Pith. sign in

REVIEW 5 major objections 6 minor 40 references

Real-time and Controllable Reactive Motion Synthesis via Intention Guidance

T0 review · 5 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a character's reactive motion can be synthesized online at about 7 ms per frame by first predicting the input character's key-joint intentions and then matching those intentions to an adversarially trained codebook…

desk verdict Plausible online reactive-motion method with clean ablations, but the headline claim against Motion Matching is contradicted by the paper's own table. read the letter →

arxiv 2507.09704 v1 pith:VJBHWPSW submitted 2025-07-13 cs.GR

Xiaotang Zhang , Ziyi Chang , Qianhui Men , Hubert Shum This is my paper · ORCID
classification cs.GR
keywords reactivemotionsynthesisintentionpredictioncodebookmatchingadversarialtrainingreal-timecharactercontrolhumaninteractionanimationcaptureuser-controlled
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 single character's reactive motion can be synthesized online, frame by frame, from only the observed history of both characters plus user-specified control, without knowing the input character's future motion. The key move is to insert an intention predictor that forecasts the future trajectories of five key joints of the input character, turning an ill-posed prediction problem into a better-conditioned one. That predicted intention is encoded into the latent space of the reactive motion and matched against a learned codebook, with adversarial training aligning the estimated and target latent distributions. If the claims hold, interactive animation systems can produce long, realistic, user-steerable interactions in real time instead of relying on precomputed full trajectories or large motion databases. The paper's ablations support the claim that the intention predictor, not just the codebook, is what stabilizes the output.

What carries the argument

The central object is the intention predictor combined with an adversarial codebook matching module. The intention predictor is a small fully-connected network that takes the current input features of both characters and outputs the predicted future trajectories of five key joints of the input character; these trajectories are represented in the local root space of the reactive character so that the model sees relative interaction rather than absolute positions. The codebook is a learned set of $c=16$ vectors of dimension $128$; the input encoder and output encoder each produce a categorical distribution over these vectors via Gumbel-Softmax, the quantized vector is decoded into the next pose, and a discriminator forces the estimated latent distribution to align with the target distribution before quantization. That adversarial alignment is what the paper claims prevents distinct interaction patterns from collapsing onto the same codebook entry, which they identify as the failure mode of plain codebook matching on two-character interaction.

What would settle it

Measure the intention predictor's error, for example MPJPE between predicted and actual future trajectories of the five key joints, at inference time on user-controlled test trajectories: if intention error grows large precisely when output quality degrades or collapses to the noisy history-only baseline, the paper's central explanation is falsified. A complementary check is to run the system on non-periodic inputs such as a standing character clapping, where the paper itself predicts trouble, and see whether the reactive character freezes or jitters beyond acceptable quality.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that reactive motion synthesis, previously treated as an offline problem in which the input character's full future is known, can be made online and controllable by predicting a compact 'intention'—the future trajectories of the head, hands, and legs of the input character in the reactor's local space—and using that prediction as an auxiliary conditioning signal for adversarial codebook matching. With this intention signal, the same historical input no longer maps to ambiguous codebook vectors, so the output poses stay coherent over long recursive generation. The method predicts the next pose stochastically by sampling a categorical distribution through Gumbel-Softmax, then decodes the sampled codebook vector into the pose, and a discriminator aligns the continuous latent space so that similar interactions map to similar regions. The authors report that this beats motion matching and plain codebook matching on MPJPE, FID, diversity, and codebook matching error, and that it keeps producing interactive motion even when user control pushes the input trajectories far from anything in the training set.

Load-bearing premise

The whole method stands on the assumption that the future motion of the input character can be predicted well enough from history alone that the predicted intention is a reliable conditioning signal; if that prediction becomes inaccurate under user steering, the system degrades to history-only codebook matching, which the paper's own ablation shows is markedly worse.

