Pith. sign in

REVIEW 5 major objections 6 minor 39 references

Pose and Facial Expression Transfer by using StyleGAN

T0 review · 5 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Given a source and a target face portrait, this paper claims a single linear mapping in StyleGAN2's latent space can transfer the source's pose and expression onto the target's identity, trained without labels.

desk verdict The architecture is simple and the write-up is honest, but in the best configuration nothing directly supervises cross-identity pose/expression transfer, so the central claim rests on an unverified disentanglement assumption. read the letter →

arxiv 2504.13021 v1 pith:JFDNXVOP submitted 2025-04-17 cs.CV cs.AI

Petr Jahoda , Jan Cech This is my paper · ORCID
classification cs.CVcs.AI
keywords facereenactmentposetransferexpressionStyleGAN2latentspacemanipulationself-supervisedlearningimage-to-imagetranslationvideo-drivenanimation
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 proposes a method that takes two face portraits, a source and a target, and outputs a new image of the target person wearing the source's pose and facial expression. The authors argue that the task can be solved without 3D models, manual labels, or training a new generator: a motion encoder and a fixed identity encoder feed concatenated embeddings through one linear mapping into the latent space of a pretrained StyleGAN2, which renders the result. Training is self-supervised on unlabeled video clips of many identities, using reconstruction, perceptual, and face-recognition losses, plus an optional contrastive loss on motion embeddings. The reported numbers put identity preservation at 0.80 cosine similarity, close to the 0.83 ceiling of inverting the target image alone, which suggests the transfer step itself costs little identity fidelity. If the method works, it makes photorealistic face reenactment from a single still image practical, since the driving video can be rendered at about 20 frames per second.

What carries the argument

The load-bearing component is the mapping network $M$: a single fully connected linear layer that takes the concatenation of the motion embedding $E_m(s)$ and the identity embedding $E_i(t)$, both living in the extended latent space $W^+$ of dimension $18\times 512$, and produces the StyleGAN2 latent code that generates the output. Because the generator $G$ and the identity encoder $E_i$ are frozen, the entire learning problem reduces to finding a linear projection that mixes motion and identity without confusing them; the motion encoder $E_m$ is the only other trained part. The identity encoder is a pretrained ReStyle iteration of the pSp encoder, chosen after the authors found the alternative e4e configuration struggled to preserve identity. A practical detail is that the L2 and LPIPS losses are computed on a central $188\times 188$ crop, which the authors report was important for faithful expression transfer, since the surrounding hair and background otherwise distracted training.

What would settle it

Take a video of one person, feed each frame through the fixed identity encoder, and train a linear classifier or regressor to predict head pose (yaw/pitch) from the resulting identity embeddings; if pose can be predicted well above chance from those embeddings, the claimed identity-motion separation fails, and the linear mapping cannot be expected to transfer pose independently across identities.

Watch

Extended reading notes

Core claim

The central claim is that pose and expression form an identity-independent motion code that can be extracted from one face and re-injected into another face's identity code, all inside StyleGAN2's latent space, without any explicit 3D face model. Formally, the system computes $g_{s\rightarrow t}=G(M(E_m(s)\oplus E_i(t)))$, where $E_m$ embeds the source's motion, $E_i$ embeds the target's identity with a fixed pretrained encoder, the linear mapping $M$ mixes the two, and the fixed generator $G$ renders the image. The motion encoder and mapping are trained on VoxCeleb2 video clips by reconstructing same-identity pairs with L2 and LPIPS losses applied to a central crop, by enforcing identity preservation with ArcFace cosine similarity, and optionally by a CosFace contrastive loss that pulls together motion embeddings of the same expression and pushes apart different ones. The final model reports mean absolute pose error of 7.67 on yaw and pitch, facial-landmark expression correlation of 0.62, and identity cosine similarity of 0.80, beating the latent-arithmetic baselines on pose and identity. The authors also show the same pipeline can impose a source's pose and expression onto randomly generated StyleGAN identities, which makes the motion code usable as a control signal for synthesizing new faces.

