Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

LS-GAN: Human Motion Synthesis with Latent-space GANs

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A simple GAN in a frozen motion latent space matches latent diffusion quality for text-to-motion at a fraction of the compute.

desk verdict Useful efficiency test of latent GANs for motion synthesis, but the headline FID-FLOPs number is not reliably attributed until Table 2 is fixed. read the letter →

arxiv 2501.01449 v1 pith:J2J3QAE6 submitted 2024-12-30 cs.CV cs.AI

Avinash Amballa , Gayathri Akkinapalli , Vinitra Muralikrishnan This is my paper · ORCID
classification cs.CVcs.AI
keywords text-to-motionsynthesishumanmotionlatentspaceGANconditionalWGAN-GPML3Daction-to-motion
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 tries to establish that a plain GAN can replace the diffusion process inside a frozen motion latent space and still generate high-quality, text-conditioned 3D human motion. On the HumanML3D benchmark, the best variant, a deep Wasserstein GAN with gradient penalty, reaches an FID of 0.482, close to the latent diffusion model MLD's 0.473, while requiring roughly 91% fewer FLOPs for inference (2.665 G versus 29.86 G FLOPs per 2048 clips). The same recipe transfers to action-to-motion on HumanAct12, where it matches or beats existing models and achieves the highest diversity. A sympathetic reader would care because the result suggests that expensive iterative diffusion sampling is not necessary for motion synthesis once a good latent representation exists, opening the way to faster training and near-real-time generation.

What carries the argument

The load-bearing object is the frozen VAE latent space borrowed from MLD: a transformer-based encoder-decoder pair that maps a motion sequence $x_{1:L}$ to a low-dimensional latent and back, trained with reconstruction and KL losses and then fixed. The LS-GAN generator takes a random vector and a condition, a CLIP text embedding for text or a learned embedding for action labels, and outputs a latent code in the VAE space; the discriminator tries to tell apart real latents $E(x)$ from fake latents $G(z,c)$. The version that carries the paper's headline result is the Deep WGAN-GP: residual blocks in both networks plus the Wasserstein loss with gradient penalty, which the t-SNE visualizations show separates action clusters more cleanly than the vanilla GAN variants and yields higher MultiModality.

What would settle it

Run the reported Deep WGAN-GP model through the official HumanML3D evaluation code; if the measured FID is not near 0.482 or the FLOPs for 2048 clips do not come out near 2.665 G, the central result fails.

Watch

Extended reading notes

Core claim

The central claim is that motion synthesis conditioned on text or action labels can be carried out by a small conditional GAN that generates latent codes inside a frozen VAE latent space, rather than by a diffusion model. The paper reports that its Deep WGAN-GP model, built from fully connected layers plus two residual blocks and trained with Wasserstein loss and gradient penalty, achieves an FID of 0.482 on HumanML3D, essentially matching MLD's 0.473 while cutting FLOPs from 29.86 G to 2.665 G for 2048 motion clips, a 91.1% reduction. The same model also achieves the highest MultiModality score (3.501) among all compared text-to-motion methods, and on HumanAct12 it produces the highest Diversity (6.850) while staying competitive on FID and accuracy. In short, the paper argues that the generative model class matters less than the latent space it operates in: once motions are encoded into a compact, structured latent space by a VAE, a simple GAN can compete with latent diffusion.

Load-bearing premise

The method assumes that the frozen VAE latent space is smooth and informative enough that a small GAN trained entirely on its encoded latents can generate diverse, high-quality motions; if the latent space has discontinuities or poorly separated modes, the GAN will collapse or decode into poor motion.

