upper portion of the strand causes decreased radius and increased wavelenth. (b) When stretched, twist angles cause the helix to appear to be missi… view at source ↗
Figure 5
Figure 5. Figure 5: Top left: The DCT-based method of [25] is used to bootstrap a rough approxi￾mation to the centerline (shown in blue) of a discretized strand S. Bottom left: After mapping to the canonical space, the errors in centerline prediction become more obvi￾ous (note the low frequency oscilations in the red strand). Bottom right: We improve the centerline approximation to better disentangle texture from geometry (no… view at source ↗
Figure 6
Figure 6. Figure 6: Inferencing strand texture parameters: To create a dataset for training, we construct strands following Section 4. Then, the process described in Section 5.1 is used to compute curvature, torsion, and speed. A network is trained to inference 𝑇˜ 𝑓 directly from the computed values of curvature, torsion, and speed. A second network inferences 𝜏˜ from 𝑇˜ 𝑓 and the computed values of curvature and torsion. 𝜅˜ … view at source ↗
Figure 7
Figure 7. Figure 7: Texture editing pipeline. The new texture can be specified either via a small set of target strands or via texture parameters (blue and red arrows illustrate the only differences between the two approaches). After extracting style guides from the input groom (top left), the radii of the input and target textures are used to determine the amount of style guide resampling necessary to adequately support the … view at source ↗
Figure 9
Figure 9. Figure 9: Modifying the texture of an input groom (bottom left) without changing the density of the style guides (bottom middle) can lead to unrealistic results (bottom right). In this specific case where the texture radius decreases, denser style guides (top middle) are required to obtain more realistic results (top right). 7.1 Style guide resampling It is typically necessary to adjust the style guide density in or… view at source ↗
Figure 8
Figure 8. Figure 8: The diffusion model is able to generate a diverse set of texture-consistent strands (red) even when it is trained on only two input strands (green). Note how various features (twisted bundles of curls, sections that jut to the left and right, and a hook-like curl at the root) of the input strands manifest differently in each generated strand while still preserving the overall look and feel of the texture. … view at source ↗
Figure 10
Figure 10. Figure 10: This figure illustrates the efficacy of the guidance mechanism discussed in Section 7.2, when it is used on the pre-trained model from Section 6.3. For each parameter, we increase the specified target value (from left to right) and show the resulting inferenced strand along with its labeled (via Section 5.2) parameter value. 7.3 Adjusting Strand Length Although length is supervised, the lengths of generat… view at source ↗
Figure 12
Figure 12. Figure 12: In order to demonstrate disentanglement between style and texture, we illustrate the results obtained by using our method on four disparate styles (left) and four disparate textures (top). See [PITH_FULL_IMAGE:figures/full_fig_p012_12.png] view at source ↗
Figure 11
Figure 11. Figure 11: Strands of different textures (green) next to their elongated versions (red), after scaling the speed channel by a factor of 2. 7.4 Treating Interpenetrations After the inferenced strands are lengthened/shortened (if necces￾sary), they are mapped to the world space style guides following Section 4.3. Even though interpolated style guides should not inter￾penetrate the head, ears, or neck (and are deleted … view at source ↗
Figure 15
Figure 15. Figure 15: Same as [PITH_FULL_IMAGE:figures/full_fig_p013_15.png] view at source ↗
Figure 14
Figure 14. Figure 14: We struggled to generate curly and afro-textured hair using the methods proposed in [125, 166]; However, we were able to improve upon the results by using our method to disentangle geometry and texture and subsequently swap in a better texture. Note that we did not change their generated style (i.e. the centerline geometry), which can also be problematic. See [PITH_FULL_IMAGE:figures/full_fig_p013_14.png] view at source ↗
Figure 16
Figure 16. Figure 16: Leftmost two columns: input groom and extracted style guides. Middle two columns: outputs from our texture editing pipeline. Rightmost two columns: final results, using artist created render hairs (from the guide strands in the middle two columns). The diverse results illustrated in the last four rows (last two columns) emphasize the impact of texture variations, especially when noting that all four rows … view at source ↗
Figure 17
Figure 17. Figure 17: The parameter entries in the Guide Groom node (left) and the Hair Generate node (center and right) represent the entirety of settings we change from the defaults when generating render hairs. Specifically, in the Guide Groom node under “Guide Creation”, we change the Mode and Geometry, select Snap to Skin, and deselect Resam￾ple To Match Fur Segments . In the Hair Generate node, in the “General” tab (cent… view at source ↗
Figure 18
Figure 18. Figure 18: With the simple setup outlined in [PITH_FULL_IMAGE:figures/full_fig_p018_18.png] 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

