Pith. sign in

REVIEW 5 major objections 7 minor 23 references

Learning Complex Non-Rigid Image Edits from Multimodal Conditioning

T0 review · 5 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims that fine-tuning a Stable Diffusion inpainting model on filtered video-frame pairs with GPT-4V-generated scene-difference captions and 2D pose conditioning yields identity-preserving, text-controllable non-rigid edits in…

desk verdict Solid engineering and a valuable public-data baseline, but the paper's headline claim—that text improves person-object interactions—is contradicted by its own Table 3. read the letter →

arxiv 2412.10219 v1 pith:QZ6UGCGM submitted 2024-12-13 cs.CV

Nikolai Warner , Jack Kolb , Meera Hahn , Vighnesh Birodkar , Jonathan Huang , Irfan Essa This is my paper · ORCID
classification cs.CV
keywords diffusionmodelsimageeditingidentitypreservationnon-rigidposeeditsconditioningtextmultimodalcaptionsperson-objectinteraction
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

The paper tries to establish that a diffusion model can insert a single photo of a person into a new scene and change that person's pose in ways controlled by a short text prompt, while keeping the person's identity intact. The training signal is weak and cheap: pairs of frames from human-action videos, with the pose change between frames described automatically by a multimodal LLM rather than by human annotators. The authors claim that adding 2D pose skeletons alongside these noisy text captions materially improves the model's handling of person-object interactions, which they identify as the hardest case of identity preservation. If the claim holds, user-controllable, identity-preserving image editing can be learned without expensive paired annotation, extending to scenes outside the training distribution. The paper also contributes an open-source dataset of 13,487 captioned video-frame pairs and reimplements the leading prior baseline on public data.

What carries the argument

The key mechanism is joint multimodal conditioning in an inpainting diffusion framework. A masked target frame provides the scene to insert into; a segmented reference crop carries identity through CLIP-image cross-attention; a GPT-4V-generated scene-difference caption enters through the CLIP text encoder; and a 17-joint 2D skeleton, flattened to 51 values and projected through a linear layer, carries pose. These three embedding streams are concatenated into one (batch, 335, 768) tensor that conditions the U-Net, and classifier-free guidance is implemented with separate unconditional representations for image, text, and pose. The other load-bearing component is the data pipeline: keyframes are sampled per video using a minimum pose distance (shoulder-to-head length) and histogram-similarity bounds, and RTMPose is used to filter scenes with a single full skeleton and to produce the conditioning poses.

What would settle it

If, in a held-out test set of pairs whose captions describe a leftward or rightward movement, the model consistently mirrors the opposite direction (following its captioning errors), the claim that it learned the text-to-edit mapping from the noisy captions would be falsified; equivalently, a larger preregistered user study that finds no significant difference between image-pose and image-pose-text on object-interaction plausibility would falsify the central thesis that pose+text improves person-object interactions.

Watch

Extended reading notes

Core claim

The central discovery is that multimodal conditioning—a reference image of the person, a text caption describing the pose change, and a 2D pose skeleton—can be fused in a Stable Diffusion inpainting model fine-tuned on self-supervised video-frame pairs, and that this fusion yields identity-preserving, text-controllable non-rigid edits on in-the-wild images. On scenes without object interactions, the combined image-pose-text model achieves the highest identity-preservation rating (68.5%) and strong control-signal adherence (51%) in a user study, at the cost of slightly lower FID than image-only conditioning. On scenes with person-object interactions, where identity preservation rates drop to 25-50%, pose and text jointly produce the most plausible interactions (33%), confirming the paper's thesis that weak caption supervision plus robust pose conditioning improves person-object interaction quality. The paper also documents the limits: identity preservation is not always achieved, long narrow objects are brittle, and automatically generated captions contain left/right and ordering errors.

Load-bearing premise

The training pairs from the same video actually show the same person performing a plausible pose change, so that the masked target, reference crop, and auto-generated caption describe one coherent edit; if filters let through identity switches or captions that reverse the motion direction, the learned text-to-edit mapping is corrupted.