Load-bearing premise

The load-bearing assumption is that the pretrained identity encoder produces an embedding that contains identity only, cleanly separated from pose and expression, so a single linear mapping over the concatenated codes can swap motion without disturbing identity; the paper does not measure how entangled the identity code actually is.

Editorial extensions

If this is right

  • A single target still plus any driving video yields a reenacted video: after one half-second inversion of the target, the model renders up to 20 frames per second.
  • Identity similarity of 0.80 versus a 0.83 inversion ceiling means the transfer step costs almost no identity fidelity, so the encoder is the bottleneck rather than the mapping.
  • Because the generator is frozen, the method inherits StyleGAN2's coverage; expressions the generator cannot render, such as fully closed eyes, will fail until the generator is fine-tuned or regularized.
  • The method beats both ReStyle-based latent-arithmetic baselines on pose error and identity preservation, so the proposed mapping is a more reliable way to reenact faces than direct vector arithmetic in the latent space.
  • Random StyleGAN identities can be driven by the same motion code, so pose and expression control extends beyond real photographs to synthetic faces.

Reading between the lines

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

  • If the identity code from the pretrained encoder is genuinely free of pose and expression, the motion code could be driven by non-image inputs such as audio or body landmarks, since the mapping accepts any vector in the motion-embedding space; the paper only demonstrates image-driven motion.
  • The linearity of the mapping suggests that simple vector arithmetic on motion embeddings, such as averaging several source expressions or interpolating between poses, would transfer cleanly across identities; the paper does not test this, but it follows from the architecture.
  • The paper's expression metric is a proxy based on landmark aspect ratios, so it may understate the method on asymmetric expressions or eye movements; a FAU-based evaluation on data with strong activations could reveal where the proxy misleads.
  • Because random identities must be rendered and then inverted to enter the model's operating subspace, applications that need arbitrary StyleGAN latent codes would have to round-trip through the encoder, suggesting the mapping lives in a restricted subspace of $W^+$.
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

5 major / 6 minor

Summary. The paper proposes a one-shot face reenactment method that transfers pose and expression from a source face image onto the identity of a target face image. The architecture consists of a motion encoder (ResNet-IR SE50), a fixed identity encoder (ReStyle pSp), a single linear mapping network, and a fixed pre-trained StyleGAN2 generator. Em and M are trained self-supervised on unlabeled VoxCeleb2 video sequences using a composite loss: L2 and LPIPS losses on same-identity reconstruction outputs, an ArcFace identity loss on cross-identity outputs, an optional CosFace motion-alignment loss that is disabled in the best configuration, and cropped variants of the reconstruction losses. Experiments on a VoxCeleb2 test split report pose MAE of 7.67 degrees, expression landmark cross-correlation of 0.62, and identity cosine similarity of 0.80, with two latent-space-editing baselines. The paper also describes a procedure for transferring pose/expression to randomly sampled StyleGAN2 identities and reports close-to-real-time inference after a one-time identity encoding step.

Significance. If the claims hold, the main contribution is a simple, fast, and reproducible pipeline built from open-source pretrained components that demonstrates how a linear fusion of separate motion and identity embeddings can drive StyleGAN2 for pose/expression transfer. The self-supervised video-based training, the use of standard components, and the reported inference speed are genuine strengths. However, the quantitative evidence is not yet conclusive: the cross-identity output is never directly supervised for pose and expression, the quantitative evaluation lacks statistical validation and a strong single-image baseline, and the expression metric is a coarse proxy. The paper is a plausible workshop-level result, but for a journal version the evaluation and the architectural justification need substantial reinforcement.