Editorial extensions

If this is right

  • Real-time interaction systems such as games, VR, and social simulation can generate reactive partner motions at about 7 ms per frame with no access to the partner's future motion.
  • Users can steer the input character's path with keyboard controls, and the reactor continues to produce plausible, non-frozen motion even for trajectories outside the training distribution.
  • Long-term synthesis is feasible by recursively feeding generated poses back as history, since the intention predictor and matcher operate on one frame at a time.
  • Stochasticity is preserved through Gumbel-Softmax temperature, so the same input can yield multiple plausible reactions while remaining stable across samples.
  • The intention predictor is the load-bearing component: removing it degrades codebook matching error and FID substantially, so future methods in this line must keep or replace that conditioning signal.

Reading between the lines

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

  • Beyond the paper: the method's success likely hinges on how predictable the input motion is; the paper's own limitation section notes that non-periodic motions like clapping cause jitter or frozen output, so an immediate testable extension is to measure intention prediction error on periodic versus non-periodic inputs.
  • Beyond the paper: because the intention predictor is trained only on ground-truth future trajectories, online errors from feeding predicted intentions back into the encoders are never measured; an inference-time evaluation of intention error under user steering would show how much of the claimed stability survives prediction drift.
  • Beyond the paper: the same intention-plus-codebook recipe could transfer to other paired interaction tasks, such as two-agent navigation, handover motions, or co-speech gesture pairs, wherever a compact future signal can be estimated from history.
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 / 6 minor

Summary. The paper proposes a real-time reactive motion synthesis method that, given the history of two interacting characters and an optional user-controlled future trajectory for the input character, predicts the next pose of the reactive character. The main ideas are an intention predictor that forecasts future trajectories of five key joints of the input character, and an adversarial codebook matching module that encodes input history plus predicted intention into a latent space, matches it to a codebook, and decodes the next pose. The method is intended to run online at about 7 ms per frame, with recursive feedback for long-term synthesis. Quantitative experiments are reported on a combination of ReMoCap, boxing, and DD100 datasets, comparing against Motion Matching and the codebook matching method of SSH*24, plus ablations that remove the intention predictor, the adversarial loss, or train the intention predictor end-to-end. User-controlled trajectory experiments are presented qualitatively.

Significance. If the claims were fully supported, the work would be a useful contribution to interactive character animation: it addresses a genuinely online setting where the reactive character cannot see the future of the input character, and it demonstrates a plausible mechanism for conditioning generation on a predicted intention while retaining real-time performance. The adaptation of categorical codebook matching to two-character interaction, the use of a relative coordinate frame, and the scheduled-sampling study are positive elements. However, the paper's central claim of superiority over matching-based approaches is not established by the reported evidence, because the strongest matching baseline dominates the proposed method on all offline metrics, and the user-controlled regime where the method is claimed to win is only evaluated qualitatively. The work is therefore promising but needs a substantial strengthening of the evaluation before the main conclusions can be accepted.