Editorial extensions

If this is right

  • A user can take an unseen photo of a person and a target scene, and request multiple different non-rigid edits (turn, lunge, raise arms) by changing only the text prompt, without retraining or per-edit tuning.
  • Weakly supervised video-frame pairs suffice to train this controllability: 5,787 captioned Kinetics videos plus annotated pairs from NTU-RGBD, Charades, and Fit3D produce the identity-preserving edits, meaning the approach can scale to unlabeled video.
  • The best identity-preservation and control adherence come from combining reference image, text, and pose; pose alone or text alone is worse, establishing the joint-conditioning recipe as the method's core.
  • FID is an imperfect proxy for this task: the image-only model wins on FID yet performs worse on identity and control in user studies, arguing for control-aware evaluation of editing models.
  • Open-sourcing the captioned dataset and the public reimplementation lets other methods be compared on identity-preserving non-rigid editing without the closed data used by prior work.

Reading between the lines

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

  • If the left/right and reverse-order caption errors are the main noise source, then improving caption fidelity (for example, by providing frame order explicitly to the LLM, or by filtering captions against the measured pose delta) could push identity preservation and control adherence above the reported 68.5% and 51%.
  • The same joint conditioning might transfer to articulated non-human subjects (animals, robots) as long as a pose skeleton and paired frames are available, making the method a general recipe for controllable subject insertion.
  • The authors hypothesize that more accurate captions would improve the image-text model; a testable extension is to use the pose delta to reject captions that disagree with the measured motion, which could raise interaction plausibility beyond the current 33% ceiling.
  • The method's dependence on a single pretrained captioner suggests a self-training loop: generate an edit, caption the difference with the same LLM, and add the pair back into the training set—whether such a loop converges or amplifies errors is untested.
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 / 7 minor

Summary. This paper addresses the task of inserting a person from a reference image into a novel scene with controllable non-rigid edits. The authors fine-tune a Stable Diffusion inpainting model on pairs of video frames, using GPT-4V-generated scene-difference captions and RTMPose 2D skeletons as additional conditioning signals. They compare four configurations (image-only, image-text, image-pose, image-pose-text) using FID, PCKh, and a human user study, and they reimplement the Kulal et al. baseline on public data. The paper also contributes a processed dataset and caption annotations.

Significance. If the central claim were established, this would be a useful step: combining weak text supervision with pose conditioning for identity-preserving, non-rigid human insertion, plus a public-data reimplementation of a previously closed-source method. The paper has clear strengths: it reimplements Kulal et al. on public data, documents captioning failures honestly, and provides qualitative evidence that text and pose together enable controllability. However, the headline claim that weak captions plus pose improves person-object interaction quality is not supported by the paper's own quantitative table, and the user study is not reported with enough statistical detail to distinguish the observed differences from noise. The contribution is therefore promising but not yet demonstrated at the level claimed.

