Pith. sign in

REVIEW 3 major objections 5 minor 52 references

FSGAN: Subject Agnostic Face Swapping and Reenactment

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

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

Pith's one-line read FSGAN establishes a subject-agnostic face swapping and reenactment pipeline: one system transfers pose, expression, and identity between any two faces without per-subject training.

desk verdict Real engineering gains in face reenactment, but a self-contradictory training description undercuts the subject-agnostic claim until clarified. read the letter →

arxiv 1908.05932 v1 pith:UJ5EV6NW submitted 2019-08-16 cs.CV cs.GRcs.LG

Yuval Nirkin , Yosi Keller , Tal Hassner This is my paper · ORCID
classification cs.CVcs.GRcs.LG
keywords faceswappingreenactmentsubjectagnosticgenerativeadversarialnetworksPoissonblendinginpaintingviewinterpolationvideomanipulation
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

This paper introduces FSGAN, a generative adversarial system for face swapping and face reenactment that can be applied to any pair of faces without being trained on them. The central claim is that subject agnosticism is achievable by breaking large pose and expression changes into small recurrent steps, interpolating over a triangulated space of face views, and then completing and blending the result into the target image. The authors further claim that their system preserves target pose and expression better than prior methods while keeping source identity and image quality competitive, and they support this with quantitative pose and landmark errors on held-out videos. If the claim holds, face manipulation becomes dramatically easier for non-experts, since no per-person collection or training is required.

What carries the argument

The recurrent reenactment generator Gr is the load-bearing component: given a face image and a heatmap of target landmarks, it repeatedly transforms the image through interpolated intermediate landmark positions, splitting a large pose or expression change into several smaller steps instead of one risky jump. This stepwise, recurrent design is what lets a network trained only on same-subject frame pairs generalize to new identities. Around it, the system uses Delaunay triangulation and barycentric coordinates to interpolate between multiple source views, a completion generator Gc for occluded regions, and a blending generator Gb trained with a Poisson blending loss that combines the classic Poisson gradient constraint with a perceptual loss.

What would settle it

Run the system on unseen source/target pairs while sweeping the target yaw from frontal to about 90 degrees, and compare the output's face-verification similarity to the source identity; if identity similarity collapses toward the target as the pose gap grows, then the subject-agnostic claim fails precisely in the large-pose regime the paper itself identifies as a limitation.

Watch

Extended reading notes

Core claim

The paper claims that FSGAN is the first method to simultaneously manipulate pose, expression, and identity without person-specific or pair-specific training. The system consists of a recurrent reenactment generator that maps a source face onto the target pose and expression through interpolated intermediate landmarks, a segmentation network, a face inpainting network that fills occluded or missing regions, and a blending network trained with a new Poisson blending loss that combines Poisson optimization with perceptual loss. For videos, the method builds an appearance map from multiple source views using Delaunay triangulation and barycentric coordinates, allowing continuous interpolation of face views rather than relying on a single image. On held-out video pairs, it reports a pose error of 2.49 degrees and a landmark error of 22.2 pixels, lower than the compared baselines.

Load-bearing premise

The reenactment generator is trained only on pairs of frames of the same person, yet it is assumed that what it learns about moving one person's own face transfers unchanged to moving one person's face into another person's head, without any explicit separation of identity from pose and expression.

Editorial extensions

If this is right

  • Face swapping and reenactment become usable on people the system has never seen, removing the expensive per-subject data collection and training required by earlier methods.
  • For video, the appearance-map interpolation means an entire source video can serve as input, giving continuous pose coverage rather than requiring a single carefully chosen source frame.
  • The inpainting and blending stages let the method handle occluded faces and match the transferred face to the target's skin tone and lighting, which the ablation study shows improves pose and expression retention.
  • A single pipeline now covers swapping, reenactment, view interpolation, occlusion handling, and blending, making these tasks substantially more accessible to non-experts.
  • The paper argues that making such synthesis accessible should accelerate the development of technical counter-measures and informed policy for detecting manipulated media.