Editorial extensions

If this is right

  • Near-parity FID (0.482 versus 0.473) at 91.1% fewer FLOPs means text-to-motion quality no longer requires iterative diffusion sampling, making real-time or embedded deployment plausible.
  • The same frozen latent space supports both text conditioning and action conditioning, so the GAN approach transfers across conditions without architectural changes.
  • The method beats the raw-motion diffusion model MDM on FID, R-precision, MM Dist, and MModality, indicating that operating in latent space carries most of the benefit.
  • The MultiModality score of 3.501 is the highest among the compared methods, so the GAN does not sacrifice per-prompt diversity for speed.
  • Training is also cheaper, because GAN optimization replaces the diffusion objective and the VAE is never retrained.

Reading between the lines

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

  • If the frozen latent space is the quality bottleneck, then improving VAE training, through better priors, more disentanglement, or longer training, should raise the ceiling for both GAN and diffusion generators; the paper does not vary the VAE.
  • The single-forward-pass inference advantage should widen for longer sequences or higher frame rates, where iterative diffusion sampling becomes increasingly costly; this is a testable prediction the paper does not directly measure.
  • The recipe should transfer to any conditional generation domain that already has a fixed latent space, such as audio, facial motion, or hand motion, though the paper only studies full-body motion.
  • A controlled comparison that keeps the VAE and CLIP conditioning identical while swapping only the generative model would directly quantify how much of the GAN's near-parity owes to the latent space; this follows naturally from the paper's setup.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes LS-GAN, a text- and action-conditioned motion synthesis method that replaces the latent diffusion model of MLD with a conditional GAN operating on a frozen VAE latent space. Using the MLD VAE architecture and CLIP text embeddings, the authors train four GAN variants (vanilla/deep architectures with BCE or WGAN-GP losses). On HumanML3D they report a best FID of 0.482 at 2.665 G FLOPs per 2048 motion clips, a 91.1% inference-FLOPs reduction relative to MLD at 50 DDIM steps; on HumanAct12 they report competitive action-to-motion results for a Deep WGAN-GP variant. The paper also includes t-SNE visualizations and a limitations section.

Significance. If the headline operating point is confirmed, the paper provides evidence that a simple GAN in a pretrained motion latent space can approach latent-diffusion quality at a fraction of the inference cost, which is an interesting and practically relevant result. The evaluation is grounded in an external benchmark (HumanML3D) with standard metrics, and the fixed VAE/CLIP stack prevents self-referential evaluation. The inference-FLOPs figures are concrete and falsifiable. However, the central evidence table mislabels the model behind the headline number, training-time claims are unmeasured, and at least one direct comparison sentence is contradicted by the paper's own Table 1. These issues must be resolved before the efficiency-quality claim can be accepted as stated.