major comments (5)
  1. [Sec. 3.2 and Sec. 3.4, Eqs. (1)-(6)] The cross-identity output g_{sA->tB} is supervised only by the ArcFace identity loss L_ID (Eq. 3); the L2 and LPIPS losses and their cropped variants are defined only for the same-identity output g_{sA->tA}, and the CosFace loss is disabled in the best model (w_cos = 0). Consequently, the central claim that the source pose and expression are transferred onto the target identity is not directly trained. The method instead relies on the untested assumption that Ei(t) is free of pose/expression information and that a single linear layer on Em(s) concatenated with Ei(t) yields a valid latent code. This assumption is load-bearing: if Ei(t) leaks pose or expression, the cross-identity output can satisfy Eq. (3) while retaining the target's own pose and expression, and the reported pose/expression transfer would be an artifact of the evaluation setup. The authors should provide a disentanglement analysis, e.g., train a linear probe on Ei(t) to predict yaw/pitch and expression, and should ablate by adding a mild reconstruction or LPIPS loss on the cross-identity output or by engaging the CosFace loss, reporting the effect on all Table 1 metrics. Without such evidence, the mechanism behind the transfer remains unjustified.
  2. [Table 1 and Sec. 4.3] The quantitative comparison does not report error bars, confidence intervals, significance tests, or the number of evaluated frames/videos. The differences among methods are small in absolute terms (pose MAE: 7.673 vs. 7.968 vs. 8.491; expression FL CORR: 0.620 vs. 0.656 for Base pSp), and the expression metric is actually lower for the proposed method than for Base pSp. Without per-video standard errors and a paired significance test over the 120 test identities, the claim that the proposed method improves pose transfer over the baselines is not statistically supported. The authors should report the mean and standard error over videos (or bootstrap confidence intervals) and conduct a paired test (e.g., Wilcoxon signed-rank) between the methods.
  3. [Sec. 4.1] The baseline comparison is not a single-image reenactment comparison: the baseline does not use the actual target image but a frontalized, average-expression reconstruction computed from a video of the target identity (via vertical flipping and averaging latent codes). This gives the baseline two disadvantages unrelated to the core task: it requires a video rather than a single image, and the averaged frontal frame may itself be a poor proxy for the target's identity and expression. The comparison therefore does not isolate the benefit of the proposed method over latent-space editing for the stated one-shot setting. The authors should either replace the baseline with a true single-image latent-editing baseline (e.g., applying the edit vector computed from a matched source frame pair directly to the target latent code) or clearly delimit the claim to the video-based latent-arithmetic setting and explain why the baseline is appropriate.
  4. [Sec. 4.3, expression fidelity metric] The FL statistic is a proxy that computes cross-correlations of aspect ratios of eyes, mouth, and eyebrows, and the paper itself acknowledges it cannot capture asymmetric expressions, eyeball movements, or many subtle expression changes. The reported value of 0.62 is not anchored to human perception or to a known-good reference, so it is unclear what level of correlation constitutes successful expression transfer. The authors should supplement this with a user study, a facial action unit detector, or an expression classification metric to substantiate the expression-transfer claim, or at least calibrate the metric on a trivial baseline (e.g., correlation between the source and an identity-preserving but expression-frozen reconstruction).
  5. [Sec. 4.3, ID CSIM] The identity metric reported in Table 1 is the ArcFace cosine similarity, and the training loss L_ID in Eq. (3) is exactly its complement. The ID CSIM of 0.80 therefore partly reflects direct optimization of the evaluation metric, not an independent verification of identity preservation. The authors should explicitly acknowledge this circularity or report a second identity metric based on a different face recognition model to strengthen the claim.
minor comments (6)
  1. [Sec. 3.3 vs. Sec. 3.1] The paper states in Sec. 3.1 that input images are 'aligned and resized to 256 x 256 pixels,' but Sec. 3.3 says the official StyleGAN preprocessing script normalizes images to 224 x 224 pixels. This inconsistency should be resolved or clarified.
  2. [Sec. 3.2, Eqs. (4)-(5)] The CosFace loss is typeset in a way that is hard to read: the expression 'd(zi,z j)' and the fraction in Eq. (5) would benefit from standard mathematical notation and consistent subscript spacing.
  3. [Sec. 3.2] There is a typo: 'In praticular' should be 'In particular'.
  4. [Sec. 4.3, FL statistic] The description of the facial-landmark aspect ratios and the cross-correlation computation is vague; please provide the exact formulas and state whether the cross-correlation is normalized (e.g., Pearson correlation) and how the reported average is computed over videos.
  5. [Sec. 4.3, dataset size] The paper should state the number of test videos, frames, and source-target pairs used for the numbers in Table 1, and whether the metrics are averaged over frames, videos, or identities.
  6. [Sec. 2] The related work discusses X2Face [33] and MegaPortraits [9], but the experiments include no comparison with these methods; a sentence explaining why they are not compared (e.g., no public code, different input requirements) would improve the framing.