Reading between the lines

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

  • Because the reenactment generator is trained only on same-subject frame pairs, the system does not explicitly enforce separation of identity from pose and expression; an explicit identity-invariance constraint or adversarial identity loss could further strengthen cross-subject transfer.
  • The recurrent stepwise design suggests a predictable scaling behavior: the larger the pose gap, the more intermediate steps are needed, so one could tune the number of steps automatically based on the angular difference between source and target.
  • The Delaunay/barycentric view-interpolation scheme is not inherently face-specific; the same triangulation of a low-dimensional pose space could be applied to other objects, such as hands or bodies, where continuous pose control is needed.
  • A direct stress test would be extreme poses and unusual expressions: if identity fidelity degrades smoothly as the target deviates from the source's pose, that would confirm the mechanism is genuinely pose-driven rather than identity-driven.
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. The paper introduces FSGAN, a face swapping and reenactment system that is claimed to be subject agnostic, meaning it can operate on pairs of faces without per-subject or per-pair training. The method comprises a recurrent reenactment generator (Gr) that iteratively adjusts a source face to a target pose and expression using landmark heatmaps, a segmentation network (Gs), a face inpainting network (Gc), and a face blending network (Gb) trained with a novel Poisson blending loss. Training uses IJB-C video data, and evaluation is done on FaceForensics++ against DeepFakes, Nirkin et al., and Face2Face. The paper reports quantitative results for identity verification, SSIM, pose error, and landmark error, plus an ablation study and qualitative examples.

Significance. If validated, the subject-agnostic claim is practically significant because it removes the need for expensive per-subject data collection and training, making face swapping and reenactment accessible for a wide range of applications. The modular pipeline, the recurrent stepwise reenactment mechanism for large pose changes, and the proposed Poisson blending loss are useful technical contributions. The paper provides quantitative comparisons and an ablation study on an external benchmark, which is a strength, and it also acknowledges limitations such as degradation under large yaw angles and resolution constraints. However, the central subject-agnostic claim depends on training details that are currently ambiguous, so the contribution is defensible but not yet fully established.

major comments (3)
  1. [Sec. 3.2, Eq. (7); Sec. 4.1] placeholder
  2. [Sec. 3.2, Eq. (6)-(7)] placeholder
  3. [Table 1, Sec. 5.4] placeholder
minor comments (5)
  1. [Sec. 3.2] placeholder
  2. [Eq. (11)] placeholder
  3. [Supplementary Material, Fig. 9 caption] placeholder
  4. [Sec. 3.3] placeholder
  5. [Table 2] placeholder

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FSGAN's subject-agnostic claim is tested on external FaceForensics++ benchmarks, and self-citations are non-load-bearing baselines or component-level resources.

full rationale

FSGAN's central claim of subject-agnostic reenactment and swapping is supported by an end-to-end trainable pipeline and is evaluated on FaceForensics++ videos not seen during training (Sec. 5.2, Fig. 6). The reported quantitative metrics (Table 1) measure pose and expression retention on that external benchmark; lower landmark error follows from conditioning the reenactment generator on target landmark heatmaps, which is a standard functional objective, not a fitted parameter renamed as a prediction. The self-citations to Nirkin et al. [35] appear as a baseline for comparison and as a source of segmentation training data; neither is load-bearing for the subject-agnostic claim. The inconsistency between Sec. 3.2's same-subject training pairs and Sec. 4.1's 'two randomly chosen subjects' is a reproducibility and generalization ambiguity, not a circular step: the model is not defined in terms of its evaluation metric, and no equation reduces to its own input by construction. The limitations stated in Sec. 6 (angular degradation, resolution limits, sparse landmark tracking) are acknowledged weaknesses, not circular reasoning. I find no step where a claimed prediction is equivalent to an input, a fitted parameter, or a self-citation chain.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claim depends on standard supervised-learning assumptions plus the domain-specific transfer assumption that same-subject reenactment training transfers to cross-subject use. No new physical or conceptual entities are postulated; the appearance map is an algorithmic structure.