171 extracted references · 44 canonical work pages · 17 internal anchors

  1. [1]

    Yuval Alaluf, Or Patashnik, and Daniel Cohen-Or. 2021. ReStyle: A Residual- Based StyleGAN Encoder via Iterative Refinement. arXiv:2104.02699 [cs] doi:10. 48550/arXiv.2104.02699

    Pith/arXiv arXiv 2021

  2. [2]

    Katlyn Sofaea Alo Alapati. 2022. Hair: Why Representation and Beauty Culture Matter. https://www.sfchronicle.com/projects/2022/hair/representation/

    2022

  3. [3]

    allure.com. 2021. Matrix Introduces Hair Diversity Matrix to Ensure All Its Products Work Across All Hair Types

    2021

  4. [4]

    Dmitry Ammosov, Tina Mai, and Juan Galvis. 2024. Generalized Multiscale Finite Element Method for a Nonlinear Elastic Strain-Limiting Cosserat Model. J. Comput. Phys.519 (Dec. 2024), 113428. doi:10.1016/j.jcp.2024.113428

    work page doi:10.1016/j.jcp.2024.113428 2024

  5. [6]

    Ken-ichi Anjyo, Yoshiaki Usami, and Tsuneya Kurihara. 1992. A Simple Method for Extracting the Natural Beauty of Hair.ACM SIGGRAPH Computer Graphics 26, 2 (July 1992), 111–120. doi:10.1145/142920.134021

    work page doi:10.1145/142920.134021 1992

  6. [7]

    Autodesk. 2026. Maya. Autodesk, INC

    2026

  7. [8]

    Lindelöf B, Forslind B, Hedblad Ma, and Kaveus U. 1988. Human Hair Form. Morphology Revealed by Light and Scanning Electron Microscopy and Computer Aided Three-Dimensional Reconstruction.Archives of dermatology124, 9 (Sept. 1988)

    1988

  8. [9]

    Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2016. Neural Machine Translation by Jointly Learning to Align and Translate. arXiv:1409.0473 [cs] doi:10.48550/arXiv.1409.0473

    Pith/arXiv arXiv doi:10.48550/arxiv.1409.0473 2016

  9. [10]

    Miklós Bergou, Max Wardetzky, Stephen Robinson, et al. 2008. Discrete Elastic Rods. InACM SIGGRAPH 2008 Papers (SIGGRAPH ’08). Association for Comput- ing Machinery, New York, NY, USA, 1–12. doi:10.1145/1399504.1360662

    doi:10.1145/1399504.1360662 2008

  10. [11]

    Bruno A Bernard. 2017. The Hair Follicle Enigma.Experimental dermatology26, 6 (2017), 472–477

    2017

  11. [12]

    Maurice Bernstein and Sylvan Robertson. 1927. Racial and Sexual Differences in Hair Weight.American Journal of Physical Anthropology10 (1927), 379–385

    1927

  12. [13]

    Florence Bertails, Basile Audoly, Marie-Paule Cani, et al. 2006. Super-Helices for Predicting the Dynamics of Natural Hair.ACM Transactions on Graphics25, 3 (July 2006), 1180–1187. doi:10.1145/1141911.1142012