major comments (5)
  1. [Section 5.2, Table 2] Table 2 attributes FID 0.482 and FLOPs 2.665 G to a row labeled 'Deep GAN,' but Table 1 reports Deep GAN FID 3.036 and Deep WGAN-GP FID 0.482. The surrounding text says 'Our Deep WGAN-GP achieves a FID of 0.482 ... as shown in Table 2.' Consequently, the headline operating point (FID 0.482 at 2.665 G) is not attached to a consistently specified model in the paper. Please relabel Table 2 to match Table 1, or report FLOPs for all four variants and state explicitly which model achieves FID 0.482. This is load-bearing because it is precisely the central efficiency-quality claim.
  2. [Section 5.2] The sentence 'our LS-GAN outperforms cross-modal models such as Seq2Seq [37], LJ2P [3], T2G [5], Hier [9], TEMOS [36], T2M [15] across all evaluation metrics' is contradicted by Table 1: Deep WGAN-GP has R-precision top-1 0.391 ± 0.002, whereas TEMOS reports 0.424 ± 0.002 and T2M reports 0.457 ± 0.002. The claim should be weakened or restricted to the metrics and models for which it actually holds.
  3. [Abstract; Section 1; Section 5.2; Section 8] The paper repeatedly claims faster training and inference time, but no training time, wall-clock convergence, or total training compute measurement appears anywhere. The reported 91% FLOPs reduction is an inference-compute estimate, not a training-time result, and GAN training can require substantially different total compute and hyperparameter tuning than diffusion training. Either report measured training time against MLD under comparable conditions or replace the training-time claims with inference-FLOPs claims.
  4. [Section 3.3; Section 5.1] The architecture description is too incomplete to reproduce the central result. The paper specifies only that the vanilla generator has three fully connected layers, the discriminator has four, and the deep variants add two residual blocks, without giving hidden widths, residual block details, number of GAN training iterations, discriminator-to-generator update ratio, gradient-penalty coefficient, or other hyperparameters. No code is released. Since the contribution is a 'remarkably simple' GAN, please provide a complete configuration, including exact layer sizes and the full training schedule.
  5. [Table 2] Please clarify whether the MLD and MDM FLOPs values were counted with the same THOP procedure as the LS-GAN rows or were taken from the original papers. If the comparison mixes counting methodologies, the 91% reduction statement is not necessarily an apples-to-apples comparison. State explicitly which rows were measured by the authors and which were borrowed.
minor comments (5)
  1. [Table 2] The column heading 'Parameter' is misleading: the entries list motion or latent dimensionalities (e.g., x ∈ R196×512, z ∈ R1×256), not parameter counts. Rename the column or report actual parameter counts.
  2. [Figure 4] The PDF text of Figure 4 contains corrupted font sequences such as '/uni00000014/uni00000013...' in the caption. Please regenerate the figure with embedded text rendered properly.
  3. [Section 5.1] The text says c ∈ R1×10 for the action-to-motion task, but HumanAct12 has 12 action categories. Please verify and correct the conditioning dimension.
  4. [References] Several references are duplicated: [2]/[3], [13]/[14]/[15], [34]/[35], and [48]/[49] point to the same works. Consolidate them into single entries.
  5. [Section 3.1; Section 5.1] Section 3.1 says the VAE is 'borrowed from MLD,' while Section 5.1 says 'Our VAE model' is trained with the same loss configuration. Clarify whether the VAE weights are inherited from MLD or retrained by the authors, since this affects the interpretation of the reported VAE metrics in Table 4.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the central FID/FLOPs claim is evaluated on external HumanML3D benchmarks using a fixed external VAE and text encoder; no fitted parameter is renamed as a prediction.

full rationale

The paper's derivation chain is self-contained in the sense relevant to circularity. The VAE and CLIP text encoder are borrowed from MLD [7] and kept fixed, and the GAN is trained to map noise plus condition to the frozen latent space, with quality measured by FID, R-precision, diversity, and multimodality on the external HumanML3D and HumanAct12 test sets using the standard motion encoder from T2M [15]. The headline FID of 0.482 and the 91% FLOPs reduction are empirical results compared against MLD's reported numbers, not quantities implied by the model definitions or by the training loss. There is no self-citation chain: the paper contains no references to prior work by its own authors, and no 'uniqueness theorem' or ansatz is imported from the authors' earlier papers. The heavy reliance on MLD is a dependence on prior work for architecture and baselines, but it is not circular: the VAE is a fixed preprocessing stage, and the GAN's performance is not defined in terms of MLD's FID. The internal inconsistency between Table 1 (assigning FID 0.482 to Deep WGAN-GP) and Table 2 (labeling the 0.482/FLOPS 2.665 row as Deep GAN) is a reproducibility and reporting concern, not a circularity concern, because neither table entry is constructed from the other or from the claimed conclusion. Likewise, reporting metrics from the training checkpoint with the lowest test FID may be a model-selection risk, but it is not a case of fitting a parameter to a target and then re-labeling the fit as a prediction. No load-bearing step reduces, by the paper's own equations or by self-citation, to its own inputs.