free parameters (6)
  • perceptual loss weight lambda_perc = 1
    Hand-chosen in Eq. (3); balances VGG perceptual loss against pixel loss in all generators.
  • pixel loss weight lambda_pixel = 0.1
    Hand-chosen in Eq. (3); controls low-frequency reconstruction fidelity.
  • adversarial loss weight lambda_adv = 0.001
    Hand-chosen weight for multi-scale GAN loss in Eqs. (4)-(7).
  • segmentation loss weight lambda_seg = 0.1
    Weight for mask loss in Eq. (7).
  • reenactment steps n
    Number of recursive reenactment iterations in Eq. (6); the paper does not state the value used in the quantitative experiments, only that multiple iterations help (Fig. 5).
  • Adam learning rate = 0.0002
    Optimizer hyperparameter, halved every 10 epochs (Sec 4.2).
assumptions (4)
  • domain assumption dlib facial landmarks and Euler angles capture pose and expression sufficiently for all subjects
    Reenactment conditions on landmark heatmaps H(p) (Eq. 6); landmark failures in extreme expressions would break the pipeline.
  • domain assumption Same-subject training pairs generalize to cross-subject reenactment
    Gr is trained with Is and It from the same video subject (Sec 4.1) and applied to arbitrary identities; no explicit identity-pose disentanglement is enforced.
  • domain assumption IJB-C training videos provide sufficient pose and expression coverage
    The reenactment generator learns pose/expression control from 5,500 high-definition IJB-C videos (Sec 4.1); coverage limits the method's generality.
  • domain assumption VGG perceptual networks pretrained on VGGFace2 and CelebA are suitable face-similarity measures
    The perceptual loss (Eq. 1) uses these networks, so their feature geometry shapes what 'realistic' means in the loss.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FSGAN: Subject Agnostic Face Swapping and Reenactment." pith.science (2026). https://pith.science/paper/UJ5EV6NW

@misc{pith2026190805932,
  author       = {Pith},
  title        = {Pith review of: FSGAN: Subject Agnostic Face Swapping and Reenactment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UJ5EV6NW}},
  note         = {Machine review of arXiv:1908.05932}
}
read the original abstract

We present Face Swapping GAN (FSGAN) for face swapping and reenactment. Unlike previous work, FSGAN is subject agnostic and can be applied to pairs of faces without requiring training on those faces. To this end, we describe a number of technical contributions. We derive a novel recurrent neural network (RNN)-based approach for face reenactment which adjusts for both pose and expression variations and can be applied to a single image or a video sequence. For video sequences, we introduce continuous interpolation of the face views based on reenactment, Delaunay Triangulation, and barycentric coordinates. Occluded face regions are handled by a face completion network. Finally, we use a face blending network for seamless blending of the two faces while preserving target skin color and lighting conditions. This network uses a novel Poisson blending loss which combines Poisson optimization with perceptual loss. We compare our approach to existing state-of-the-art systems and show our results to be both qualitatively and quantitatively superior.

Figures

Figures reproduced from arXiv: 1908.05932 by the authors.