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1145/1141911.1142012 2006

  13. [14]

    Richard L. Bishop. 1975. There Is More than One Way to Frame a Curve.The American Mathematical Monthly82, 3 (1975), 246–251. jstor:2319846 doi:10. 2307/2319846

    1975

  14. [15]

    Blender Online Community. 2026. Blender. Blender Foundation

    2026

  15. [16]

    2008.Hair Growth and Disorders

    Ulrike Blume-Peytavi. 2008.Hair Growth and Disorders. Springer, Berlin

    2008

  16. [17]

    Breakspear, B

    S. Breakspear, B. Nöcker, and C. Popescu. 2024. Chemical Bonds and Hair Behaviour- A Review.International Journal of Cosmetic Science46, 5 (2024), 806–814. doi:10.1111/ics.12967

    work page doi:10.1111/ics.12967 2024

  17. [18]

    Hans Brunner and Brian J. Coman. 1974.The Identification of Mammalian Hair. Inkata Press

    1974

  18. [19]

    Edwin Catmull and Raphael Rom. 1974. A CLASS OF LOCAL INTERPOLATING SPLINES. InComputer Aided Geometric Design. Academic Press, 317–326. doi:10. 1016/B978-0-12-079050-0.50020-5

    1974

  19. [20]

    Seunggyu Chang, Gihoon Kim, and Hayeon Kim. 2023. HairNeRF: Geometry- Aware Image Synthesis for Hairstyle Transfer. InProceedings of the IEEE/CVF International Conference on Computer Vision. 2448–2458

    2023

  20. [21]

    Wesley Chang, Andrew Russell, Stephane Grabli, et al . 2025. Transforming Unstructured Hair Strands into Procedural Hair Grooms.ACM Trans. Graph.44, August 2025 (May 2025). doi:10.1145/3731168

    doi:10.1145/3731168 2025

  21. [22]

    Raphaël Charrondière, Sébastien Neukirch, and Florence Bertails-Descoubes

  22. [23]

    Graph.43, 5 (Aug

    MERCI: Mixed Curvature-Based Elements for Computing Equilibria of Thin Elastic Ribbons.ACM Trans. Graph.43, 5 (Aug. 2024), 160:1–160:26. doi:10. 1145/3674502

    2024

  23. [24]

    Huanyu Chen, Jiahao Wen, and Jernej Barbič. 2025. ANIME-Rod: Adjustable Nonlinear Isotropic Materials for Elastic Rods.ACM Trans. Graph.44, 4 (July 2025), 87:1–87:23. doi:10.1145/3731208

    work page doi:10.1145/3731208 2025

  24. [25]

    Ting Chen. 2023. On the Importance of Noise Scheduling for Diffusion Models. arXiv:2301.10972 [cs] doi:10.48550/arXiv.2301.10972

    Pith/arXiv arXiv doi:10.48550/arxiv.2301.10972 2023

  25. [26]

    Yunlu Chen, Francisco Vicente Carrasco, Christian Häne, et al. 2024. Doubly Hierarchical Geometric Representations for Strand-based Human Hairstyle Generation. InAdvances in Neural Information Processing Systems, A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang (Eds.), Vol. 37. Curran Associates, Inc., Pittsburgh, PA, USA,...

    2024

  26. [27]

    Yufan Chen, Lizhen Wang, Qijing Li, et al. 2024. Monogaussianavatar: Monocular Gaussian Point-Based Head Avatar. InACM SIGGRAPH 2024 Conference Papers. 1–9

    2024

  27. [28]

    Mccann, et al

    Hyungjin Chung, Jeongsol Kim, Michael T. Mccann, et al . 2024. Diffusion Posterior Sampling for General Noisy Inverse Problems. arXiv:2209.14687 [stat] doi:10.48550/arXiv.2209.14687

    Pith/arXiv arXiv doi:10.48550/arxiv.2209.14687 2024

  28. [29]

