Pith. sign in

REVIEW 3 major objections 5 minor 49 references

FACSIMILE: Fast and Accurate Scans From an Image in Less Than a Second

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read FACSIMILE claims that a single RGB photo is enough to produce a detailed, textured, poseable 3D body scan in under one second by supervising geometry with per-pixel surface normals instead of depth.

desk verdict FAX's real contribution is showing that a normals-only loss can drive detailed single-image body geometry; the evaluation never measures the detail it claims, but the ablation supports the core design insight. read the letter →

arxiv 1909.00883 v1 pith:OLDMWVFO submitted 2019-09-02 cs.CV

classification cs.CV
keywords single-imagebodyreconstructionsurfacenormalsupervisionimage-to-imagetranslationmonoculardepthestimationalbedoSMPLavatarfittingsynthetictrainingdataDynamicFAUST
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

FACSIMILE (FAX) claims that a single ordinary RGB photograph is enough to produce a detailed, textured, poseable 3D body scan in under a second. The method treats body capture as image-to-image translation: a network maps the photo to front and back depth maps, front and back surface normals, a foreground mask, and front and back albedo, all at the original image resolution. The counterintuitive core claim is that geometry should be supervised primarily with a loss on per-pixel surface normals rather than depth; the authors argue, and support with an ablation, that normals alone recover high-quality depth up to scale, while a direct depth loss yields noisy, unusable surfaces. If true, this closes much of the gap between an easily acquired photo and a detailed avatar, at a fraction of the cost of multi-image or depth-sensor systems. The price is a controlled setup: the method assumes a cooperative subject, frontal arms-down pose, and minimal clothing, and it is trained entirely on synthetic renders.

What carries the argument

The load-bearing object is an image-to-image translation network built from the global generator of [46], with transposed convolutions replaced by upsample-convolutions to avoid checkerboard artifacts, trained with a multi-scale PatchGAN discriminator for albedo and without adversarial losses for geometry. Behind the depth branch sits a fixed spatial-differentiation layer $\delta$: a local difference filter, parameterized by focal length and the predicted mask, that turns the network's depth output into surface normals. Because differentiation is the inverse of integration, supervising the output of $\delta$ with an L1 plus VGG loss on normals implicitly constrains depth without requiring depth ground truth or an explicit integration step. A second contribution is the two-values-per-pixel design: each pixel predicts both the closest and farthest body surface intersecting its ray, yielding front and back depth, normals, and albedo maps that are triangulated into a scan and later completed by SMPL fitting.

What would settle it

Render a large set of synthetic test bodies under the paper's training distribution, measure bidirectional mesh-to-mesh error for the normals-only variant, and compare with a version trained on a direct depth loss; if the depth-supervised version matches or beats normals-only on detail metrics, the paper's central claim that normals dominate depth supervision fails. More decisively, capture real subjects in varied poses and loose clothing with a high-resolution 3D scanner and measure scan error; if real-image errors far exceed the 3 to 9 mm synthetic range, the synthetic-to-real transfer premise is false.

Watch

Extended reading notes

Core claim

The central claim is that high-detail body geometry can be regressed from a single RGB image by predicting per-pixel surface normals, then differentiating them to depth through a fixed layer, rather than predicting depth directly. FAX's geometry network outputs front depth, back depth, and a mask; a fixed spatial-differentiation layer $\delta$ converts the depth channels into front and back normal maps, using the mask to avoid differentiating across boundaries and the focal length to correct perspective. The objective is an L1 loss plus a VGG perceptual loss on the predicted normals, with an optional L1 depth loss for global scale; in the ablation, removing the normal terms hurts more than removing the depth term. After scan inference, a mesh-alignment step fits SMPL (a skinned multi-person linear body model) to the scan with pose and shape priors, closing holes and producing a reposable avatar in about seven seconds. On synthetic renderings of Dynamic FAUST subjects, the single-image estimates reach bidirectional mesh-to-mesh errors around 3 to 9 mm depending on settings, compared with 3.3 to 5.7 mm for a multi-image video-based baseline.