major comments (5)
  1. [Section 4.3, Table 2] The central claim that the method 'outperforms other matching-based motion synthesis approaches' is directly contradicted by the quantitative comparison with Motion Matching (MM): MM achieves lower MPJPE (1.88 vs 6.91), lower FID (0.47 vs 28.36), higher Diversity closer to ground truth (81.65 vs 61.77 against GT 85.35), and lower codebook error (0.011 vs 0.048). Since MM is the matching-based baseline used in the paper, the evidence as presented supports the opposite of the claimed superiority in the offline accuracy setting. The argument that MM lacks generalizability is deferred to qualitative controlled experiments, so the claim needs to be restricted or replaced by quantitative evidence in the setting where the method is claimed to win.
  2. [Section 4.3.2, Figures 7 and 9] The user-controlled regime, which is the main claimed advantage over Motion Matching, is evaluated only qualitatively. No quantitative metrics are reported for the proposed method or for MM and [SSH*24] under modified root trajectories, and the paper does not report error bars, perceptual studies, or any measure of physical plausibility such as foot sliding or codebook error in the controlled setting. Figure 13 shows a density plot, but it lacks axis labels and numeric comparisons. Without such evidence, the conclusion that the method 'outperforms' MM under user control is unsupported.
  3. [Section 3.2, Eqs. (4)-(5), Section 4.3.2] The intention predictor is trained with MSE against ground-truth future five-key-joint trajectories, but at inference the user modifies the future root trajectory T^{0→+1}_{X_t}. The paper never measures the intention prediction error under user-controlled inputs or how this error degrades as the controlled trajectory diverges from training data. Since the codebook encoders are conditioned on the intention feature, an inaccurate intention prediction could reduce the method to history-only codebook matching, which the ablation shows performs worse. Reporting inference-time intention error for controlled trajectories is necessary to validate the key mechanism.
  4. [Section 4.4, Table 3] The improvements attributed to the adversarial loss and to the intention predictor are small (e.g., MPJPE 6.91 vs 7.13 vs 7.79; codebook error 0.048 vs 0.054 vs 0.063) and are reported without variance across seeds or statistical significance. The paper also does not report the sensitivity of these results to the codebook capacity and loss weights, even though lambda_G/lambda_C is shown to have a sharp optimum at 500 in Figure 14. The evidence for the specific contribution of adversarial training is therefore suggestive rather than conclusive.
  5. [Section 4.4.2, Figure 13] The claim that 'Motion Matching exhibits significantly poorer codebook alignment' under user control conflicts with Table 2, where MM has the lowest codebook error (0.011 vs 0.048) in the standard evaluation. If Figure 13 is restricted to the controlled setting, this needs to be stated explicitly, and the axes and quantitative values must be provided; as written, the figure cannot support the comparison.
minor comments (6)
  1. [Section 4.1] The reference 'Figure ??' should be a numbered figure reference, and Table 1 is discussed before being introduced.
  2. [Section 3.2] The text says the data representation is 'as stated in Section 5' and the five key joints are 'as described in Section 5', but both are described in Section 3.1.
  3. [Section 4.4.1, Eq. (15)] The scheduled sampling probability p=e/E increases with epoch, so early training uses ground truth and later training uses predictions; the direction of the schedule is unusual and should be stated explicitly, since standard scheduled sampling typically decreases teacher forcing over time.
  4. [Table 2] The table header contains the typo 'Diversiy' instead of 'Diversity'.
  5. [Section 4.3] The paper says 'throughout' where 'thorough' is intended, and in Section 4.3.1 'lays' should be 'lies'.
  6. [Section 5] The limitation about periodic motions is useful, but it also implies the method's performance depends heavily on dataset regularity; the paper should connect this limitation to the claimed generalizability under user control.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the intention predictor is supervised against ground-truth future trajectories and the reactive synthesis is trained against ground-truth outputs with external baselines and public datasets.

full rationale

The paper's derivation chain is not circular. The intention predictor (Eq. 4) is supervised by MSE against ground-truth future joint trajectories (Eq. 5), so the "intention" is an externally supervised quantity rather than a fitted quantity that is then renamed as a prediction. The reactive pose synthesis is trained by a reconstruction loss (Eq. 10) against ground-truth outputs, and the codebook and adversarial objectives (Eqs. 11-14) are standard training losses evaluated against external baselines (Motion Matching, SSH24) and public datasets (ReMoCap, DD100, synthetic boxing). No equation reduces to its own input by construction, and no load-bearing claim depends on an unverified self-citation. The self-citations ([MSHL22], [SKY07], [SKY10], [CKCS25], [CWKS25], [CFZS23]) appear in the literature review or as dataset sources and do not carry the central argument. The evaluation caveat that FID and Diversity are computed in the encoder's latent space, while adversarial training aligns that space, is a metric-validity concern rather than a circular derivation; it does not make the reported losses or predictions equivalent to the inputs. Concerns about Motion Matching outperforming the method on Table 2 and the absence of quantitative user-control metrics are correctness and evidence issues, not circularity.

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

The central claim rests on supervised learning assumptions rather than mathematical derivation. The main free parameters are loss weights and codebook capacity, selected empirically. The method assumes key-joint future trajectories capture intention, that training datasets cover interactive motion, and that relative-coordinate codebook matching can represent two-character interactions. No invented physical entities are introduced.