    Khumalo, Jennifer C

    Elsabe Cloete, Nonhlanhla P. Khumalo, Jennifer C. Van Wyk, and Malebogo N. Ngoepe. 2019. Systems Approach to Human Hair Fibers: Interdependence Between Physical, Mechanical, Biochemical and Geometric Properties of Natural Healthy Hair.Frontiers in PhysiologyVolume 10 - 2019 (2019). doi:10.3389/fphys. 2019.00112

    work page doi:10.3389/fphys 2019

  29. [30]

    David Csuka, Ella Csuka, Margit Juhász, et al. 2022. A Systematic Review on the Lipid Composition of Human Hair.International Journal of Dermatology62 (Feb. 2022). doi:10.1111/ijd.16109

    work page doi:10.1111/ijd.16109 2022

  30. [31]

    Clever Curl. 2020. HOW TO DETERMINE HAIR POROSITY AND WHAT IT MEANS FOR YOUR HAIR

    2020

  31. [32]

    Liliane-Joy Dandy, Michele Vidulis, Yingying Ren, and Mark Pauly. 2024. Tencers: Tension-Constrained Elastic Rods.ACM Transactions on Graphics43, 6 (Dec. 2024), 1–13. doi:10.1145/3687967

    work page doi:10.1145/3687967 2024

  32. [33]

    Westgate

    Gabriela Daniels, Ashiana Fraser, and Gillian E. Westgate. 2023. How Different Is Human Hair? A Critical Appraisal of the Reported Differences in Global Hair Fibre Characteristics and Properties towards Defining a More Relevant Framework for Hair Type Classification.International Journal of Cosmetic Science 45, 1 (2023), 50–61. doi:10.1111/ics.12819

    work page doi:10.1111/ics.12819 2023

  33. [34]

    Gilles Daviet. 2023. Interactive Hair Simulation on the GPU Using ADMM. In Special Interest Group on Computer Graphics and Interactive Techniques Conference Conference Proceedings. ACM, Los Angeles CA USA, 1–11. doi:10.1145/3588432. 3591551

    doi:10.1145/3588432 2023

  34. [35]

    Yu Deng, Duomin Wang, Xiaohang Ren, et al. 2024. Portrait4D: Learning One- Shot 4D Head Avatar Synthesis Using Synthetic Data. arXiv:2311.18729 [cs] doi:10.48550/arXiv.2311.18729

    Pith/arXiv arXiv doi:10.48550/arxiv.2311.18729 2024

  35. [36]

    Helisa Dhamo, Yinyu Nie, Arthur Moreau, et al . 2024. HeadGaS: Real-Time Animatable Head Avatars via 3D Gaussian Splatting. arXiv:2312.02902 [cs] doi:10.48550/arXiv.2312.02902

    Pith/arXiv arXiv doi:10.48550/arxiv.2312.02902 2024

  36. [38]

    Yun (Raymond) Fei, Christopher Batty, Eitan Grinspun, and Changxi Zheng. 2019. A Multi-Scale Model for Coupling Strands with Shear-Dependent Liquid.ACM Transactions on Graphics38, 6 (Dec. 2019), 1–20. doi:10.1145/3355089.3356532

    doi:10.1145/3355089.3356532 2019

  37. [39]

    1850.Sur La Théorie Des Courbes à Double Courbure

    Jean Frédéric Frenet. 1850.Sur La Théorie Des Courbes à Double Courbure. Ph. D. Dissertation. Université de Toulouse

  38. [40]

    Xiao Fu, Wei Yin, Mu Hu, et al . 2024. GeoWizard: Unleashing the Diffusion Priors for 3D Geometry Estimation from a Single Image. arXiv:2403.12013 [cs] doi:10.48550/arXiv.2403.12013

    Pith/arXiv arXiv doi:10.48550/arxiv.2403.12013 2024

  39. [41]