Load-bearing premise

The method is trained only on synthetic images of people in a fixed frontal, arms-down, minimal-clothing setup, and its numerical accuracy is measured on synthetic test renders; if that distribution fails to transfer to real photographs, the claimed single-photo scan capability collapses.

Editorial extensions

If this is right

  • A single forward pass of either network takes about 100 ms; with the 7-second SMPL alignment, an almost complete textured, reposable avatar is produced in under 10 seconds, orders of magnitude faster than detail-preserving multi-image methods.
  • Because geometry is supervised through normals rather than depth, the approach does not need depth ground truth or RGB-D sensors for training, as long as synthetic renders supply normal maps.
  • The front and back two-depth-per-pixel representation recovers both sides of the body from one view, though the seam between front and back scans remains and is closed by model fitting.
  • The ablation identifies the normal L1 plus VGG loss as the most important geometry term: removing it degrades accuracy more than removing the depth loss, and using 2 downsampling scales hurts accuracy.

Reading between the lines

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

  • The normals-over-depth design principle likely transfers to other single-image shape-recovery problems where global scale is ambiguous, such as clothed objects or animals: any domain with renderable synthetic ground-truth normals could train a fast image-to-image scanner.
  • Because the paper's quantitative evidence is synthetic-only, the most direct test of the real-world claim is measuring scan error on real photographs with high-resolution 3D ground truth; until that exists, the single-photo-scan capability on arbitrary real bodies remains unestablished.
  • If scale is truly unrecoverable from normals alone, then downstream applications requiring absolute measurements, such as apparel sizing, would need an external scale cue such as camera distance, a known object in the scene, or a body-height prior.
  • The paper's handling of a missing depth term by independently scaling the back vertices during alignment suggests a testable prediction: without the depth loss, front and back scales drift apart, so the geometry loss alone does not tie the two sides of the body to a common scale.
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 / 5 minor

Summary. This paper introduces FACSIMILE (FAX), a single-image body capture method that uses a conditional image-translation network to predict front/back depth, surface normals, an alpha mask, and front/back albedo at input resolution, followed by an SMPL-based mesh alignment stage that yields a reposable avatar. The key methodological claim is that a per-pixel L1/VGG loss on surface normals, computed from predicted depth through a fixed spatial-differentiation layer, is more effective than a direct loss on depth, and that normals alone can produce plausible geometry. The network is trained on 40,000 synthetic renders of SMPL bodies with in-lab displacement and texture maps under a fixed camera (f=720 pixels), and is evaluated quantitatively on synthetic renderings of D-FAUST subjects using bidirectional point-to-mesh distance after unposing and SMPL fitting, with an ablation study over loss terms, network depth, scales, and blur augmentation, plus qualitative real-image results.

Significance. If the claims are supported, FAX would be a meaningful step for single-image body digitization: it recovers per-pixel geometry at image resolution, runs in about 100 ms for the network and about 7 s for the SMPL alignment, and it makes a crisp, non-obvious design point that surface-normal supervision is more informative than depth supervision for this task. The ablation in Table 2 and the qualitative comparison in Figure 2 support that design insight. The paper is also honest about its controlled assumptions (frontal arms-down pose, minimal clothing, in-lab capture), and the comparison against the multi-image method [3] on D-FAUST is a reasonable sanity check. The main weaknesses are that the quantitative evaluation measures coarse, model-regularized error rather than recovered detail, and that real-image evidence is only qualitative, so the paper's headline claim of 'detail previously unseen' is not yet quantitatively established.