major comments (5)
  1. [§4.4, Table 3] The abstract and conclusion claim that combining weak captions with pose improves the quality of person-object interactions, but Table 3 reports interaction plausibility of 33% for both Img-Pose and Img-Pose-Text, with Img-Text at 24.5% and Img-Only at 24%. The only measurable gain over the image-only baseline comes from pose conditioning; text adds zero marginal improvement to interaction plausibility. The identity-preservation gain of Img-Pose-Text (41%) over Img-Pose (25%) is a different result and should not be presented as an interaction-quality improvement. The central claim needs to be either supported with a statistically significant contrast between Img-Pose-Text and Img-Pose on interaction plausibility, or revised to state that pose drives interaction quality while text improves identity preservation.
  2. [§4.1, §4.4] The user study is not reported with enough statistical detail to support the quantitative claims. Only eight raters and 50 scenes per configuration are used, answers are binary, and no confidence intervals, significance tests, or inter-rater agreement measures are given. With these numbers, a difference of 33% versus 24% (roughly 4-5 binary responses out of 50) can easily be within sampling noise. The authors should report per-rater variation, compute confidence intervals or a paired significance test, and ideally increase the number of scenes or raters for the key interaction comparison.
  3. [§4.3, Figure 3] The paper claims superior performance compared to baselines such as ControlNet, MASACtrl, and PIDM, but the actual comparisons to these baselines are qualitative only. There is no quantitative evaluation of these baselines on the same scenes, and no user-study ratings for them. To substantiate the third contribution, the authors should evaluate at least one state-of-the-art baseline under the same identity-preservation and controllability protocol used in Tables 2 and 3.
  4. [§3.3, A.3] The captioning pipeline is a central contribution, but its error rate is never quantified. The appendix documents left/right reversals, reversed ordering, and occasional misidentification of objects, yet no human evaluation of caption accuracy or an analysis of how caption errors affect controllability is provided. A small human-annotated subset with caption correctness statistics would strengthen the claim that weak captions are usable and would let readers assess the effect of caption noise.
  5. [§3.2, §3.4] The frame-pair filtering criteria are described qualitatively: minimum pose distance is said to be equivalent to the length of the pose skeleton's shoulder-to-head distance, and histogram similarity is said to have minimum and maximum thresholds, but no numerical values are reported. Since the data-quality argument is load-bearing for the method, the exact thresholds and a sensitivity analysis should be reported for reproducibility.
minor comments (7)
  1. [§2.3] The sentence 'These methods maintain subject identities, however are non-rigid edits' appears to be missing a negation; the intended meaning is likely that these methods are not non-rigid edits.
  2. [§4.4] The text says 'Responses are averaged for 4 users each into the tables below,' but earlier it says eight raters were recruited; clarify whether each scene was rated by four or eight raters.
  3. [Appendix A.5] The reference to 'PIDM [ ?]' is an unresolved placeholder and should be replaced with the actual reference [3].
  4. [Table 2] The caption states that 'Adding text or pose to image conditioning improves identity preservation,' but the Img-Text row shows identity at 55%, which is lower than the Img-Only value of 61%; adjust the caption to match the numbers.
  5. [Figure 3 caption] The caption uses 'our's' where 'ours' is intended.
  6. [§4.6] The statement that 'image-text and image-pose models achieve the best results' conflicts with the user-study tables, where image-pose-text is highest on identity in both Table 2 and Table 3; rephrase for clarity.
  7. [General] The paper states that the dataset is open-source, but no URL or release instructions are provided; an availability statement should be added.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: text and pose conditioning are externally sourced and edit quality is externally evaluated, so the central claim does not reduce to its own inputs.

full rationale

The paper's derivation chain is an empirical finetuning pipeline, not a formal derivation whose conclusion is encoded in its premises. The conditioning signals come from independent external models: GPT-4V generates scene-difference captions (Section 3.3), RTMPose extracts 2D pose skeletons (Section 3.4), and video frame pairs provide the image supervision. The reported outcomes are measured by human raters on held-out scenes in Tables 2 and 3 and by FID/PCKh in Table 1; none of these metrics is defined in terms of the fitted conditioning inputs. A generated edit is not forced by its caption or pose by construction, and the model is not evaluated on its own training captions. The re-implementation of Kulal et al. on public data is a baseline reproduction, not a self-citation chain, and no uniqueness theorem or author-imported assumption is invoked to make the approach forced. The main internal concern is that the abstract's claim that adding text to pose improves person-object interactions is not supported by Table 3, where Img-Pose and Img-Pose-Text both score 33% on interactions; however, that is an evidence or correctness objection, not circularity. No load-bearing step reduces a prediction to a fitted input or to a self-citation, so the appropriate finding is no significant circularity.