    2021.Representations of Black Women in Hollywood

    Geena Davis Institute on Gender in Media. 2021.Representations of Black Women in Hollywood. Technical Report. Mount Saint Mary’s University,

    2021

  40. [42]

    Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, et al

    Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, et al . 2014. Generative Adversarial Networks. arXiv:1406.2661 [stat] doi:10.48550/arXiv.1406.2661

    Pith/arXiv arXiv doi:10.48550/arxiv.1406.2661 2014

  41. [43]

    2000.Human Hair Diversity

    John Gray. 2000.Human Hair Diversity. Wiley-Blackwell, Abingdon, Oxon

    2000

  42. [44]

    A Guttmann. 2024. Media Usage in the United States

    2024

  43. [45]

    Duane P Harland, Richard J Walls, James A Vernon, et al . 2014. Three- Dimensional Architecture of Macrofibrils in the Human Scalp Hair Cortex. Journal of structural biology185, 3 (2014), 397–404

    2014

  44. [46]

    2022.A Cultural History of Hair in Antiquity

    Mary Harlow, Geraldine Biddle-Perry, Edith Snook, et al . 2022.A Cultural History of Hair in Antiquity. Bloomsbury Academic

    2022

  45. [47]

    Chengan He, Junxuan Li, Artem Sevastopolsky, et al . 2025. 3DGH: 3D Head Generation with Composable Hair and Face. arXiv:2506.20875 [cs] doi:10.48550/ arXiv.2506.20875

    Pith/arXiv arXiv 2025

  46. [48]

    Chengan He, Xin Sun, Zhixin Shu, et al. 2025. Perm: A Parametric Representation for Multi-Style 3D Hair Modeling. arXiv:2407.19451 [cs] doi:10.48550/arXiv.2407. 19451

    Pith/arXiv arXiv doi:10.48550/arxiv.2407 2025

  47. [49]

    Jorge Alejandro Amador Herrera, Yi Zhou, Xin Sun, et al . 2024. Augmented Mass-Spring Model for Real-Time Dense Hair Simulation. arXiv:2412.17144 [cs] doi:10.48550/arXiv.2412.17144

    work page doi:10.48550/arxiv.2412.17144 2024

  48. [50]

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. 2020. Denoising Diffusion Probabilis- tic Models. arXiv:2006.11239 [cs] doi:10.48550/arXiv.2006.11239

    Pith/arXiv arXiv doi:10.48550/arxiv.2006.11239 2020

  49. [51]

    Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long Short-Term Memory. Neural Computation9 (Nov. 1997), 1735–1780. doi:10.1162/neco.1997.9.8.1735

    doi:10.1162/neco.1997.9.8.1735 1997

  50. [52]

    Jerry Hsu, Tongtong Wang, Zherong Pan, et al. 2023. Sag-Free Initialization for Strand-Based Hybrid Hair Simulation.ACM Transactions on Graphics42, 4 (Aug. 2023), 1–14. doi:10.1145/3592143

    doi:10.1145/3592143 2023

  51. [53]

    Jerry Hsu, Tongtong Wang, Zherong Pan, et al . 2024. Real-Time Physically Guided Hair Interpolation.ACM Transactions on Graphics43, 4 (July 2024), 1–11. doi:10.1145/3658176

    doi:10.1145/3658176 2024

  52. [54]

    Jerry Hsu, Tongtong Wang, Kui Wu, and Cem Yuksel. 2025. Stable Cosserat Rods. InProceedings of the Special Interest Group on Computer Graphics and Interactive Techniques Conference Conference Papers. ACM, Vancouver BC Canada, 1–10. doi:10.1145/3721238.3730618

    doi:10.1145/3721238.3730618 2025

  53. [55]

    Liwen Hu, Chongyang Ma, Linjie Luo, and Hao Li. 2015. Single-View Hair Mod- eling Using A Hairstyle Database.ACM Transactions on Graphics (Proceedings SIGGRAPH 2015)34, 4 (July 2015)