Figure 1
Figure 1. Face swapping and reenactment. Left: Source face swapped onto target. Right: Target video used to control the expressions of the face appearing in the source image. In both cases, our results appears in the middle. For more information please visit our website: https://nirkin.com/fsgan. Abstract We present Face Swapping GAN (FSGAN) for face swap￾ping and reenactment. Unlike previous work, FSGAN is subject agnostic a… view at source ↗
Figure 2
Figure 2. Overview of the proposed FSGAN approach. (a) The recurrent reenactment generator [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Face view interpolation. (a) Shows an example of an appearance map of the source subject (Donald Trump). The green dots represent different views of the source subject, the blue lines represent the Delaunay Triangulation of those views, and the red X marks the location of the current target’s pose. (b) The interpolated views associated with the vertices of the selected triangle (represented by the yellow dots). (c) … view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Qualitative face reenactment results. Row 1: The source face for reenactment. Row 2: Our reenactment results (without background removal). Row 3: The target face from which to transfer the pose and expression [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Reenactment limitations. Top left image trans￾formed onto each of the images in Row 1 (using the same subject for clarity). Row 2: Reenactment with one iteration. Row 3: Three iterations. Euler angles of Fb to the original target image, It. Similarly, the accuracy of t…
Figure 6
Figure 6. Figure 6: Qualitative face swapping results on [39]. Results for source photo swapped onto target provided for Nirkin et al. [35], DeepFakes [12] and our method on images of faces of subjects it was not trained on. Method verification ↓ SSIM ↑ euler ↓ landmarks ↓ FSGAN (Gr) 0.38…
Figure 7
Figure 7. Figure 7: Comparison to Face2Face [44] on FaceForen￾sics++ [39]. As demonstrated, our method exhibits far less artifacts than Face2Face. policy makers if it is misused. References [1] Oleg Alexander, Mike Rogers, William Lambeth, Matt Chi￾ang, and Paul Debevec. Creating a photor…
Figure 8
Figure 8. Figure 8: Ablation study. From columns 3 and 5, without the completion network, Gc, the transferred face does not cover the entire target face, leaving obvious artifacts. Columns 3 and 4 show that without the blending network, Gb, the skin color and lighting conditions of the tr…
Figure 9
Figure 9. Figure 9: Additional qualitative face swapping results on on the Caltech Occluded Faces in the Wild (COFW) dataset [8]. Supplementary Material A. Additional qualitative results We offer additional quantitative face swapping results in [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Additional qualitative face swapping comparison to Nirkin et al. [35] and DeepFakes [12] on FaceForen￾sics++ [39] [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: Additional qualitative face reenactment comparison to Face2Face [44] on FaceForensics++ [39] [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]
Figure 12
Figure 12. Figure 12: Generator architectures. (a) The global generator is based on a residual variant of the U-Net [38] CNN, using a number of bottleneck layers per resolution. We replace the simple convolutions with bottleneck blocks (in purple), the concatenation with summation (plus si…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 38 canonical work pages

  1. [1]

    Creating a photoreal digital actor: The digital emily project

    Oleg Alexander, Mike Rogers, William Lambeth, Matt Chi- ang, and Paul Debevec. Creating a photoreal digital actor: The digital emily project. In Conf. Visual Media Production, pages 176–187. IEEE, 2009

    work page 2009

  2. [2]

    Wasserstein gan

    Martin Arjovsky, Soumith Chintala, and L ´eon Bottou. Wasserstein gan. arXiv preprint arXiv:1701.07875, 2017

    arXiv 2017

  3. [3]

    Bringing portraits to life

    Hadar Averbuch-Elor, Daniel Cohen-Or, Johannes Kopf, and Michael F Cohen. Bringing portraits to life. ACM Transac- tions on Graphics (TOG), 36(6):196, 2017

    work page 2017

  4. [4]

    Face swapping: automatically replacing faces in photographs

    Dmitri Bitouk, Neeraj Kumar, Samreen Dhillon, Peter Bel- humeur, and Shree K Nayar. Face swapping: automatically replacing faces in photographs. ACM Trans. on Graphics , 27(3):39, 2008

    work page 2008

  5. [5]

    Face identification across different poses and illuminations with a 3d morphable model

    V olker Blanz, Sami Romdhani, and Thomas Vetter. Face identification across different poses and illuminations with a 3d morphable model. In Int. Conf. on Automatic Face and Gesture Recognition, pages 192–197, 2002

    work page 2002

  6. [6]

    Exchanging faces in images

    V olker Blanz, Kristina Scherbaum, Thomas Vetter, and Hans-Peter Seidel. Exchanging faces in images. Comput. Graphics F orum, 23(3):669–676, 2004

    work page 2004

  7. [7]

    Face recognition based on fitting a 3d morphable model

    V olker Blanz and Thomas Vetter. Face recognition based on fitting a 3d morphable model. Trans. Pattern Anal. Mach. Intell., 25(9):1063–1074, 2003

    work page 2003

  8. [8]

    Robust face landmark estimation under occlusion

    Xavier P Burgos-Artizzu, Pietro Perona, and Piotr Doll ´ar. Robust face landmark estimation under occlusion. In Proc. Int. Conf. Comput. Vision, pages 1513–1520. IEEE, 2013

    work page 2013

Show all 52 references
  1. [9]