Circularity Check

1 steps flagged · score 2.0 of 10

No load-bearing circularity; one minor overlap: the identity-preservation score is exactly the ArcFace training objective.

  1. fitted input called prediction [This occurs in Section 3.2, Equation (3), and in Section 4.3 / Table 1.]
    "Identity loss . To ensure that the generated image preserves the identity of the target image, we employ the pre-trained facial recognition model ArcFace [8]. ... LID = 1−⟨D(tB),D (gsA→tB )⟩ ... For the evaluation of identity preservation, we use the ArcFace [8]. The ID error is the cosine similarity between the generated and the target (identity) frame descriptors."

    The reported identity metric ID(CSIM)=0.801 in Table 1 is the same cosine similarity that the training loss L_ID maximizes. Therefore the identity-preservation number is partly the training objective evaluated on the test split, not an independently derived validation of identity preservation. This overlap is minor rather than load-bearing because the central pose/expression transfer claim is evaluated with external pose and landmark metrics that are not training losses.

full rationale

The central claim of pose and expression transfer is not circular: the pose error uses an external CNN head-pose estimator and the expression fidelity uses Dlib facial landmarks, neither of which appears in the training objective. The cross-identity output is trained only with the ArcFace identity loss while the CosFace motion-alignment loss is disabled in the best model, but that is an empirical weakness about unverified disentanglement of the fixed ReStyle identity encoder, not a circular derivation. The self-citations in the paper, such as the architectural inspiration from [25] and the landmark aspect-ratio evaluation from [24], are not load-bearing; they support component choices and an evaluation proxy rather than proving the transfer result. The only quantitative overlap is the identity metric: Eq. (3) trains with ArcFace cosine similarity and Sec. 4.3 reports the same cosine similarity as ID(CSIM). Since the test identities are disjoint and the pose/expression results rest on independent external estimators, the paper is largely self-contained and the score is low.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

No new physical or mathematical entities are introduced. The central claim rests on pretrained components (StyleGAN2, ReStyle, ArcFace, LPIPS) and hand-tuned loss weights and data-selection choices. The ArcFace metric is used in both training and evaluation, which is the main circularity signal.

free parameters (4)
  • Loss weights (w_LPIPS, w_ID, w_L2_crop, w_LPIPS_crop; w_L2=0, w_cos=0) = 0.05, 0.3, 2, 0.3; 0, 0 (final model)
    Chosen by hand based on validation performance; the paper states 'with the best performance, we used the following hyperparameters' (Sec. 3.4). These weights affect the balance of reconstruction, identity, and expression fidelity.
  • CosFace margin parameters a and b = a=5, b=0.2
    Set as hyperparameters (Sec. 3.2); not in final model since w_cos=0, but used in variant Ours-Cos.
  • Central crop size 188x188 = 188x188
    Chosen after observing that full-frame losses struggle with expression transfer (Sec. 3.2).
  • Training frame sampling and extreme-pose filtering threshold = 10 frames at half-second intervals; extreme poses eliminated
    Manual data selection choices (Sec. 3.3) that affect pose diversity and could bias evaluation.
assumptions (5)
  • domain assumption Pretrained StyleGAN2 can generate faces over the required range of poses, expressions, and identities, including closed eyes.
    The method relies on StyleGAN2 as a frozen decoder; the limitation section admits closed eyes are problematic because StyleGAN rarely sees them (Sec. 5).
  • domain assumption ReStyle pSp encoder maps real images to a W+ subspace that is compatible with the mapping network and preserves identity.
    The identity encoder is fixed and used without fine-tuning (Sec. 3.1); the paper relies on its inversion quality and notes the upper bound of 0.83 ArcFace similarity (Sec. 4.3).
  • domain assumption VoxCeleb2 provides sufficient identity and pose/expression diversity for the self-supervised task.
    The training uses video clips from VoxCeleb2 with the stated assumption that interviews offer a broad spectrum of poses and expressions (Sec. 3.3).
  • domain assumption ArcFace cosine similarity is a valid proxy for identity preservation in both the loss and the evaluation.
    Used as identity loss (Eq. 3) and as evaluation metric (Sec. 4.3), which introduces partial circularity.
  • domain assumption The facial-landmark aspect-ratio cross-correlation is a valid proxy for expression transfer fidelity.
    The paper states it does not capture eyeball movements or asymmetric expressions but 'seems to correlate with subjective quality' (Sec. 4.3).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Pose and Facial Expression Transfer by using StyleGAN." pith.science (2026). https://pith.science/paper/JFDNXVOP