    2015

  54. [56]

    Liangxiao Hu, Hongwen Zhang, Yuxiang Zhang, et al. 2024. GaussianAvatar: Towards Realistic Human Avatar Modeling from a Single Video via Animatable 16•Sarah Jobalia, Yitong Deng, Carolyn Smith, and Ronald Fedkiw 3D Gaussians

    2024

  55. [57]

    Xinrong Hu, Qing Chang, Junjie Huang, et al . 2023. HSSAN: Hair Synthesis with Style-Guided Spatially Adaptive Normalization on Generative Adversarial Network.The Visual Computer39, 8 (Aug. 2023), 3311–3318. doi:10.1007/s00371- 023-02998-5

    doi:10.1007/s00371- 2023

  56. [58]

    Li Huang, Fan Yang, Chendi Wei, et al. 2023. Towards Realtime: A Hybrid Physics- based Method for Hair Animation on GPU.Proceedings of the ACM on Computer Graphics and Interactive Techniques6, 3 (Aug. 2023), 1–18. doi:10.1145/3606937

    work page doi:10.1145/3606937 2023

  57. [59]

    2024.Natural: Black Beauty and the Politics of Hair

    Chelsea Mary Elise Johnson. 2024.Natural: Black Beauty and the Politics of Hair. NYU Press. jstor:jj.28248903

    2024

  58. [60]

    Tero Karras, Miika Aittala, Timo Aila, and Samuli Laine. 2022. Elucidating the Design Space of Diffusion-Based Generative Models. arXiv:2206.00364 [cs] doi:10.48550/arXiv.2206.00364

    Pith/arXiv arXiv doi:10.48550/arxiv.2206.00364 2022

  59. [61]

    Tero Karras, Samuli Laine, and Timo Aila. 2019. A Style-Based Generator Architecture for Generative Adversarial Networks. arXiv:1812.04948 [cs] doi:10. 48550/arXiv.1812.04948

    Pith/arXiv arXiv 2019

  60. [62]

    Vaibhav Kaushik, Ajeet Kumar, Nitya Nand Gosvami, et al . 2022. Benefit of Coconut-Based Hair Oil via Hair Porosity Quantification.International Journal of Cosmetic Science44, 3 (2022), 289–298. doi:10.1111/ics.12774

    work page doi:10.1111/ics.12774 2022

  61. [63]

    Bernhard Kerbl, Georgios Kopanas, Thomas LeimkÃŒhler, and George Drettakis

  62. [64]

    3D Gaussian Splatting for Real-Time Radiance Field Rendering

  63. [66]

    Sasikarn Khwanmuang, Pakkapon Phongthawee, Patsorn Sangkloy, and Supa- sorn Suwajanakorn. 2023. StyleGAN Salon: Multi-View Latent Optimization for Pose-Invariant Hairstyle Transfer. InIEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    2023

  64. [67]

    Jong-Hyun Kim and Jung Lee. 2025. Porous Models for Enhanced Representation of Saturated Curly Hairs: Simulation and Learning.IEEE Access13 (2025), 84517– 84540. doi:10.1109/ACCESS.2025.3569324

    work page doi:10.1109/access.2025.3569324 2025

  65. [68]

    Ted Kim. 2020. The Racist Legacy of Computer-Generated Humans

    2020

  66. [69]

    Taewoo Kim, Chaeyeon Chung, Yoonseo Kim, et al. 2022. Style Your Hair: Latent Optimization for Pose-Invariant Hairstyle Transfer via Local-Style-Aware Hair Alignment. arXiv:2208.07765 [cs] doi:10.48550/arXiv.2208.07765

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2208.07765 2022

  67. [70]

    1876.Vorlesungen Über Mathematische Physik: Mechanik(1 ed.)