free parameters (4)
  • GAN-to-codebook loss weight ratio lambda_G / lambda_C = 500
    Selected in Section 4.4.2 by sweeping codebook matching error; no separate held-out validation protocol is described, and the same metric is later reported as an evaluation result.
  • Codebook capacity (channels c=16, dimension d=128) = 2048 codes
    Architectural choice increased from SSH24 because the model encodes two characters with 52 joints; not derived from data.
  • Hidden layer dimensions for intention predictor and codebook matching = 512 and 2048
    Chosen by hand in Section 4.1; no ablation of these values is reported.
  • Scheduled sampling probability p = e/E = epoch fraction
    Used in Eq. 15 for end-to-end training; an ad hoc schedule that affects the end-to-end ablation results.
assumptions (4)
  • domain assumption Future joint trajectories of five key joints represent the interaction intention of the input character.
    Section 3.2 states that the dynamics of the five joints are representative of overall body movement; if false, intention guidance cannot disambiguate reactions.
  • domain assumption The training datasets (ReMoCap, boxing, DD100) cover the interactive motion space needed for real-time generalization, including user-controlled trajectory edits.
    Section 4.1 trains on these sets, and Section 4.3.2 expects the model to generalize to user paths outside the data distribution.
  • domain assumption Relative coordinates in the reactor's root space are a learnable and sufficient representation for two-character interaction.
    Section 3.3.1 and Figure 5 justify this representation, but the paper itself notes the increased diversity makes codebook coverage difficult.
  • standard math Gumbel-Softmax with straight-through estimator provides a valid differentiable approximation to categorical sampling.
    Invoked in Eq. 8 and Section 3.3.1; a standard reparameterization technique, not a contribution of this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Real-time and Controllable Reactive Motion Synthesis via Intention Guidance." pith.science (2026). https://pith.science/paper/VJBHWPSW

@misc{pith2026250709704,
  author       = {Pith},
  title        = {Pith review of: Real-time and Controllable Reactive Motion Synthesis via Intention Guidance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VJBHWPSW}},
  note         = {Machine review of arXiv:2507.09704}
}
read the original abstract

We propose a real-time method for reactive motion synthesis based on the known trajectory of input character, predicting instant reactions using only historical, user-controlled motions. Our method handles the uncertainty of future movements by introducing an intention predictor, which forecasts key joint intentions to make pose prediction more deterministic from the historical interaction. The intention is later encoded into the latent space of its reactive motion, matched with a codebook which represents mappings between input and output. It samples a categorical distribution for pose generation and strengthens model robustness through adversarial training. Unlike previous offline approaches, the system can recursively generate intentions and reactive motions using feedback from earlier steps, enabling real-time, long-term realistic interactive synthesis. Both quantitative and qualitative experiments show our approach outperforms other matching-based motion synthesis approaches, delivering superior stability and generalizability. In our method, user can also actively influence the outcome by controlling the moving directions, creating a personalized interaction path that deviates from predefined trajectories.

Figures

Figures reproduced from arXiv: 2507.09704 by the authors.