@misc{pith2026250413021,
  author       = {Pith},
  title        = {Pith review of: Pose and Facial Expression Transfer by using StyleGAN},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JFDNXVOP}},
  note         = {Machine review of arXiv:2504.13021}
}
read the original abstract

We propose a method to transfer pose and expression between face images. Given a source and target face portrait, the model produces an output image in which the pose and expression of the source face image are transferred onto the target identity. The architecture consists of two encoders and a mapping network that projects the two inputs into the latent space of StyleGAN2, which finally generates the output. The training is self-supervised from video sequences of many individuals. Manual labeling is not required. Our model enables the synthesis of random identities with controllable pose and expression. Close-to-real-time performance is achieved.

Figures

Figures reproduced from arXiv: 2504.13021 by the authors.

Figure 1
Figure 1. Results of our method. Pose and expression [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The architecture of the proposed model. The Motion encoder and Mapping network weights are trained, while [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Pose and expression transfer results. The top row depicts the target (identity) input images, leftmost column [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Pose and expression transfer comparison. The top two rows represent the input: source and target images. The [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 35 canonical work pages

  1. [1]

    Abdal, Y

    R. Abdal, Y . Qin, and P. Wonka. Image2StyleGAN: How to embed images into the StyleGAN latent space? In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision, pages 4432– 4441, 2019. 2

    work page 2019

  2. [2]

    Abdal, Y

    R. Abdal, Y . Qin, and P. Wonka. Image2stylegan++: How to Edit the Embedded Images? In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8296–8305, 2020. 2

    work page 2020

  3. [3]

    Abdal, P

    R. Abdal, P. Zhu, N. J. Mitra, and P. Wonka. Style- flow: Attribute-conditioned exploration of stylegan- generated images using conditional continuous nor- malizing flows. ACM Transactions on Graphics (ToG), 40(3):1–21, 2021. 2

    work page 2021

  4. [4]

    Alaluf, O

    Y . Alaluf, O. Patashnik, and D. Cohen-Or. Restyle: A Residual-based StyleGAN Encoder via Iterative Refinement. In Proceedings of the IEEE/CVF In- ternational Conference on Computer Vision , pages 6711–6720, 2021. 2, 5

    work page 2021

  5. [5]

    Blanz and T

    V . Blanz and T. Vetter. A morphable model for the synthesis of 3d faces. In Proceedings of the 26th annual conference on Computer graphics and inter- active techniques, pages 187–194, 1999. 2

    work page 1999

  6. [6]

    J. S. Chung, A. Nagrani, and A. Zisserman. V ox- Celeb2: Deep Speaker Recognition. In Proc. Inter- speech 2018, pages 1086–1090, 2018. 4, 7

    work page 2018

  7. [7]

    Creswell and A

    A. Creswell and A. A. Bharath. Inverting the Gen- erator of a Generative Adversarial Network. IEEE transactions on neural networks and learning sys- tems, 30(7):1967–1974, 2018. 2

    work page 1967

  8. [8]

    J. Deng, J. Guo, N. Xue, and S. Zafeiriou. Arc- face: Additive Angular Margin Loss for Deep Face Recognition. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 4690–4699, 2019. 4, 7

    work page 2019

Show all 39 references
  1. [9]