    Gustav Robert Kirchhoff. 1876.Vorlesungen Über Mathematische Physik: Mechanik(1 ed.). B. G. Teubner, Leipzig

  68. [71]

    Koch, Mark D

    Sandra L. Koch, Mark D. Shriver, and Nina G. Jablonski. 2019. Variation in Human Hair Ultrastructure among Three Biogeographic Populations.Journal of Structural Biology205, 1 (Jan. 2019), 60–66. doi:10.1016/j.jsb.2018.11.008

    work page doi:10.1016/j.jsb.2018.11.008 2019

  69. [72]

    Tina Kubrak. 2020. Impact of Films: Changes in Young People’s Attitudes after Watching a Movie.Behavioral Sciences10, 5 (2020). doi:10.3390/bs10050086

    work page doi:10.3390/bs10050086 2020

  70. [73]

    1901.Beitrag zur näherungsweisen Integration totaler Differen- tialgleichungen

    Wilhelm Kutta. 1901.Beitrag zur näherungsweisen Integration totaler Differen- tialgleichungen. Teubner

    1901

  71. [74]

    Hill Kymberlee, Abimbola Adesola, Dangal Prajjwhal, et al . 2018. Towards Creation of a Curl Pattern Recognition System. InProceedings of the International Conference on Image Processing, Computer Vision, and Pattern Recognition (IPCV). The Steering Committee of The World Congress in Computer Science, Computer . . . , 30–33

    2018

  72. [75]

    Kristy L LaMar and Helen N Rolle. 2022. How Media Influence about Hair Texture Impacts Internalized Racial Oppression and Why The Crown Act Simultaneously Promotes Necessary Change and Yet Familiar Defeat.Journal of Psychology10, 2 (2022), 1–8

    2022

  73. [76]

    Tina Lasisi, Shosuke Ito, Kazumasa Wakamatsu, and Colin N. Shaw. 2016. Quan- tifying Variation in Human Scalp Hair Fiber Shape and Pigmentation.American Journal of Physical Anthropology160, 2 (June 2016), 341–352. doi:10.1002/ajpa. 22971

    work page doi:10.1002/ajpa 2016

  74. [77]

    Minjae Lee, David Hyde, Kevin Li, and Ronald Fedkiw. 2019. A Robust Volume Conserving Method for Character-Water Interaction

    2019

  75. [78]

    Fernández

    Fabien Lespagnol, Céline Grandmont, Paolo Zunino, and Miguel A. Fernández

  76. [79]

    2024), 117316

    A Mixed-Dimensional Formulation for the Simulation of Slender Structures Immersed in an Incompressible Flow.Computer Methods in Applied Mechanics and Engineering432 (Dec. 2024), 117316. doi:10.1016/j.cma.2024.117316

    work page doi:10.1016/j.cma.2024.117316 2024

  77. [80]

    Xueting Li, Ye Yuan, Shalini De Mello, et al. 2024. SimAvatar: Simulation-Ready Avatars with Layered Hair and Clothing. arXiv:2412.09545 [cs] doi:10.48550/ arXiv.2412.09545

    Pith/arXiv arXiv 2024

  78. [81]

    Di Liu, Teng Deng, Giljoo Nam, et al. 2025. LUCAS: Layered Universal Codec Avatars. arXiv:2502.19739 [cs] doi:10.48550/arXiv.2502.19739

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2502.19739 2025

  79. [82]

    Ilya Loshchilov and Frank Hutter. 2019. Decoupled Weight Decay Regularization. arXiv:1711.05101 [cs] doi:10.48550/arXiv.1711.05101

    Pith/arXiv arXiv doi:10.48550/arxiv.1711.05101 2019

  80. [83]

    Loussouarn

    G. Loussouarn. 2001. African Hair Growth Parameters.British Journal of Dermatology145, 2 (Aug. 2001), 294–297. doi:10.1046/j.1365-2133.2001.04350.x

    work page doi:10.1046/j.1365-2133.2001.04350.x 2001

Showing first 80 references.