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

The method introduces no new physical or mathematical entities. It relies on several domain assumptions: a well-behaved VAE latent space, sufficient CLIP conditioning, and comparable evaluation pipelines. The free parameters are mostly architectural and training hyperparameters chosen by hand without ablations.

free parameters (5)
  • Generator input latent dimension = 100
    Chosen by hand in Section 5.1; no ablation or justification.
  • Generated latent dimension = 256
    Chosen by hand in Section 5.1; matches MLD's latent size.
  • VAE checkpoint iteration = 1250
    Selected as the checkpoint with best FID in Table 4; introduces selection bias into the GAN evaluation.
  • Learning rate = 1e-4
    Fixed for all models, no schedule or sensitivity analysis reported.
  • GAN architecture depth = 3 FC layers (Vanilla), plus 2 residual blocks (Deep)
    Hand-chosen; no ablation study compares other depths or widths.
assumptions (3)
  • domain assumption The VAE latent space is smooth and informative enough for a small GAN to generate diverse, high-quality motions.
    The GAN is trained entirely on latent codes from a frozen VAE (Section 3.1, 3.2); if the VAE distorts or collapses the latent manifold, the GAN cannot recover.
  • domain assumption CLIP text embeddings provide adequate conditioning for text-to-motion in this pipeline.
    The method uses a pre-trained CLIP-ViT-L-14 text encoder without fine-tuning on motion data (Section 3.1); if CLIP embeddings are not well aligned with motion latents, text conditioning will be weak.
  • domain assumption The evaluation metrics computed with the T2M motion encoder are reliable and directly comparable across methods.
    The paper borrows many baseline numbers from MLD and assumes the evaluation pipeline is identical (Section 4.3, Table 1 caption); any difference in preprocessing or metric code would break comparability.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LS-GAN: Human Motion Synthesis with Latent-space GANs." pith.science (2026). https://pith.science/paper/J2J3QAE6

@misc{pith2026250101449,
  author       = {Pith},
  title        = {Pith review of: LS-GAN: Human Motion Synthesis with Latent-space GANs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J2J3QAE6}},
  note         = {Machine review of arXiv:2501.01449}
}
read the original abstract

Human motion synthesis conditioned on textual input has gained significant attention in recent years due to its potential applications in various domains such as gaming, film production, and virtual reality. Conditioned Motion synthesis takes a text input and outputs a 3D motion corresponding to the text. While previous works have explored motion synthesis using raw motion data and latent space representations with diffusion models, these approaches often suffer from high training and inference times. In this paper, we introduce a novel framework that utilizes Generative Adversarial Networks (GANs) in the latent space to enable faster training and inference while achieving results comparable to those of the state-of-the-art diffusion methods. We perform experiments on the HumanML3D, HumanAct12 benchmarks and demonstrate that a remarkably simple GAN in the latent space achieves a FID of 0.482 with more than 91% in FLOPs reduction compared to latent diffusion model. Our work opens up new possibilities for efficient and high-quality motion synthesis using latent space GANs.

Figures

Figures reproduced from arXiv: 2501.01449 by the authors.

Figure 1
Figure 1. Qualitative results of text-to-motion shown by LS-GAN [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Qualitative results of text-to-motion shown by our best model (Deep WGAN-GP) [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Method overview: The Generator G maps the condi￾tional input [z, c] to a latent z ′ . The Discriminator D learns to differentiate real E(x) vs. fake motion latent z ′ . Finally at gener￾ation, we maps the learned latent z ′ to motion sequence using the decoder D(z ′ ) 3.3. GAN architectures We experiment with two different LS-GAN architectures in the latent space setting. Vanilla GAN: The Generator comprises three f… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Visualization of the t-SNE results on latent codes of LS-GAN compared to MLD. We sample 30 motions for each action label [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Qualitative results of our method shown by Vanilla GAN [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Qualitative results of our method shown by Deep GAN [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Qualitative results of our method shown by Vanilla WGAN-GP [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Qualitative results of our method shown by Deep WGAN GP [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Multimodal Generative AI with Autoregressive LLMs for Human Motion Understanding and Generation: A Way Forward

    cs.CV 2025-05 conditional novelty 3.0 of 10

    A survey paper reviews multimodal generative AI and autoregressive LLMs for text-driven human motion generation, with comparative tables of models, datasets, and metrics.