    Drobyshev, J

    N. Drobyshev, J. Chelishev, T. Khakhulin, A. Ivakhnenko, V . Lempitsky, and E. Zakharov. Megaportraits: One-shot megapixel neural head avatars. arXiv preprint arXiv:2207.07621, 2022. 2, 4, 8

    2022 arXiv

  2. [10]

    Ekman and W

    P. Ekman and W. V . Friesen. Facial action coding system. Environmental Psychology & Nonverbal Be- havior, 1978. 7

    1978

  3. [11]

    Goodfellow, J

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Ben- gio. Generative adversarial networks. InAdvances in Neural Information Processing Systems, volume 27, pages 2672–2680. Curran Associates, Inc., 2014. 1

    2014

  4. [12]

    H ¨ark¨onen, A

    E. H ¨ark¨onen, A. Hertzmann, J. Lehtinen, and S. Paris. Ganspace: Discovering interpretable gan controls. Advances in Neural Information Process- ing Systems, 33:9841–9850, 2020. 2

    2020

  5. [13]

    Karras, S

    T. Karras, S. Laine, and T. Aila. Flickr-faces- hq dataset (ffhq). https://github.com/ NVlabs/ffhq-dataset, 2019. 5

    2019

  6. [14]

    Karras, S

    T. Karras, S. Laine, M. Aittala, J. Hellsten, J. Lehti- nen, and T. Aila. Analyzing and Improving the Im- age Quality of StyleGAN. In Proceedings of the IEEE/CVF conference on computer vision and pat- tern recognition, pages 8110–8119, 2020. 1, 2

    2020

  7. [15]

    D. E. King. Dlib-ml: A Machine Learning Toolkit. The Journal of Machine Learning Research, 10:1755–1758, 2009. 7

    2009

  8. [16]

    Krizhevsky, I

    A. Krizhevsky, I. Sutskever, and G. E. Hinton. Im- agenet classification with deep convolutional neural networks. In F. Pereira, C. Burges, L. Bottou, and K. Weinberger, editors, Advances in Neural Infor- mation Processing Systems, volume 25. Curran As- sociates, Inc., 2012. 3

    2012

  9. [17]

    C. Luo, S. Song, W. Xie, L. Shen, and H. Gunes. Learning Multi-dimensional Edge Feature-based au Relation Graph for Facial Action Unit Recognition. arXiv preprint arXiv:2205.01782, 2022. 7

    2022 arXiv

  10. [18]

    Petr ˇzelkov´a

    N. Petr ˇzelkov´a. Face image editing in latent space of generative adversarial networks, Prague, 2021. Bachelor thesis. CTU in Prague, Faculty of Electri- cal Engineering, Department of Cybernetics. 2

    2021

  11. [19]

    Radford, L

    A. Radford, L. Metz, and S. Chintala. Unsu- pervised representation learning with deep convo- lutional generative adversarial networks. arXiv preprint arXiv:1511.06434, 2015. 2

    2015 arXiv

  12. [20]

    Richardson, Y

    E. Richardson, Y . Alaluf, O. Patashnik, Y . Nitzan, Y . Azar, S. Shapiro, and D. Cohen-Or. Encoding in Style: a StyleGAN Encoder for Image-to-Image Translation. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 2287–2296, 2021. 2, 3, 5

    2021

  13. [21]

    Roich, R

    D. Roich, R. Mokady, A. H. Bermano, and D. Cohen-Or. Pivotal Tuning for Latent-based Edit- ing of Real Images. ACM Transactions on Graphics (TOG), 42(1):1–13, 2022. 2

    2022

  14. [22]

    N. Ruiz, E. Chong, and J. M. Rehg. Fine-grained head pose estimation without keypoints. InProceed- ings of the IEEE conference on computer vision and pattern recognition workshops , pages 2074–2083,

  15. [23]

    Y . Shen, J. Gu, X. Tang, and B. Zhou. Interpreting the Latent Space of GANs for Semantic Face Edit- ing. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 9243–9252, 2020. 2

    2020

  16. [24]

    Soukupova and J

    T. Soukupova and J. Cech. Eye blink detection us- ing facial landmarks. In 21st computer vision winter workshop, Rimske Toplice, Slovenia , page 2, 2016. 7