major comments (3)
  1. [Abstract; Eq. (4); Sections 4.1, 4.5] The abstract states that FAX estimates detailed body geometry 'without any depth supervision,' but the training objective in Eq. (4) includes depth losses λd(Lf_d + Lb_d), Section 4.1 says the depth network is trained with depth ground truth, and Section 4.5 explicitly ablates a 'No depth' configuration by removing that term. The method as presented therefore uses depth supervision as part of the full objective; 'normals-only' is an ablation configuration, not the evaluated default. This is a misleading headline claim and should be corrected either by rephrasing the abstract (e.g., by saying the main loss is on normals while depth remains an auxiliary regularizer, or that the method can in principle run without depth supervision) or by reporting the normals-only variant as the primary method.
  2. [Section 4.4; Section 4.5; Tables 1 and 2] The quantitative evaluation does not measure the high-frequency detail that is the paper's stated contribution. In Table 1 and Table 2, the reported bidirectional point-to-mesh error is computed after unposing, rescaling to a fixed height, and fitting/aligning an SMPL model whose pose is initialized from ground truth; Section 4.5 explicitly notes that 'fitting a model to our scan regularizes problems in less robust variants.' This procedure filters out the very displacements (folds, compression artifacts in chest, waist, hips) that Section 1 and Figure 7 claim as the advance. The ablation still provides a valid comparison of loss terms under this coarse metric, but the numbers cannot support the claim of recovered detail. I ask for a quantitative evaluation on the raw scan surfaces before SMPL fitting, for example bidirectional distance between the estimated front/back depth meshes and the ground-truth synthetic scan after a rigid/scale alignment, ideally separated by spatial-frequency band.
  3. [Section 4.2; Section 4.4; Figure 7 caption] The evidence for real-image performance is only qualitative and is collected under conditions very close to the training distribution. Section 4.2 states that test subjects stand in an in-lab environment in an A-pose with tight-fitting clothing, and the Figure 7 caption notes that most subjects wear garments similar to the synthetic training data. The quantitative D-FAUST evaluation is on synthetic renderings made with the same fixed camera and rendering assumptions as training. Consequently, the paper's broader claim that a single photo suffices for a detailed, textured, poseable body scan is not established for arbitrary photographs. Since in-lab 3D scans appear to exist for the displacement maps, a quantitative real-image validation against ground-truth scans would substantially strengthen the verification; without it, the scope of the claims should be explicitly narrowed in the abstract and conclusion.
minor comments (5)
  1. [Table 2] The ablation table reports a single average error per configuration with no variance or statistical significance; given the small differences (for example Baseline 6.89 mm versus 5 res blocks 6.76 mm), error bars or per-subject standard deviations should be reported.
  2. [Section 3.2] The notation Gi_d and Gi_n is used before the index i is defined; please state explicitly that i ∈ {f,b} and give the tensor shapes of all predicted channels.
  3. [Section 4.1] The timing statement is inconsistent: the abstract says 'in less than one second' for a scan, while Section 4.1 reports 100 ms for the network but 7 s for the SMPL alignment; please clarify whether 'scan' refers to the raw depth/albedo output or the fitted avatar.
  4. [Section 4.2] The sampling notation 'x∼[−0.5, 0.5], y∼[0.0, 0.4], z∼[−2.2,−1.5]' does not state the distribution; please specify uniform distributions explicitly.
  5. [Throughout] There are several typos, including 'targetted' in Section 1, 'depthefficiently' in Section 3.2, 'Zollhoffer' in the related work, and 'qualititative' in Section 1; a proofreading pass is recommended.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the normals-loss result is tested against synthetic ground truth and external D-FAUST data; self-citations to SMPL/D-FAUST are benchmarks, not fitted inputs.

full rationale

The claimed derivation chain is empirical and self-contained. FAX is trained on 40,000 synthetic renders with ground-truth depth, normals, and albedo, and the network outputs depth, which is converted to normals by a fixed spatial-differentiation layer (Sections 3.1-3.2, Eqs. 1-4). The claim that a normals loss can recover depth up to scale is justified by the mathematical derivative/integral relation and by the ablation study in Table 2, not by fitting a parameter and then predicting it. The evaluation uses D-FAUST, a public dataset of measured human scans, and compares against the external method [3]; the dataset's author overlap is not circular because the ground-truth meshes are external measurements, not outputs of this paper's fitted values. The mesh-alignment procedure from [6] is a component used to repose and regularize scans, not evidence for the core geometry claim, so the self-citation is minor and not load-bearing. The paper itself concedes verification gaps: quantitative metrics are computed after SMPL fitting, so high-frequency detail is only qualitatively demonstrated, and real-image tests are in-lab near the training distribution (Sections 4.2, 4.5). These are limitations in evidence strength, not circular reductions of a prediction to its input.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on domain assumptions about synthetic-to-real transfer, controlled capture geometry, the validity of normals-as-supervision, and SMPL as a prior. No new physical entities are introduced. Most assumptions are stated in the paper, but none are independently verified outside the controlled test setup.