Figure 1
Figure 1. Our model generates instant reactive movements for the red character based on anticipated intentions. The user controls the movement direction of the input blue character, and our system dynamically generates corresponding motions for both characters in response. Abstract We propose a real-time method for reactive motion synthesis based on the known trajectory of input character, predicting instant reactions using o… view at source ↗
Figure 2
Figure 2. The overall pipeline of our intention-aware reactive motion synthesis system [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. During inference, the predicted intention will be encoded and mapped to a trained latent space from which the Output De￾coder predicts the next pose. modeling was also essential. [CTO∗ 23] was a pioneering work in this area, employing an attention-based network to model both tem￾poral and spatial dependencies. [GDG∗ 24] utilized diffusion mod￾els and introduced a reaction loss to enhance prediction accuracy. Similar… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Data representation of input and output. The black lines behind characters represent their historical joints and root trajec￾tories, while the lines in front of the character are the future trajec￾tories. 3.1. Representation We represent the conditioned input character…
Figure 5
Figure 5. Figure 5: Visual comparison of different trajectory representations. The left example shows future trajectories represented in the char￾acter’s own root space, while the right example uses the reactor’s root space as the reference frame. Accurate interaction prediction in our ta…
Figure 6
Figure 6. Figure 6: Architecture details of the adversarial codebook match￾ing. and dimension d = 128 as the size of codebook. The number of codebook channels is increased compared to the original setting in [SSH∗ 24], as our model encodes the concatenated motion fea￾tures of two interact…
Figure 7
Figure 7. Figure 7: Left: Performance using original motion from the testing dataset as input. Middle and Right: Performance using controlled motion (red and green trajectories) as input [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: For better visualization, we remove character’s mesh to showcase the variation of different samples. Left: Our model pro￾duces stable predictions across all samples. Right: [SSH∗ 24] tends to generate noisy results due to the high uncertainty in predicting the input ch…
Figure 10
Figure 10. Figure 10: The black lines with orange dots are the predicted in￾tentions. Left: Performance with intention predictor. Right: Perfor￾mance without intention predictor [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 11
Figure 11. Figure 11: By adjusting the temperature of the Gumbel-Softmax sampling, the model can generate non-deterministic reactive poses (red skeletons) at each step from the same input conditions. at early training stage, we employ scheduled sampling to help the network adapt to its own…
Figure 12
Figure 12. Figure 12: T-SNE results on latent spaces. For better visualization, we only choose 1000 frames for calculation. Blue: Target latent space. Red: Estimated latent space. Left: Result of using adversar￾ial training. Right: Result without adversarial training [PITH_FULL_IMAGE:figu…
Figure 13
Figure 13. Figure 13: Density of codebook matching error. Adversarial train￾ing helps to increase robustness of codebook matching in user￾controlled environment while Motion Matching lacks generalizabil￾ity in this case. in Figures 12 and 13 [PITH_FULL_IMAGE:figures/full_fig_p009_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 37 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTIO...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry.original add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 i...

  3. [3]

    Chang Z., Findlay E. J. C., Zhang H., Shum H. P. H. : Unifying human motion synthesis and style transfer with denoising diffusion probabilistic models. In Proceedings of the 2023 International Conference on Computer Graphics Theory and Applications (Lisbon, Portugal, 2 2023), GRAPP '23, SciTePress, pp. 64--74

    work page 2023

  4. [4]

    A., Chang H

    Chang Z., Koulieris G. A., Chang H. J., Shum H. P. : On the design fundamentals of diffusion models: A survey. Pattern Recognition (2025), 111934

    work page 2025

  5. [5]

    : Ready-to-react: Online reaction policy for two-character interaction generation

    Cen Z., Pi H., Peng S., Shuai Q., Shen Y., Bao H., Zhou X., Hu R. : Ready-to-react: Online reaction policy for two-character interaction generation. In The Thirteenth International Conference on Learning Representations (Singapore EXPO, Singapore)

    work page

  6. [6]

    : Interaction transformer for human reaction generation

    Chopin B., Tang H., Otberdout N., Daoudi M., Sebe N. : Interaction transformer for human reaction generation. IEEE Transactions on Multimedia 25 (2023), 8842--8854

    work page 2023

  7. [7]

    B., van de Panne M

    Cohan S., Tevet G., Reda D., Peng X. B., van de Panne M. : Flexible motion in-betweening with diffusion models. In ACM SIGGRAPH 2024 Conference Papers (Colorado Convention Center, Denver, CO, USA, 2024), ACM SIGGRAPH, pp. 1--9

    work page 2024

  8. [8]

    Large-Scale Multi-Character Interaction Synthesis

    Chang Z., Wang H., Koulieris G. A., Shum H. P. : Large-scale multi-character interaction synthesis. arXiv preprint arXiv:2505.14087 (2025)

    work page Pith review arXiv 2025

Show all 40 references
  1. [9]

    : Remos: 3d motion-conditioned reaction synthesis for two-person interactions

    Ghosh A., Dabral R., Golyanik V., Theobalt C., Slusallek P. : Remos: 3d motion-conditioned reaction synthesis for two-person interactions. In European Conference on Computer Vision (ECCV) (Milan, Italy, 2024), Springer

    2024

  2. [10]

    G., Wang S., Cheng L

    Guo C., Mu Y., Javed M. G., Wang S., Cheng L. : Momask: Generative masked modeling of 3d human motions. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (Seattle Convention Center, Seattle, WA, USA, 2024), IEEE/CVF, pp. 1900--1910

    2024

  3. [11]