    2016

  17. [25]

    Subrtova, J

    A. Subrtova, J. Cech, and V . Franc. Hairstyle trans- fer between face images. 2021 16th IEEE Interna- tional Conference on Automatic Face and Gesture Recognition (FG 2021), pages 1–8, 2021. 3

    2021

  18. [26]

    Thies, M

    J. Thies, M. Zollh ¨ofer, M. Nießner, L. Valgaerts, M. Stamminger, and C. Theobalt. Real-time Expres- sion Transfer for Facial Reenactment. ACM Trans. Graph., 34(6):183–1, 2015. 2

    2015

  19. [27]

    Thies, M

    J. Thies, M. Zollhofer, M. Stamminger, C. Theobalt, and M. Nießner. Face2Face: Real-time Face Capture and Reenactment of RGB Videos. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2387–2395, 2016. 2

    2016

  20. [28]

    O. Tov, Y . Alaluf, Y . Nitzan, O. Patashnik, and D. Cohen-Or. Designing an Encoder for StyleGAN Image Manipulation. ACM Transactions on Graph- ics (TOG), 40(4):1–14, 2021. 2, 3, 5

    2021

  21. [29]

    Tzaban, R

    R. Tzaban, R. Mokady, R. Gal, A. Bermano, and D. Cohen-Or. Stitch it in Time: GAN-based Facial Editing of Real Videos. In SIGGRAPH Asia 2022 Conference Papers, pages 1–9, 2022. 2

    2022

  22. [30]

    ˇSubrtov´a, D

    A. ˇSubrtov´a, D. Futschik, J. ˇCech, M. Luk ´aˇc, E. Shechtman, and D. S ´ykora. ChunkyGAN: Real image inversion via segments. InProceedings of Eu- ropean Conference on Computer Vision, pages 189– 204, 2022. 2

    2022

  23. [31]

    H. Wang, Y . Wang, Z. Zhou, X. Ji, D. Gong, J. Zhou, Z. Li, and W. Liu. Cosface: Large Margin Cosine Loss For Deep Face Recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 5265–5274, 2018. 4

    2018

  24. [32]

    T.-C. Wang, A. Mallya, and M.-Y . Liu. One-shot Free-view Neural Talking-head Synthesis for Video Conferencing. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 10039–10049, 2021. 2

    2021

  25. [33]

    Wiles, A

    O. Wiles, A. Koepke, and A. Zisserman. X2face: A Network For Controlling Face Generation Using Im- ages, Audio, and Pose Codes. In Proceedings of the European conference on computer vision (ECCV) , pages 670–686, 2018. 2

    2018

  26. [34]

    Wright and N

    L. Wright and N. Demeure. Ranger21: a synergis- tic deep learning optimizer. CoRR, abs/2106.13731,

    arXiv

  27. [35]

    Yang and S.-N

    C. Yang and S.-N. Lim. Unconstrained facial ex- pression transfer using style-based generator. arXiv preprint arXiv:1912.06253, 2019. 3

    1912 arXiv

  28. [36]

    Zakharov, A

    E. Zakharov, A. Shysheya, E. Burkov, and V . Lem- pitsky. Few-shot adversarial learning of realistic neural talking head models. In Proceedings of the IEEE/CVF international conference on computer vi- sion, pages 9459–9468, 2019. 2

    2019

  29. [37]

    Zhang, P

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang. The Unreasonable Effectiveness of Deep Features as a Perceptual Metric. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 586–595, 2018. 3

    2018

  30. [38]

    J.-Y . Zhu, P. Kr¨ahenb¨uhl, E. Shechtman, and A. A. Efros. Generative visual manipulation on the natural image manifold. In Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Nether- lands, October 11-14, 2016, Proceedings, Part V 14, pages 597–613. Springer, 2016. 2

    2016

  31. [39]

    P. Zhu, R. Abdal, Y . Qin, J. Femiani, and P. Wonka. Improved StyleGAN Embedding: Where are the Good Latents? arXiv preprint arXiv:2012.09036 ,

    2012 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.