free parameters (3)
  • Geometry loss weights λd, λn, λmsk, λ_L1, λVGG = Not reported
    The balance of depth, normal, mask, and VGG losses in Eq. (4) is chosen by hand and not stated; the ablation shows the result is sensitive to removing these terms.
  • Albedo loss weights λFM and λVGG = Not reported
    Adopted from [46] with only "weighted as in [46]" given; these weights shape the albedo output.
  • Back-vertex scale variable (opt back) = Optimized per subject
    Used in the no-depth variant to reconcile front/back scale during mesh fitting; it is a fitted value, not learned by the network.
assumptions (5)
  • domain assumption Synthetic training distribution transfers to real images.
    Section 4.2 trains exclusively on 40,000 synthetic renders; Section 4.3 evaluates real images only qualitatively. The method's real-world accuracy depends on this transfer.
  • domain assumption Fixed camera and fixed focal length f=720 pixels.
    Section 3.2 and Section 4.1 assume a fixed focal length; the spatial differentiation layer uses f to correct perspective, so different capture setups would invalidate the normal-to-depth conversion.
  • ad hoc to paper Spatial differentiation layer δ with mask is an adequate normal-to-depth link.
    Section 3.2 introduces the fixed δ layer to convert depth to normals and backpropagate normal loss; this assumes the discrete difference operator with mask is sufficient for training geometry.
  • domain assumption SMPL body model provides an adequate prior for alignment and hole closure.
    Section 3.3 aligns the raw scan to SMPL; errors in SMPL's shape/pose distribution will propagate into the final avatar.
  • domain assumption D-FAUST synthetic renderings and [3]'s bidirectional distance metric measure what matters.
    Section 4.4 uses D-FAUST synthetic renders and the evaluation procedure from [3]; the metric may not capture perceptual detail that the visual results emphasize.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FACSIMILE: Fast and Accurate Scans From an Image in Less Than a Second." pith.science (2026). https://pith.science/paper/OLDMWVFO

@misc{pith2026190900883,
  author       = {Pith},
  title        = {Pith review of: FACSIMILE: Fast and Accurate Scans From an Image in Less Than a Second},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OLDMWVFO}},
  note         = {Machine review of arXiv:1909.00883}
}
read the original abstract

Current methods for body shape estimation either lack detail or require many images. They are usually architecturally complex and computationally expensive. We propose FACSIMILE (FAX), a method that estimates a detailed body from a single photo, lowering the bar for creating virtual representations of humans. Our approach is easy to implement and fast to execute, making it easily deployable. FAX uses an image-translation network which recovers geometry at the original resolution of the image. Counterintuitively, the main loss which drives FAX is on per-pixel surface normals instead of per-pixel depth, making it possible to estimate detailed body geometry without any depth supervision. We evaluate our approach both qualitatively and quantitatively, and compare with a state-of-the-art method.

Figures

Figures reproduced from arXiv: 1909.00883 by the authors.