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

The learning system is built on several inherited components: pretrained Stable Diffusion, CLIP, RTMPose, and GPT-4V. The central claim treats their outputs as reliable enough for supervision. No new physical or conceptual entities are introduced.

free parameters (3)
  • Minimum pose distance threshold (shoulder-to-head length) = not specified numerically
    Used in Section 3.4 to select video frame pairs with meaningful motion; the threshold is chosen by hand and changes the training distribution.
  • Frame histogram similarity min/max thresholds = not specified
    Used in Section 3.4 to keep background consistent while avoiding redundant frames; the values are not reported and are selected by hand.
  • Unconditional pose image for classifier-free guidance = one neutral standing pose selected from the dataset
    Defined in Section 3.6 as the null pose embedding; the choice of a single neutral image is ad hoc and not ablated.
assumptions (5)
  • domain assumption Stable Diffusion inpainting checkpoint and CLIP embeddings provide a sufficiently strong prior for photorealistic generation and identity preservation after finetuning.
    Invoked throughout Sections 3.5 and 3.6; the entire method inherits the quality and biases of these pretrained components.
  • domain assumption Paired frames from the same video, after filtering, show the same person in a plausible pose transition.
    Used in Sections 3.2 and 3.4 to create training pairs; if identity switches or implausible jumps survive filtering, supervision is invalid.
  • domain assumption GPT-4V scene-difference captions, though noisy, contain enough signal about relative pose changes to supervise text conditioning.
    Used in Section 3.3 and Appendix Figure 3; the paper documents caption failures including left/right reversal and reverse order.
  • domain assumption RTMPose 2D 17-joint skeletons capture the pose information needed for controllability and person-object interaction.
    Used in Sections 3.5, 4, and 4.2; pose conditioning depends entirely on the accuracy of this external detector.
  • domain assumption Human binary ratings with 8 raters and 50 scenes per configuration are a valid proxy for identity preservation, controllability, and interaction plausibility.
    Used in Section 4.1; the conclusions rely on this evaluation without confidence intervals, significance tests, or inter-rater agreement.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning Complex Non-Rigid Image Edits from Multimodal Conditioning." pith.science (2026). https://pith.science/paper/QZ6UGCGM

@misc{pith2026241210219,
  author       = {Pith},
  title        = {Pith review of: Learning Complex Non-Rigid Image Edits from Multimodal Conditioning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QZ6UGCGM}},
  note         = {Machine review of arXiv:2412.10219}
}
read the original abstract

In this paper we focus on inserting a given human (specifically, a single image of a person) into a novel scene. Our method, which builds on top of Stable Diffusion, yields natural looking images while being highly controllable with text and pose. To accomplish this we need to train on pairs of images, the first a reference image with the person, the second a "target image" showing the same person (with a different pose and possibly in a different background). Additionally we require a text caption describing the new pose relative to that in the reference image. In this paper we present a novel dataset following this criteria, which we create using pairs of frames from human-centric and action-rich videos and employing a multimodal LLM to automatically summarize the difference in human pose for the text captions. We demonstrate that identity preservation is a more challenging task in scenes "in-the-wild", and especially scenes where there is an interaction between persons and objects. Combining the weak supervision from noisy captions, with robust 2D pose improves the quality of person-object interactions.

Figures

Figures reproduced from arXiv: 2412.10219 by the authors.

Figure 1
Figure 1. Given a single image, multiple controllable identity [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Complex edits are achievable through weakly annotated [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Comparison of our approach to baselines for identity preservation and controllability of in-the-wild images. The input image [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: System diagram illustrating the process of generating a desired edit using multiple inputs including noise target latent, binary [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Using pose conditioning is insufficient to specify person [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 1
Figure 1. Figure 1: Our data encompasses a variety of action [PITH_FULL_IMAGE:figures/full_fig_p011_1.png]
Figure 3
Figure 3. Figure 3: Our supervision is noisy, yet we achieve [PITH_FULL_IMAGE:figures/full_fig_p012_3.png]
Figure 4
Figure 4. Figure 4: Pose conditioning works well in various scenarios. Visualization of pose adherence for the image-pose model. The model generally adheres well to pose conditioning but can struggle with complex scenes, backgrounds, and many object-person inter￾actions. Larger filtered d…
Figure 5
Figure 5. Figure 5: Failure cases of the PIDM [?] model on ”In-The-Wild” data across various poses and input images. Rows 1 and 3 represent different pose varia￾tions. Our model, in contrast, is trained to maintain identity preservation and generate plausible, control￾lable results in mor…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 12 canonical work pages