Reference graph

Works this paper leans on

55 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [37]

    Learning a bidirectional mapping between human whole- body motion and natural language using deep recurrent neu- ral networks

    Matthias Plappert, Christian Mandery, and Tamim Asfour. Learning a bidirectional mapping between human whole- body motion and natural language using deep recurrent neu- ral networks. Robotics and Autonomous Systems, 109:13–26,

    work page

  2. [3]

    Lan- guage2pose: Natural language grounded pose forecasting

    Chaitanya Ahuja and Louis-Philippe Morency. Lan- guage2pose: Natural language grounded pose forecasting. In 2019 International Conference on 3D Vision (3DV), pages 719–728. IEEE, 2019. 4, 5

    work page 2019

  3. [5]

    Text2gestures: A transformer-based network for generating emotive body gestures for virtual agents

    Uttaran Bhattacharya, Nicholas Rewkowski, Abhishek Banerjee, Pooja Guhan, Aniket Bera, and Dinesh Manocha. Text2gestures: A transformer-based network for generating emotive body gestures for virtual agents. In 2021 IEEE Vir- tual Reality and 3D User Interfaces (VR), pages 1–10. IEEE,

    work page 2021

  4. [9]

    Synthesis of compositional animations from textual descriptions

    Anindita Ghosh, Noshaba Cheema, Cennet Oguz, Christian Theobalt, and Philipp Slusallek. Synthesis of compositional animations from textual descriptions. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 1396–1406, 2021. 4, 5

    work page 2021

  5. [36]

    Black, and G ¨ul Varol

    Mathis Petrovich, Michael J. Black, and G ¨ul Varol. TEMOS: Generating diverse human motions from textual descriptions. In European Conference on Computer Vision (ECCV), 2022. 4, 5

    work page 2022

  6. [15]

    Generating diverse and natural 3d human motions from text

    Chuan Guo, Shihao Zou, Xinxin Zuo, Sen Wang, Wei Ji, Xingyu Li, and Li Cheng. Generating diverse and natural 3d human motions from text. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5152–5161, June 2022. 4, 5

    work page 2022

  7. [1]

    Text2action: Generative adversarial synthesis from language to action

    Hyemin Ahn, Timothy Ha, Yunho Choi, Hwiyeon Yoo, and Songhwai Oh. Text2action: Generative adversarial synthesis from language to action. 2018 IEEE International Confer- ence on Robotics and Automation (ICRA), pages 1–5, 2017. 3

    work page 2018

  8. [2]

    Lan- guage2pose: Natural language grounded pose forecasting

    Chaitanya Ahuja and Louis-Philippe Morency. Lan- guage2pose: Natural language grounded pose forecasting. 2019 International Conference on 3D Vision (3DV) , pages 719–728, 2019. 1

    work page 2019

Show all 55 references
  1. [4]

    Wasserstein gan, 2017

    Martin Arjovsky, Soumith Chintala, and L ´eon Bottou. Wasserstein gan, 2017. 4

    2017

  2. [6]

    Implicit neural representations for variable length human motion generation

    Pablo Cervantes, Yusuke Sekikawa, Ikuro Sato, and Koichi Shinoda. Implicit neural representations for variable length human motion generation. In European Conference on Com- puter Vision, 2022. 5

    2022

  3. [7]

    Executing your commands via motion diffusion in latent space, 2023

    Xin Chen, Biao Jiang, Wen Liu, Zilong Huang, Bin Fu, Tao Chen, Jingyi Yu, and Gang Yu. Executing your commands via motion diffusion in latent space, 2023. 1, 2, 3, 4, 5

    2023

  4. [8]