    Vggface2: A dataset for recognising faces across pose and age

    Qiong Cao, Li Shen, Weidi Xie, Omkar M Parkhi, and Andrew Zisserman. Vggface2: A dataset for recognising faces across pose and age. In 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018), pages 67–74. IEEE, 2018

    2018

  2. [10]

    FacePoseNet: Making a case for landmark-free face alignment

    Feng-Ju Chang, Anh Tran, Tal Hassner, Iacopo Masi, Ram Nevatia, and G´erard Medioni. FacePoseNet: Making a case for landmark-free face alignment. In Proc. Int. Conf. Com- put. Vision Workshops, 2017

    2017

  3. [11]

    Deep, landmark-free fame: Face alignment, modeling, and expression estimation

    Feng-Ju Chang, Anh Tuan Tran, Tal Hassner, Iacopo Masi, Ram Nevatia, and G ´erard Medioni. Deep, landmark-free fame: Face alignment, modeling, and expression estimation. Int. J. Comput. Vision, 127(6-7):930–956, 2019

    2019

  4. [12]

    FaceSwap

    DeepFakes. FaceSwap. https://github.com/ deepfakes/faceswap. Accessed: 2019-02-06

    2019

  5. [13]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. In Advances in neural information processing systems, pages 2672–2680, 2014

    2014

  6. [14]

    Effective face frontalization in unconstrained images

    Tal Hassner, Shai Harel, Eran Paz, and Roee Enbar. Effective face frontalization in unconstrained images. In Proc. Conf. Comput. Vision Pattern Recognition, 2015

    2015

  7. [15]

    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

    2016

  8. [16]

    Computer graphics: principles and prac- tice

    John F Hughes, Andries Van Dam, James D Foley, and Steven K Feiner. Computer graphics: principles and prac- tice. Pearson Education, 2014

    2014

  9. [17]

    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. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 1125–1134, 2017

    2017

  10. [18]

    Perceptual losses for real-time style transfer and super-resolution

    Justin Johnson, Alexandre Alahi, and Li Fei-Fei. Perceptual losses for real-time style transfer and super-resolution. In European Conference on Computer Vision , pages 694–711. Springer, 2016

    2016

  11. [19]

    Augmenting crfs with boltzmann machine shape pri- ors for image labeling

    Andrew Kae, Kihyuk Sohn, Honglak Lee, and Erik Learned- Miller. Augmenting crfs with boltzmann machine shape pri- ors for image labeling. In Proceedings of the IEEE con- ference on computer vision and pattern recognition , pages 2019–2026, 2013

    2019

  12. [20]

    Progressive growing of gans for improved quality, stability, and variation

    Tero Karras, Timo Aila, Samuli Laine, and Jaakko Lehtinen. Progressive growing of gans for improved quality, stability, and variation. arXiv preprint arXiv:1710.10196, 2017

    2017 arXiv

  13. [21]

    Transfiguring portraits

    Ira Kemelmacher-Shlizerman. Transfiguring portraits. ACM Trans. on Graphics, 35(4):94, 2016

    2016

  14. [22]

    Figure 8: Ablation study

    Hyeongwoo Kim, Pablo Carrido, Ayush Tewari, Weipeng Xu, Justus Thies, Matthias Niessner, Patrick P ´erez, Chris- tian Richardt, Michael Zollh ¨ofer, and Christian Theobalt. Figure 8: Ablation study. From columns 3 and 5, without the completion network,Gc, the transferred face ...

    2018

  15. [23]

    Davis E. King. Dlib-ml: A machine learning toolkit. Journal of Machine Learning Research, 10:1755–1758, 2009

    2009

  16. [24]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 , 2014

    2014 arXiv

  17. [25]

    Photo- realistic single image super-resolution using a generative ad- versarial network

    Christian Ledig, Lucas Theis, Ferenc Husz´ar, Jose Caballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, et al. Photo- realistic single image super-resolution using a generative ad- versarial network. In Proceedings of the IE...