  1. [16]

    Efros, and Karsten Kreis Singh

    Siddhesh Kulal, Tim Brooks, Alex Aiken, Jiajun Wu, Jianwei Yang, Jiajun Lu, Alexei A. Efros, and Karsten Kreis Singh. Putting people in their place: Affordance-aware human insertion into scenes. arXiv preprint arXiv:2304.14406, 2023. 2, 4, 5, 6, 7

    arXiv 2023

  2. [1]

    Assessing phenotypic correlation through the multivariate phylogenetic latent liability model

    Mykhaylo Andriluka, Leonid Pishchulin, Peter Gehler, and Bernt Schiele. 2d human pose estimation: New be nchmark and state of the art analysis. arXiv preprint arXiv:1406.3863, 2014. 6

    work page Pith review arXiv 2014

  3. [2]

    Identity-preserving aging of face images via latent diffusion models

    Sudipta Banerjee, Govind Mittal, Ameya Joshi, Chin- may Hegde, and Nasir Memon. Identity-preserving aging of face images via latent diffusion models. In 2023 IEEE International Joint Conference on Biomet- rics (IJCB), pages 1–8. IEEE, 2023. 4

    work page 2023

  4. [4]

    Tim Brooks and Alexei A. Efros. Hallucinating pose- compatible scenes. arXiv preprint arXiv:2112.07933,

    work page Pith review arXiv

  5. [5]

    Tim Brooks, Aleksander Holynski, and Alexei A. Efros. Instructpix2pix: Learning to follow image editing in- structions. arXiv preprint arXiv:2211.09800 , 2022. 5

    arXiv 2022

  6. [6]

    Sparks of artificial general intelli- gence: Early experiments with gpt-4

    Sébastien Bubeck, Varun Chandrasekaran, Ronen El- dan, Johannes Gehrke, Eric Horvitz, Ece Kamar, Pe- ter Lee, Yin Tat Lee, Yuanzhi Li, Scott Lundberg, Harsha Nori, Hamid Palangi, Marco Tulio Ribeiro, and Yi Zhang. Sparks of artificial general intelli- gence: Early experiments with gpt-4. arXiv preprint arXiv:2303.12712, 2023. 5

    arXiv 2023

  7. [7]

    Physics-informed radial basis network (PIRBN): A local approximating neural network for solving nonlinear PDEs

    Mingdeng Cao, Xintao Wang, Zhongang Qi, Ying Shan, Xiaohu Qie, and Yinqiang Zheng. Masac- trl: Tuning-free mutual self-attention control for con- sistent image synthesis and editing. arXiv preprint arXiv:2304.06234, 2023. 3, 7

    work page Pith review arXiv 2023

  8. [8]

    Patchy particles by self-assembly of star copolymers on a spherical substrate: Thomson solutions in a geometric problem with a color constraint

    Joao Carreira, Eric Noland, Chloe Hillier, and Andrew Zisserman. A short note on the kinetics-700 human action dataset. arXiv preprint arXiv:1907.08103, 2019. 4

    work page Pith review arXiv 1907

Show all 23 references
  1. [9]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at...