    Eduardo de Campos Valadares and Cleber P. A. Anconi. Dancing to the music. The Physics Teacher , 38:404–404, 2000

    2000

  5. [10]

    Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio

    Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial networks, 2014. 1

    2014

  6. [11]

    Improved training of wasserstein gans, 2017

    Ishaan Gulrajani, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, and Aaron Courville. Improved training of wasserstein gans, 2017. 1, 4

    2017

  7. [12]

    Tm2t: Stochastic and tokenized modeling for the reciprocal genera- tion of 3d human motions and texts

    Chuan Guo, Xinxin Xuo, Sen Wang, and Li Cheng. Tm2t: Stochastic and tokenized modeling for the reciprocal genera- tion of 3d human motions and texts. ArXiv, abs/2207.01696,

    arXiv

  8. [13]

    Generating diverse and natural 3d human motions from text

    Chuan Guo, Shihao Zou, Xinxin Zuo, Sen Wang, Wei Ji, Xingyu Li, and Li Cheng. Generating diverse and natural 3d human motions from text. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5142–5151, 2022. 1

    2022

  9. [14]

    Generating diverse and natural 3d human motions from text

    Chuan Guo, Shihao Zou, Xinxin Zuo, Sen Wang, Wei Ji, Xingyu Li, and Li Cheng. Generating diverse and natural 3d human motions from text. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 5152–5161, 2022. 1, 3

    2022

  10. [16]

    Ac- tion2motion: Conditioned generation of 3d human motions

    Chuan Guo, Xinxin Zuo, Sen Wang, Shihao Zou, Qingyao Sun, Annan Deng, Minglun Gong, and Li Cheng. Ac- tion2motion: Conditioned generation of 3d human motions. Proceedings of the 28th ACM International Conference on Multimedia, 2020. 1, 3

    2020

  11. [17]

    Deep residual learning for image recognition, 2015

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition, 2015. 3

    2015

  12. [18]

    Denoising diffu- sion probabilistic models, 2020

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models, 2020. 1

    2020

  13. [19]

    Motiongpt: Human motion as a foreign language,

    Biao Jiang, Xin Chen, Wen Liu, Jingyi Yu, Gang Yu, and Tao Chen. Motiongpt: Human motion as a foreign language,

  14. [20]

    Scaling up gans for text-to-image synthesis, 2023

    Minguk Kang, Jun-Yan Zhu, Richard Zhang, Jaesik Park, Eli Shechtman, Sylvain Paris, and Taesung Park. Scaling up gans for text-to-image synthesis, 2023. 6

    2023

  15. [21]

    Audio-driven facial animation by joint end- to-end learning of pose and emotion

    Tero Karras, Timo Aila, Samuli Laine, Antti Herva, and Jaakko Lehtinen. Audio-driven facial animation by joint end- to-end learning of pose and emotion. ACM Transactions on Graphics (TOG), 36:1 – 12, 2017. 6

    2017

  16. [22]

    A style-based generator architecture for generative adversarial networks,

    Tero Karras, Samuli Laine, and Timo Aila. A style-based generator architecture for generative adversarial networks,

  17. [23]

    Flame: Free- form language-based motion synthesis & editing

    Jihoon Kim, Jiseob Kim, and Sungjoon Choi. Flame: Free- form language-based motion synthesis & editing. In AAAI Conference on Artificial Intelligence, 2022. 1

    2022

  18. [24]

    Flame: Free- form language-based motion synthesis & editing, 2023

    Jihoon Kim, Jiseob Kim, and Sungjoon Choi. Flame: Free- form language-based motion synthesis & editing, 2023. 2

    2023

  19. [25]

    Dance- former: Music conditioned 3d dance generation with para- metric motion transformer

    Buyu Li, Yongchi Zhao, Zhelun Shi, and Lu Sheng. Dance- former: Music conditioned 3d dance generation with para- metric motion transformer. In AAAI Conference on Artificial Intelligence, 2021. 1