    2017

  18. [26]

    Face swapping under large pose variations: A 3D model based approach

    Yuan Lin, Shengjin Wang, Qian Lin, and Feng Tang. Face swapping under large pose variations: A 3D model based approach. In Int. Conf. on Multimedia and Expo , pages 333–

  19. [27]

    Large-scale celebfaces attributes (celeba) dataset

    Ziwei Liu, Ping Luo, Xiaogang Wang, and Xiaoou Tang. Large-scale celebfaces attributes (celeba) dataset. Retrieved August, 15:2018, 2018

    2018

  20. [28]

    Least squares generative adversarial networks

    Xudong Mao, Qing Li, Haoran Xie, Raymond YK Lau, Zhen Wang, and Stephen Paul Smolley. Least squares generative adversarial networks. In Proceedings of the IEEE Interna- tional Conference on Computer Vision , pages 2794–2802, 2017

    2017

  21. [29]

    Face-specific data augmentation for un- constrained face recognition

    Iacopo Masi, Anh Tun Trn, Tal Hassner, Gozde Sahin, and G´erard Medioni. Face-specific data augmentation for un- constrained face recognition. Int. J. Comput. Vision , 127(6- 7):642–667, 2019

    2019

  22. [30]

    Iarpa janus benchmark-c: Face dataset and protocol

    Brianna Maze, Jocelyn Adams, James A Duncan, Nathan Kalka, Tim Miller, Charles Otto, Anil K Jain, W Tyler Niggel, Janet Anderson, Jordan Cheney, et al. Iarpa janus benchmark-c: Face dataset and protocol. In 2018 Inter- national Conference on Biometrics (ICB) , pages 158–165. I...

    2018

  23. [31]

    Conditional generative adversarial nets

    Mehdi Mirza and Simon Osindero. Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784, 2014

    2014 arXiv

  24. [32]

    Pho- torealistic face de-identification by aggregating donors face components

    Saleh Mosaddegh, Loic Simon, and Fr ´ed´eric Jurie. Pho- torealistic face de-identification by aggregating donors face components. In Asian Conf. Comput. Vision, pages 159–174. Springer, 2014

    2014

  25. [33]

    Fsnet: An identity-aware generative model for image-based face swapping

    Ryota Natsume, Tatsuya Yatagawa, and Shigeo Morishima. Fsnet: An identity-aware generative model for image-based face swapping. In Proc. of Asian Conference on Computer Vision (ACCV). Springer, Dec 2018

    2018

  26. [34]

    Rsgan: face swapping and editing using face and hair repre- sentation in latent spaces

    Ryota Natsume, Tatsuya Yatagawa, and Shigeo Morishima. Rsgan: face swapping and editing using face and hair repre- sentation in latent spaces. arXiv preprint arXiv:1804.03447, 2018

    2018 arXiv

  27. [35]

    On face segmentation, face swapping, and face perception

    Yuval Nirkin, Iacopo Masi, Anh Tran Tuan, Tal Hassner, and Gerard Medioni. On face segmentation, face swapping, and face perception. In Automatic Face & Gesture Recognition (FG 2018), 2018 13th IEEE International Conference on , pages 98–105. IEEE, 2018

    2018

  28. [36]

    Pois- son image editing

    Patrick P ´erez, Michel Gangnet, and Andrew Blake. Pois- son image editing. ACM Transactions on graphics (TOG) , 22(3):313–318, 2003

    2003

  29. [37]

    Ganimation: Anatomically-aware facial animation from a single image

    Albert Pumarola, Antonio Agudo, Aleix M Martinez, Al- berto Sanfeliu, and Francesc Moreno-Noguer. Ganimation: Anatomically-aware facial animation from a single image. In Proceedings of the European Conference on Computer Vi- sion (ECCV), pages 818–833, 2018

    2018

  30. [38]

    U- net: Convolutional networks for biomedical image segmen- tation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U- net: Convolutional networks for biomedical image segmen- tation. In International Conference on Medical image com- puting and computer-assisted intervention , pages 234–241. Springer, 2015

    2015

  31. [39]