    E., Alexanderson S., Beskow J

    Henter G. E., Alexanderson S., Beskow J. : Moglow: Probabilistic and controllable motion synthesis using normalising flows. ACM Transactions on Graphics (TOG) 39, 6 (2020), 1--14

    2020

  4. [12]

    : Learned motion matching

    Holden D., Kanoun O., Perepichka M., Popa T. : Learned motion matching. ACM Transactions on Graphics (TOG) 39, 4 (2020), 53--1

    2020

  5. [13]

    : Phase-functioned neural networks for character control

    Holden D., Komura T., Saito J. : Phase-functioned neural networks for character control. ACM Transactions on Graphics (TOG) 36, 4 (2017), 1--13

    2017

  6. [14]

    : Gans trained by a two time-scale update rule converge to a local nash equilibrium

    Heusel M., Ramsauer H., Unterthiner T., Nessler B., Hochreiter S. : Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems 30 (2017)

    2017

  7. [15]

    : A deep learning framework for character motion synthesis and editing

    Holden D., Saito J., Komura T. : A deep learning framework for character motion synthesis and editing. ACM Transactions on Graphics (TOG) 35, 4 (2016), 1--11

    2016

  8. [16]

    : Learning motion manifolds with convolutional autoencoders

    Holden D., Saito J., Komura T., Joyce T. : Learning motion manifolds with convolutional autoencoders. In SIGGRAPH Asia 2015 technical briefs. ACM, 2015, pp. 1--4

    2015

  9. [17]

    : Categorical reparameterization with gumbel-softmax

    Jang E., Gu S., Poole B. : Categorical reparameterization with gumbel-softmax. In International Conference on Learning Representations (Toulon, France, 2017)

    2017

  10. [18]

    : Decoupled weight decay regularization

    Loshchilov I., Hutter F. : Decoupled weight decay regularization. In International Conference on Learning Representations (New Orleans, Louisiana, USA, 2019)

    2019

  11. [19]

    : Questenvsim: Environment-aware simulated motion tracking from sparse sensors

    Lee S., Starke S., Ye Y., Won J., Winkler A. : Questenvsim: Environment-aware simulated motion tracking from sparse sensors. In ACM SIGGRAPH 2023 Conference Proceedings (Los Angeles Convention Center, Los Angeles, CA, USA, 2023), ACM SIGGRAPH, pp. 1--9

    2023

  12. [20]

    : Intergen: Diffusion-based multi-human motion generation under complex interactions

    Liang H., Zhang W., Li W., Yu J., Xu L. : Intergen: Diffusion-based multi-human motion generation under complex interactions. International Journal of Computer Vision (2024), 1--21

    2024

  13. [21]

    https://www.mixamo.com/

    Mixamo : Mixamo. https://www.mixamo.com/. Accessed: 2024-09-23

    2024

  14. [22]

    P., Ho E

    Men Q., Shum H. P., Ho E. S., Leung H. : Gan-based reactive motion synthesis with class-aware discriminators for human--human interaction. Computers & Graphics 102 (2022), 634--645

    2022

  15. [23]

    : Motion matching in the last of us part ii, 2020

    Mach M., Zhuravlov M. : Motion matching in the last of us part ii, 2020. Accessed: 29-Sep-2024

    2020

  16. [24]

    : Trajectory-aware body interaction transformer for multi-person pose forecasting

    Peng X., Mao S., Wu Z. : Trajectory-aware body interaction transformer for multi-person pose forecasting. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (Vancouver Convention Center, Vancouver, BC, Canada, 2023), IEEE/CVF, pp. 17121--17130

    2023

  17. [25]

    : Combining motion matching and orientation prediction to animate avatars for consumer-grade vr devices

    Ponton J., Yun H., Andujar C., Pelechano N. : Combining motion matching and orientation prediction to animate avatars for consumer-grade vr devices

  18. [26]

    Siyao L., Gu T., Yang Z., Lin Z., Liu Z., Ding H., Yang L., Loy C. C. : Duolando: Follower gpt with off-policy reinforcement learning for dance accompaniment. arXiv preprint arXiv:2403.18811 (2024)

    2024 arXiv

  19. [27]