    2021

  20. [26]

    Ganimator: neural motion synthesis from a single sequence

    Peizhuo Li, Kfir Aberman, Zihan Zhang, Rana Hanocka, and Olga Sorkine-Hornung. Ganimator: neural motion synthesis from a single sequence. ACM Transactions on Graphics , 41(4):1–12, July 2022. 2

    2022

  21. [27]

    Ross, and Angjoo Kanazawa

    Ruilong Li, Sha Yang, David A. Ross, and Angjoo Kanazawa. Ai choreographer: Music conditioned 3d dance generation with aist++. 2021 IEEE/CVF International Con- ference on Computer Vision (ICCV) , pages 13381–13392,

    2021

  22. [28]

    Ross, and Angjoo Kanazawa

    Ruilong Li, Shan Yang, David A. Ross, and Angjoo Kanazawa. Ai choreographer: Music conditioned 3d dance generation with aist++, 2021. 2

    2021

  23. [29]

    Piano: A parametric hand bone model from magnetic reso- nance imaging

    Yuwei Li, Minye Wu, Yuyao Zhang, Lan Xu, and Jingyi Yu. Piano: A parametric hand bone model from magnetic reso- nance imaging. In International Joint Conference on Artifi- cial Intelligence, 2021. 6

    2021

  24. [30]

    A comprehensive survey on knowledge distil- lation of diffusion models, 2023

    Weijian Luo. A comprehensive survey on knowledge distil- lation of diffusion models, 2023. 6

    2023

  25. [31]

    Troje, Ger- ard Pons-Moll, and Michael J

    Naureen Mahmood, Nima Ghorbani, Nikolaus F. Troje, Ger- ard Pons-Moll, and Michael J. Black. Amass: Archive of motion capture as surface shapes. 2019 IEEE/CVF Interna- tional Conference on Computer Vision (ICCV), pages 5441– 5450, 2019. 3

    2019

  26. [32]

    Conditional generative adversarial nets, 2014

    Mehdi Mirza and Simon Osindero. Conditional generative adversarial nets, 2014. 1, 3

    2014

  27. [33]

    Black, and G ¨ul Varol

    Mathis Petrovich, Michael J. Black, and G ¨ul Varol. Action- conditioned 3d human motion synthesis with transformer vae. 2021 IEEE/CVF International Conference on Computer Vision (ICCV), pages 10965–10975, 2021. 1, 5

    2021

  28. [34]

    Black, and G ¨ul Varol

    Mathis Petrovich, Michael J. Black, and G ¨ul Varol. Temos: Generating diverse human motions from textual descriptions. ArXiv, abs/2204.14109, 2022. 1

    2022 arXiv

  29. [35]

    Black, and G ¨ul Varol

    Mathis Petrovich, Michael J. Black, and G ¨ul Varol. Temos: Generating diverse human motions from textual descriptions,

  30. [38]

    Modi: Un- conditional motion synthesis from diverse data, 2022

    Sigal Raab, Inbal Leibovitch, Peizhuo Li, Kfir Aberman, Olga Sorkine-Hornung, and Daniel Cohen-Or. Modi: Un- conditional motion synthesis from diverse data, 2022. 2

    2022

  31. [39]

    Learning transferable visual models from natural language supervision, 2021

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision, 2021. 2

    2021

  32. [40]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision. In International Con...