    Faceforen- sics++: Learning to detect manipulated facial images

    Andreas R ¨ossler, Davide Cozzolino, Luisa Verdoliva, Chris- tian Riess, Justus Thies, and Matthias Nießner. Faceforen- sics++: Learning to detect manipulated facial images. arXiv, 2019

    2019

  32. [40]

    Triple consistency loss for pairing distributions in gan-based face synthesis

    Enrique Sanchez and Michel Valstar. Triple consistency loss for pairing distributions in gan-based face synthesis. arXiv preprint arXiv:1811.03492, 2018

    2018 arXiv

  33. [41]

    Very deep convo- lutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep convo- lutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014

    2014 arXiv

  34. [42]

    Synthesizing obama: learn- ing lip sync from audio

    Supasorn Suwajanakorn, Steven M Seitz, and Ira Kemelmacher-Shlizerman. Synthesizing obama: learn- ing lip sync from audio. ACM Transactions on Graphics (TOG), 36(4):95, 2017

    2017

  35. [43]

    Figaro, hair detection and seg- mentation in the wild

    Michele Svanera, Umar Riaz Muhammad, Riccardo Leonardi, and Sergio Benini. Figaro, hair detection and seg- mentation in the wild. In 2016 IEEE International Con- ference on Image Processing (ICIP) , pages 933–937. IEEE, 2016

    2016

  36. [44]

    Face2face: Real-time face capture and reenactment of rgb videos

    Justus Thies, Michael Zollhofer, Marc Stamminger, Chris- tian Theobalt, and Matthias Nießner. Face2face: Real-time face capture and reenactment of rgb videos. In Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 2387–2395, 2016

    2016

  37. [45]

    Cr-gan: learning complete representations for multi-view generation

    Yu Tian, Xi Peng, Long Zhao, Shaoting Zhang, and Dim- itris N Metaxas. Cr-gan: learning complete representations for multi-view generation. arXiv preprint arXiv:1806.11191, 2018

    2018 arXiv

  38. [46]

    Extreme 3D face reconstruc- tion: Looking past occlusions

    Anh Tuan Tran, Tal Hassner, Iacopo Masi, Eran Paz, Yuval Nirkin, and G ´erard Medioni. Extreme 3D face reconstruc- tion: Looking past occlusions. In Proc. Conf. Comput. Vision Pattern Recognition, 2018

    2018

  39. [47]

    High-resolution image syn- thesis 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 syn- thesis and semantic manipulation with conditional gans. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018

    2018

  40. [48]

    An eye for an eye: A single camera gaze-replacement method

    Lior Wolf, Ziv Freund, and Shai Avidan. An eye for an eye: A single camera gaze-replacement method. In Proc. Conf. Comput. Vision Pattern Recognition , pages 817–824. IEEE, 2010

    2010

  41. [49]

    Gp-gan: Towards realistic high-resolution image blending

    Huikai Wu, Shuai Zheng, Junge Zhang, and Kaiqi Huang. Gp-gan: Towards realistic high-resolution image blending. arXiv preprint arXiv:1703.07195, 2017

    2017 arXiv

  42. [50]

    Reenactgan: Learning to reenact faces via boundary transfer

    Wayne Wu, Yunxuan Zhang, Cheng Li, Chen Qian, and Chen Change Loy. Reenactgan: Learning to reenact faces via boundary transfer. In ECCV, Sept. 2018

    2018

  43. [51]

    Seman- tic image inpainting with deep generative models

    Raymond A Yeh, Chen Chen, Teck Yian Lim, Alexander G Schwing, Mark Hasegawa-Johnson, and Minh N Do. Seman- tic image inpainting with deep generative models. In Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 5485–5493, 2017

    2017

  44. [52]

    Unpaired image-to-image translation using cycle- consistent adversarial networks

    Jun-Yan Zhu, Taesung Park, Phillip Isola, and Alexei A Efros. Unpaired image-to-image translation using cycle- consistent adversarial networks. In Proceedings of the IEEE International Conference on Computer Vision , pages 2223– 2232, 2017. Figure 9: Additional qualitative fac...

    2017

Pith tools

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