    Shum H. P. H., Komura T., Yamazaki S. : Simulating competitive interactions using singly captured motions. In Proceedings of the 2007 ACM Symposium on Virtual Reality Software and Technology (New York, NY, USA, 11 2007), VRST '07, ACM, pp. 65--72

    2007

  20. [28]

    P., Komura T., Yamazaki S

    Shum H. P., Komura T., Yamazaki S. : Simulating multiple character interactions with collaborative and adversarial goals. IEEE Transactions on Visualization and Computer Graphics 18, 5 (2010), 741--752

    2010

  21. [29]

    : Deepphase: Periodic autoencoders for learning motion phase manifolds

    Starke S., Mason I., Komura T. : Deepphase: Periodic autoencoders for learning motion phase manifolds. ACM Transactions on Graphics (TOG) 41, 4 (2022), 1--13

    2022

  22. [30]

    : Categorical codebook matching for embodied character controllers

    Starke S., Starke P., He N., Komura T., Ye Y. : Categorical codebook matching for embodied character controllers. ACM Transactions on Graphics (TOG) 43, 4 (2024), 1--14

    2024

  23. [31]

    Shafir Y., Tevet G., Kapon R., Bermano A. H. : Human motion diffusion as a generative prior. In The Twelfth International Conference on Learning Representations (Vienna, Austria, 2024)

    2024

  24. [32]

    : Local motion phases for learning multi-contact character movements

    Starke S., Zhao Y., Komura T., Zaman K. : Local motion phases for learning multi-contact character movements. ACM Transactions on Graphics (TOG) 39, 4 (2020), 54--1

    2020

  25. [33]

    : Role-aware interaction generation from textual description

    Tanaka M., Fujiwara K. : Role-aware interaction generation from textual description. In Proceedings of the IEEE/CVF international conference on computer vision (Paris, France, 2023), IEEE/CVF, pp. 15999--16009

    2023

  26. [34]

    : Social diffusion: Long-term multiple human motion anticipation

    Tanke J., Zhang L., Zhao A., Tang C., Cai Y., Wang L., Wu P.-C., Gall J., Keskin C. : Social diffusion: Long-term multiple human motion anticipation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (Paris, France, 2023), IEEE/CVF, pp. 9601--9611

    2023

  27. [35]

    : Multi-person 3d motion prediction with multi-range transformers

    Wang J., Xu H., Narasimhan M., Wang X. : Multi-person 3d motion prediction with multi-range transformers. Advances in Neural Information Processing Systems 34 (2021), 6036--6049

    2021

  28. [36]

    : Stochastic multi-person 3d motion forecasting

    Xu S., Wang Y.-X., Gui L. : Stochastic multi-person 3d motion forecasting. In The Eleventh International Conference on Learning Representations (Kigali, Rwanda, 2023), ICLR

    2023

  29. [37]

    : Regennet: Towards human action-reaction synthesis

    Xu L., Zhou Y., Yan Y., Jin X., Zhu W., Rao F., Yang X., Zeng W. : Regennet: Towards human action-reaction synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (Seattle Convention Center, Seattle, WA, USA, 2024), IEEE/CVF, pp. 1759--1769

    2024

  30. [38]

    : Robots learn social skills: End-to-end learning of co-speech gesture generation for humanoid robots

    Yoon Y., Ko W.-R., Jang M., Lee J., Kim J., Lee G. : Robots learn social skills: End-to-end learning of co-speech gesture generation for humanoid robots. In 2019 International Conference on Robotics and Automation (ICRA) (Montreal, QC, Canada, 2019), IEEE, IEEE, pp. 4303--4309

    2019

  31. [39]

    : Diffusion models: A comprehensive survey of methods and applications

    Yang L., Zhang Z., Song Y., Hong S., Xu R., Zhao Y., Zhang W., Cui B., Yang M.-H. : Diffusion models: A comprehensive survey of methods and applications. ACM Computing Surveys 56, 4 (2023), 1--39

    2023

  32. [40]

    : Mode-adaptive neural networks for quadruped motion control

    Zhang H., Starke S., Komura T., Saito J. : Mode-adaptive neural networks for quadruped motion control. ACM Transactions on Graphics (TOG) 37, 4 (2018), 1--11

    2018

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.