Figure 1
Figure 1. FAX converts a single RGB image (a) into a scan (b, d) with albedo texture (c, e) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Frontal meshes estimated using (pink) an L1 loss on depth and (green) an L1 loss on normals. loss just on normals can be sufficient to reconstruct a high￾quality depth map up to scale, and that this applies for an articulated, far from spherical object. Because a single depthmap is still far from an entire avatar, we extended the system to estimate front and back￾facing geometry and albedo. Similar to the concurrent… view at source ↗
Figure 3
Figure 3. Depth, surface normals and albedo are computed [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Network architecture for geometry inference (left) and albedo inference (right) from an image a. They share the [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Each row shows two instances of synthetic data (one male, one female). For each example, left to right: RGB, [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Visual comparison of the ablation experiments (a) Baseline, (b) without [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Two columns with RGB image, scan with and without texture and alignment. Pay close attention to variation in [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 46 canonical work pages

  1. [3]

    Video based reconstruc- tion of 3d people models

    Thiemo Alldieck, Marcus Magnor, Weipeng Xu, Christian Theobalt, and Gerard Pons-Moll. Video based reconstruc- tion of 3d people models. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2018. 2, 6, 7

  2. [1]

    Learning to re- construct people in clothing from a single RGB camera

    Thiemo Alldieck, Marcus Magnor, Bharat Lal Bhatnagar, Christian Theobalt, and Gerard Pons-Moll. Learning to re- construct people in clothing from a single RGB camera. In IEEE Conference on Computer Vision and Pattern Recogni- tion (CVPR), jun 2019. 2

  3. [2]

    Detailed human avatars from monocular video

    Thiemo Alldieck, Marcus Magnor, Weipeng Xu, Christian Theobalt, and Gerard Pons-Moll. Detailed human avatars from monocular video. In 2018 International Conference on 3D Vision (3DV), pages 98–109. IEEE, 2018. 1, 2

  4. [4]

    Barron and Jitendra Malik

    Jonathan T. Barron and Jitendra Malik. Shape, illumination, and reflectance from shading. IEEE Trans. Pattern Anal. Mach. Intell, 37(8):1670–1687, 2015. 2

  5. [5]

    A morphable model for the synthesis of 3d faces

    V olker Blanz and Thomas Vetter. A morphable model for the synthesis of 3d faces. In Proceedings of the 26th Annual Conference on Computer Graphics and Interactive Tech- niques, SIGGRAPH ’99, pages 187–194, New York, NY , USA, 1999. ACM Press/Addison-Wesley Publishing Co. 2

  6. [6]

    Black, Matthew Loper, and Javier Romero

    Federica Bogo, Michael J. Black, Matthew Loper, and Javier Romero. Detailed full-body reconstructions of moving peo- ple from monocular rgb-d sequences. In Proceedings of the 2015 IEEE International Conference on Computer Vi- sion (ICCV), ICCV ’15, pages 2300–2308, Washington, DC, USA, 2015. IEEE Computer Society. 1, 4

  7. [7]

    Keep it smpl: Automatic estimation of 3d human pose and shape from a single image

    Federica Bogo, Angjoo Kanazawa, Christoph Lassner, Peter Gehler, Javier Romero, and Michael J Black. Keep it smpl: Automatic estimation of 3d human pose and shape from a single image. In European Conference on Computer Vision, pages 561–578. Springer, 2016. 1, 2, 4

  8. [8]

    Federica Bogo, Javier Romero, Gerard Pons-Moll, and Michael J. Black. Dynamic FAUST: registering human bod- ies in motion. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pages 5573–5582, 2017. 1, 2, 6

Show all 49 references
  1. [9]

    3d face morphable models ”in-the- wild”

    James Booth, Epameinondas Antonakos, Stylianos Ploumpis, George Trigeorgis, Yannis Panagakis, and Stefanos Zafeiriou. 3d face morphable models ”in-the- wild”. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), volume 00, pages 5464–5473, July 2017. 2

  2. [10]

    Cengiz ¨Oztireli, Remo Ziegler, and Markus H

    Endri Dibra, Himanshu Jain, A. Cengiz ¨Oztireli, Remo Ziegler, and Markus H. Gross. Human shape from silhou- ettes using generative hks descriptors and cross-modal neu- ral networks. In CVPR, pages 5504–5514. IEEE Computer Society, 2017. 1, 2

  3. [11]

    Shah, and Ioannis A

    Pengfei Dou, Shishir K. Shah, and Ioannis A. Kakadiaris. End-to-end 3d face reconstruction with deep neural net- works. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 1503–1512, July 2017. 2