    2010 arXiv

  2. [10]

    Aifit: Au- tomatic 3d human-interpretable feedback models for fitness training

    Mihai Fieraru, Mihai Zanfir, Silviu-Cristian Pirlea, Vlad Olaru, and Cristian Sminchisescu. Aifit: Au- tomatic 3d human-interpretable feedback models for fitness training. In The IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR) , June

  3. [12]

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

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilibrium. arXiv preprint arXiv:1706.08500, 2017. 6

    2017 arXiv

  4. [13]

    Classifier-free diffu- sion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffu- sion guidance. arXiv preprint arXiv:2207.12598, 2022. 6

    2022 arXiv

  5. [14]

    Rtmpose: Real- time multi-person pose estimation based on mmpose

    Tao Jiang, Peng Lu, Li Zhang, Ningsheng Ma, Rui Han, Chengqi Lyu, Yining Li, and Kai Chen. Rtmpose: Real- time multi-person pose estimation based on mmpose. arXiv preprint arXiv:2303.06244, 2023. 5

    2023 arXiv

  6. [15]

    Imagic: Text-based real image editing with diffu- sion models

    Bahjat Kawar, Shiran Zada, Oran Lang, Omer Tov, Huiwen Chang, Tali Dekel, Inbar Mosseri, and Michal Irani. Imagic: Text-based real image editing with diffu- sion models. arXiv preprint arXiv:2210.09276, 2022. 3

    2022 arXiv

  7. [17]

    Deepfashion: Powering robust clothes recognition and retrieval with rich annotations

    Ziwei Liu, Ping Luo, Shi Qiu, Xiaogang Wang, and Xiaoou Tang. Deepfashion: Powering robust clothes recognition and retrieval with rich annotations. In Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , pages 1096–1104. IEEE, 2016. 3

    2016

  8. [18]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. arXiv preprint arXiv:1505.04597, 2015. 2

    2015 arXiv

  9. [19]

    Ntu rgb+d: A large scale dataset for 3d human activity analysis

    Amir Shahroudy, Jun Liu, Tian-Tsong Ng, and Gang Wang. Ntu rgb+d: A large scale dataset for 3d human activity analysis. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. 5

    2016

  10. [20]

    Advancing pose-guided image synthesis with progressive conditional diffusion models

    Fei Shen, Hu Ye, Jun Zhang, Cong Wang, Xiao Han, and Wei Yang. Advancing pose-guided image synthesis with progressive conditional diffusion models. arXiv preprint arXiv:2310.08563, 2023. 3

    2023 arXiv

  11. [21]

    In- stantid: Zero-shot identity-preserving generation in seconds

    Qixun Wang, Xu Bai, Haofan Wang, Zekui Qin, An- thony Chen, Huaxia Li, Xu Tang, and Yao Hu. In- stantid: Zero-shot identity-preserving generation in seconds. arXiv preprint arXiv:2401.07519, 2024. 4

    2024 arXiv

  12. [22]

    Temporal reasoning graph for activity recogni- tion

    Jingran Zhang, Fumin Shen, Xing Xu, and Heng Tao Shen. Temporal reasoning graph for activity recogni- tion. In arXiv preprint arXiv:1908.08648, 2019. Pre- sented at the Computer Vision and Pattern Recognition Conference (CVPR). 5

    1908 arXiv

  13. [23]

    Dreambooth: Fine tun- ing text-to-image diffusion models for subject-driven generation

    Lvmin Zhang, Qingyang Chen, Chengyi Chen, Yeping Huang, Zhe Gan, and Li Ma. Dreambooth: Fine tun- ing text-to-image diffusion models for subject-driven generation. arXiv preprint arXiv:2208.12242, 2022. 2 9

    2022 arXiv

  14. [24]

    Adding conditional control to text-to-image diffusion models (controlnet)

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models (controlnet). arXiv preprint arXiv:2302.05543,

    arXiv

  15. [2023]

    3 10 Scene Ref Target Pose Det Figure 1: Our data encompasses a variety of action classes and scenes with and without object interac- tions from Kinetics. We sample a few key frames per video, according to the minimum pose distance criteria described in Methods, which allows u...

    2024 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.