    2021

  33. [41]

    High-resolution image syn- thesis with latent diffusion models, 2022

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models, 2022. 2

    2022

  34. [42]

    Improved techniques for training gans, 2016

    Tim Salimans, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, and Xi Chen. Improved techniques for training gans, 2016. 6

    2016

  35. [43]

    Stylegan-t: Unlocking the power of gans for fast large-scale text-to-image synthesis, 2023

    Axel Sauer, Tero Karras, Samuli Laine, Andreas Geiger, and Timo Aila. Stylegan-t: Unlocking the power of gans for fast large-scale text-to-image synthesis, 2023. 1, 3, 6

    2023

  36. [44]

    Adversarial diffusion distillation, 2023

    Axel Sauer, Dominik Lorenz, Andreas Blattmann, and Robin Rombach. Adversarial diffusion distillation, 2023. 6

    2023

  37. [45]

    Generative adversarial net- works (gans survey): Challenges, solutions, and future direc- tions, 2023

    Divya Saxena and Jiannong Cao. Generative adversarial net- works (gans survey): Challenges, solutions, and future direc- tions, 2023. 6

    2023

  38. [46]

    Human motion gen- eration using wasserstein gan

    Ayumi Shiobara and Makoto Murakami. Human motion gen- eration using wasserstein gan. In Proceedings of the 2021 5th International Conference on Digital Signal Processing , ICDSP ’21, page 278–282, New York, NY , USA, 2021. As- sociation for Computing Machinery. 3

    2021

  39. [47]

    Bermano, and Daniel Cohen-Or

    Guy Tevet, Brian Gordon, Amir Hertz, Amit H. Bermano, and Daniel Cohen-Or. Motionclip: Exposing human motion generation to clip space, 2022. 1

    2022

  40. [49]

    Guy Tevet, Sigal Raab, Brian Gordon, Yonatan Shafir, Daniel Cohen-Or, and Amit H. Bermano. Human motion diffusion model. ArXiv, abs/2209.14916, 2022. 1, 5

    2022 arXiv

  41. [50]

    Guy Tevet, Sigal Raab, Brian Gordon, Yonatan Shafir, Daniel Cohen-Or, and Amit H. Bermano. Human motion diffusion model, 2022. 2

    2022

  42. [51]

    Neural discrete representation learning,

    Aaron van den Oord, Oriol Vinyals, and Koray Kavukcuoglu. Neural discrete representation learning,

  43. [52]

    Actformer: A gan- based transformer towards general action-conditioned 3d hu- man motion generation, 2022

    Liang Xu, Ziyang Song, Dongliang Wang, Jing Su, Zhicheng Fang, Chenjing Ding, Weihao Gan, Yichao Yan, Xin Jin, Xi- aokang Yang, Wenjun Zeng, and Wei Wu. Actformer: A gan- based transformer towards general action-conditioned 3d hu- man motion generation, 2022. 3

    2022

  44. [53]

    Freeman, and Taesung Park

    Tianwei Yin, Micha ¨el Gharbi, Richard Zhang, Eli Shecht- man, Fredo Durand, William T. Freeman, and Taesung Park. One-step diffusion with distribution matching distillation,

  45. [54]

    T2m-gpt: Generating human motion from textual de- scriptions with discrete representations, 2023

    Jianrong Zhang, Yangsong Zhang, Xiaodong Cun, Shaoli Huang, Yong Zhang, Hongwei Zhao, Hongtao Lu, and Xi Shen. T2m-gpt: Generating human motion from textual de- scriptions with discrete representations, 2023. 2

    2023

  46. [55]

    Motiondif- fuse: Text-driven human motion generation with diffusion model

    Mingyuan Zhang, Zhongang Cai, Liang Pan, Fangzhou Hong, Xinying Guo, Lei Yang, and Ziwei Liu. Motiondif- fuse: Text-driven human motion generation with diffusion model. arXiv preprint arXiv:2208.15001, 2022. 1, 2, 4

    2022 arXiv

  47. [56]

    LS-GAN qualitative results on text-to-motion: a.A person is skipping rope

    Appendix 9.1. LS-GAN qualitative results on text-to-motion: a.A person is skipping rope. b.a person doing jumping jacks. c.a person jogs straight forward. Figure 5. Qualitative results of our method shown by Vanilla GAN a.a man kicks with something or some-one with his left le...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.