  4. [12]

    Predicting depth, surface nor- mals and semantic labels with a common multi-scale convo- lutional architecture

    David Eigen and Rob Fergus. Predicting depth, surface nor- mals and semantic labels with a common multi-scale convo- lutional architecture. CoRR, abs/1411.4734, 2014. 2

  5. [13]

    Deep ordinal regres- sion network for monocular depth estimation

    Huan Fu, Mingming Gong, Chaohui Wang, Kayhan Bat- manghelich, and Dacheng Tao. Deep ordinal regres- sion network for monocular depth estimation. CoRR, abs/1806.02446, 2018. 2

  6. [14]

    Clothed and naked human shapes estimation from a single image

    Yu Guo, Xiaowu Chen, Bin Zhou, and Qinping Zhao. Clothed and naked human shapes estimation from a single image. In CVM, 2012. 2

  7. [15]

    Multilinear pose and body shape estimation of dressed subjects from image sets

    Nils Hasler, Hanno Ackermann, Bodo Rosenhahn, Thorsten Thorm¨ahlen, and Hans-Peter Seidel. Multilinear pose and body shape estimation of dressed subjects from image sets. In The Twenty-Third IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2010, San Francisco, ...

  8. [16]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 3

  9. [17]

    Berthold K. P. Horn. Shape-from-shading: A method for ob- taining the shape of a smooth opaque object from one view. Technical Report MAC-TR-79 and AI-TR-232, AI Labora- tory, MIT, Nov. 1970. 2

  10. [18]

    Berthold K. P. Horn and Michael J. Brooks. The variational approach to shape from shading.Computer Vision, Graphics, and Image Processing, 33(2):174–208, 1986. 1

  11. [19]

    Image-to-image translation with conditional adver- sarial networks

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros. Image-to-image translation with conditional adver- sarial networks. arXiv preprint, 2017. 1, 2, 3

  12. [20]

    Jackson, Adrian Bulat, Vasileios Argyriou, and Georgios Tzimiropoulos

    Aaron S. Jackson, Adrian Bulat, Vasileios Argyriou, and Georgios Tzimiropoulos. Large pose 3d face reconstruction from a single image via direct volumetric cnn regression. In The IEEE International Conference on Computer Vision (ICCV), Oct 2017. 2

  13. [21]

    Moviereshape: Tracking and reshap- ing of humans in videos

    Arjun Jain, Thorsten Thorm ¨ahlen, Hans-Peter Seidel, and Christian Theobalt. Moviereshape: Tracking and reshap- ing of humans in videos. ACM Trans. Graph., 29(6):148:1– 148:10, Dec. 2010. 2

  14. [22]

    End-to-end recovery of human shape and pose

    Angjoo Kanazawa, Michael J Black, David W Jacobs, and Jitendra Malik. End-to-end recovery of human shape and pose. In The IEEE Conference on Computer Vision and Pat- tern Recognition (CVPR), 2018. 1, 2, 6

  15. [23]

    One millisecond face alignment with an ensemble of regression trees

    Vahid Kazemi and Josephine Sullivan. One millisecond face alignment with an ensemble of regression trees. In CVPR, pages 1867–1874. IEEE Computer Society, 2014. 4

  16. [24]

    3d face re- construction from a single image using a single reference face shape

    Ira Kemelmacher-Shlizerman and Ronen Basri. 3d face re- construction from a single image using a single reference face shape. IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, 33(2):394–405, Feb 2011. 2

  17. [25]

    Openimages: A pub- lic dataset for large-scale multi-label and multi-class im- age classification

    Ivan Krasin, Tom Duerig, Neil Alldrin, Andreas Veit, Sami Abu-El-Haija, Serge Belongie, David Cai, Zheyun Feng, Vit- torio Ferrari, Victor Gomes, et al. Openimages: A pub- lic dataset for large-scale multi-label and multi-class im- age classification. Dataset available from htt...

  18. [26]

    Black, and Peter V

    Christoph Lassner, Javier Romero, Martin Kiefel, Federica Bogo, Michael J. Black, and Peter V . Gehler. Unite the peo- ple: Closing the loop between 3d and 2d human representa- tions. In IEEE Conf. on Computer Vision and Pattern Recog- nition (CVPR), July 2017. 2

  19. [27]

    Intrinsic face image decomposition with human face priors

    Chen Li, Kun Zhou, and Stephen Lin. Intrinsic face image decomposition with human face priors. In David J. Fleet, Tom´as Pajdla, Bernt Schiele, and Tinne Tuytelaars, editors, Computer Vision - ECCV 2014 - 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Pro...

  20. [28]

    Capturing relightable human performances under general uncontrolled illumination

    Guannan Li, Chenglei Wu, Carsten Stoll, Yebin Liu, Kiran Varanasi, Qionghai Dai, and Christian Theobalt. Capturing relightable human performances under general uncontrolled illumination. Comput. Graph. Forum, 32(2):275–284, 2013. 1

  21. [29]

    Smpl: A skinned multi- person linear model

    Matthew Loper, Naureen Mahmood, Javier Romero, Gerard Pons-Moll, and Michael J Black. Smpl: A skinned multi- person linear model. ACM Transactions on Graphics (TOG), 34(6):248, 2015. 4

  22. [30]

    Least squares genera- tive adversarial networks

    Xudong Mao, Qing Li, Haoran Xie, Raymond YK Lau, Zhen Wang, and Stephen Paul Smolley. Least squares genera- tive adversarial networks. In Computer Vision (ICCV), 2017 IEEE International Conference on, pages 2813–2821. IEEE,

  23. [31]

    Siclope: Silhouette-based clothed people

    Ryota Natsume, Shunsuke Saito, Zeng Huang, Weikai Chen, Chongyang Ma, Hao Li, and Shigeo Morishima. Siclope: Silhouette-based clothed people. CoRR, abs/1901.00049,

  24. [32]

    De- convolution and checkerboard artifacts

    Augustus Odena, Vincent Dumoulin, and Chris Olah. De- convolution and checkerboard artifacts. Distill, 2016. 3

  25. [33]

    Neural body fitting: Unifying deep learning and model based human pose and shape es- timation

    Mohamed Omran, Christoph Lassner, Gerard Pons-Moll, Pe- ter Gehler, and Bernt Schiele. Neural body fitting: Unifying deep learning and model based human pose and shape es- timation. In 2018 International Conference on 3D Vision (3DV), pages 484–494. IEEE, 2018. 1, 2

  26. [34]

    Learning to estimate 3d human pose and shape from a single color image

    Georgios Pavlakos, Luyang Zhu, Xiaowei Zhou, and Kostas Daniilidis. Learning to estimate 3d human pose and shape from a single color image. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pages 459–468, 2018. 2

  27. [35]

    Learning detailed face reconstruction from a single image

    Elad Richardson, Matan Sela, Roy Or-El, and Ron Kimmel. Learning detailed face reconstruction from a single image. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pages 5553–5562, 2017. 1, 2

  28. [36]

    Unre- stricted facial geometry reconstruction using image-to-image translation

    Matan Sela, Elad Richardson, and Ron Kimmel. Unre- stricted facial geometry reconstruction using image-to-image translation. In Computer Vision (ICCV), 2017 IEEE Interna- tional Conference on, pages 1585–1594. IEEE, 2017. 2, 4

  29. [37]

    Castillo, and David W

    Soumyadip Sengupta, Angjoo Kanazawa, Carlos D. Castillo, and David W. Jacobs. Sfsnet: Learning shape, refectance and illuminance of faces in the wild. In Computer Vision and Pattern Regognition (CVPR), 2018. 2

  30. [38]

    Indirect deep structured learning for 3d human body shape and pose prediction

    Vince Tan, Ignas Budvytis, and Roberto Cipolla. Indirect deep structured learning for 3d human body shape and pose prediction. In British Machine Vision Conference 2017, BMVC 2017, London, UK, September 4-7, 2017, 2017. 2

  31. [39]

    Self-supervised multi-level face model learning for monocular reconstruction at over 250 hz

    Ayush Tewari, Michael Zollh ¨ofer, Pablo Garrido, Florian Bernard, Hyeongwoo Kim, Patrick P ´erez, and Christian Theobalt. Self-supervised multi-level face model learning for monocular reconstruction at over 250 hz. In Proceedings of Computer Vision and Pattern Recognition (CV...

  32. [40]

    MoFA: Model-based Deep Convolutional Face Autoen- coder for Unsupervised Monocular Reconstruction

    Ayush Tewari, Michael Zoll ¨ofer, Hyeongwoo Kim, Pablo Garrido, Florian Bernard, Patrick Perez, and Theobalt Chris- tian. MoFA: Model-based Deep Convolutional Face Autoen- coder for Unsupervised Monocular Reconstruction. In The IEEE International Conference on Computer Vision (ICCV),

  33. [41]

    Anh Tu ˆan Tran, Tal Hassner, Iacopo Masi, and G ´erard G. Medioni. Regressing robust and discriminative 3d mor- phable models with a very deep neural network. In 2017 IEEE Conference on Computer Vision and Pattern Recog- nition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017 ...

  34. [42]

    Extreme 3d face reconstruc- tion: Seeing through occlusions

    Anh Tu ˆan Tran, Tal Hassner, Iacopo Masi, Eran Paz, Yuval Nirkin, and G ´erard Medioni. Extreme 3d face reconstruc- tion: Seeing through occlusions. In Proc. CVPR, 2018. 2

  35. [43]

    Fouhey, Alexei A

    Shubham Tulsiani, Saurabh Gupta, David F. Fouhey, Alexei A. Efros, and Jitendra Malik. Factoring shape, pose, and layout from the 2d image of a 3d scene. CoRR, abs/1712.01812, 2017. 2

  36. [44]

    Bodynet: V ol- umetric inference of 3d human body shapes

    G ¨ul Varol, Duygu Ceylan, Bryan Russell, Jimei Yang, Ersin Yumer, Ivan Laptev, and Cordelia Schmid. Bodynet: V ol- umetric inference of 3d human body shapes. In Computer Vision - ECCV 2018 - 15th European Conference, Munich, Germany, September 8-14, 2018, Proceedings, Part VI...

  37. [45]

    Learning from synthetic humans

    G ¨ul Varol, Javier Romero, Xavier Martin, Naureen Mah- mood, Michael J Black, Ivan Laptev, and Cordelia Schmid. Learning from synthetic humans. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2017), pages 4627–4635. IEEE, 2017. 2

  38. [46]

    High-resolution image synthesis and semantic manipulation with conditional gans

    Ting-Chun Wang, Ming-Yu Liu, Jun-Yan Zhu, Andrew Tao, Jan Kautz, and Bryan Catanzaro. High-resolution image synthesis and semantic manipulation with conditional gans. arXiv preprint arXiv:1711.11585, 2017. 1, 3, 4, 5

  39. [47]

    Quality dynamic human body modeling using a single low-cost depth camera

    Qing Zhang, Bo Fu, Mao Ye, and Ruigang Yang. Quality dynamic human body modeling using a single low-cost depth camera. In CVPR, pages 676–683. IEEE Computer Society,

  40. [48]

    Shape from shading: A survey

    Ruo Zhang, Ping-Sing Tsai, James Edwin Cryer, and Mubarak Shah. Shape from shading: A survey. IEEE Trans. Pattern Anal. Mach. Intell, 21(8):690–706, 1999. 2

  41. [49]

    State of the art on monocular 3d face reconstruction, tracking, and applica- tions

    Michael Zollh ¨ofer, Justus Thies, Pablo Garrido, Thabo Bradley, Derek Beeler, Patrick P ´erez, Marc Stamminger, Matthias Nießner, and Christian Theobalt. State of the art on monocular 3d face reconstruction, tracking, and applica- tions. Computer Graphics Forum, 37(2):523–550...

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.