Pith. sign in

REVIEW 3 major objections 4 minor 106 references

LogiShot: Logically Coherent Cross-Shot Video Generation

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read LogiShot generates video clips that follow the previous shot's logic while keeping characters and objects consistent.

desk verdict LogiShot's central claim is credible and the architecture is a real step forward, but the evaluation needs code/data release and a quality-score-stratified analysis before the numbers are fully trusted. read the letter →

arxiv 2608.08820 v1 pith:W3IFDF3T submitted 2026-08-09 cs.CV

Shuai Guo , Yuhang Yang , Zeyu Zhang , Pengfei Yu , Wei Zhai , Yang Cao , Zheng-Jun Zha This is my paper · ORCID
classification cs.CV
keywords cross-shotvideogenerationlogicalcoherencevisualconsistencymultimodalcueguidancememorydiffusiontransformervision-languagemodelreasoning
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 argues that generating a video clip that follows a previous one fails when each clip is produced from isolated text or a reference image, because underspecified prompts leave the intended logical relation ambiguous. Its proposed system, LogiShot, conditions a video diffusion transformer on the context video, the user prompt, and a starting frame through two complementary paths: dense multimodal cues from a frozen vision-language model, and a latent visual memory of the context video that stays accessible during generation. The authors construct a 110,000-sample dataset with a dedicated benchmark for cross-shot logical coherence and report that LogiShot outperforms three baselines on both Logical Correctness and Visual Consistency. If correct, the paper establishes that the context video itself can supply the relational evidence that otherwise has to be spelled out in exhaustive per-shot scripts.

What carries the argument

The central mechanism is a pair of complementary conditioning paths inside a video diffusion transformer. Multimodal Cue Guidance (MCG) takes the final-layer hidden states of a frozen vision-language model that jointly processes the context video, the prompt, and the starting frame, projects them into the transformer's text-conditioning space, and concatenates them with the decoded target-event description, so generation receives visual-semantic evidence that the text alone does not carry. Visual Memory (VM) constructs latent slots from uniformly sampled context-video frames, prepends them to the noisy target-video sequence with presence masks, and lets target tokens retrieve from them in every self-attention block; the starting frame is also injected via its latent and CLIP features. Together, MCG supplies the relational semantics and VM preserves the context's visual details.

What would settle it

If a model given the prompt, the starting frame, and the decoded target-event description but not the context video matched LogiShot's Logical Correctness on the same 800-sample held-out set, the claim that context-video conditioning drives logical coherence would be falsified.

Watch

Extended reading notes

Core claim

LogiShot's central claim is that cross-shot logical coherence requires jointly establishing the logical relation between shots and preserving visual consistency, and that both can be achieved by conditioning a video diffusion transformer on the context video rather than only on a decoded description. The method feeds the context video, the prompt, and the starting frame into a frozen vision-language model, projects the model's final-layer hidden states into dense multimodal cues, and concatenates those cues with the decoded target-event description to form the transformer's conditioning sequence. Separately, uniformly sampled context-video frames are embedded as latent slots prepended to the noisy target sequence, so target tokens can attend to them in every self-attention block. The paper reports consistent gains over three baselines, with Logical Correctness improvements of 0.075-0.119 and Visual Consistency improvements of 0.041-0.087, and ablation experiments attribute the gains to the two mechanisms being complementary.

Load-bearing premise

The dataset's guarantee that each target event is inferable from the context video, the prompt, and the starting frame jointly and only from them must hold; if the automatic judge accepts pairs whose target event is deducible from the prompt alone, the reported logical-coherence advantage is inflated.

Editorial extensions

If this is right

  • Users can give underspecified instructions such as "based on the detective's reasoning, have him point to the likely culprit" and the model resolves the intended event from the context video rather than requiring a fully detailed script.
  • Because the relation types include Progression, Parallel, Causal, Conditional, and Overview/Detail, the same architecture can drive both earlier-shot and later-shot generation.
  • Visual consistency no longer depends on supplying explicit reference images or character sheets; the context-video latent prefix acts as an implicit memory.
  • The largest Logical Correctness gains appear on Conditional and Parallel relations, the cases that demand reasoning across shots rather than simple chronological continuation.
  • The released 110K-sample dataset and benchmark give the field a shared resource for measuring cross-shot logical coherence.

Reading between the lines

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

  • A natural stress test the paper leaves implicit: replace the matched context video with an unrelated one while keeping the prompt and starting frame fixed; the paper's shuffled-cue ablation already suggests Logical Correctness should drop, which would confirm the cues are context-specific rather than generic.
  • The two-path conditioning recipe is portable: any video diffusion transformer could prepend a context-video latent prefix and accept projected VLM hidden states, so the approach generalizes beyond cinema to instruction, simulation, and embodied prediction.
  • If the judge's inferability labels are trustworthy at scale, the dataset becomes reusable for next-event prediction, a neighbouring task that usually lacks large grounded context-target pairs.
  • An implicit consequence is that the method's ceiling is the vision-language model's ability to infer the target event; when the VLM misreads the relation, the dense cues propagate that error into the generated video.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. LogiShot proposes a method for cross-shot video generation that takes as input a context video, a prompt instruction, and a starting frame, and generates a target video that is logically related to the context. The method has two mechanisms: Multimodal Cue Guidance (MCG), which extracts dense multimodal cues from a frozen VLM's hidden states and concatenates them with the decoded target-event description for DiT cross-attention, and Visual Memory (VM), which retains context-video latent slots as a prefix in the DiT's self-attention. The paper also describes construction of a 110K-sample dataset with a three-stage filtering pipeline, including VLM-based prompt verification and two human audits, and an 800-sample benchmark evaluated by a VLM judge using six sub-metrics for Logical Correctness and Visual Consistency. Experiments report consistent gains over three adapted baselines, ablations of MCG and VM, a user study, and additional analyses of the judge's behavior.

Significance. If the reported results are robust, the paper's main contribution is a well-specified architecture for conditioning a video DiT on both continuous VLM representations and context-video latents, together with a large curated dataset for cross-shot logical coherence. The paper deserves credit for including a shuffled-cue ablation (Table 10) that directly tests whether the dense cues carry context-specific information, for reporting per-method judge residuals (Table 9) to examine method-specific bias, for running human audits and a judge-validation study (Table 6), and for providing a relation-type breakdown (Fig. 6). However, the evaluation rests heavily on a single VLM judge, and the benchmark averages over samples without conditioning on the quality scores that the pipeline itself records. Because the dataset's defining property is that the target event is inferable only from the joint inputs, and the post-verification audit reports a 7.3% target-event inferability failure rate on the final dataset, the magnitude of the claimed advantages over baselines is not yet fully established.

major comments (3)
  1. [Sec. 5.1 / A.1 / Table 6] The 800-sample benchmark in Sec. 5.1 averages Logical Correctness and Visual Consistency over all samples without conditioning on the quality score assigned by the prompt-verification judge, even though the paper's own post-verification audit reports a 7.3% target-event inferability failure rate and a 10.7% composite failure rate on the final dataset. Since the dataset's defining property is that the target event is inferable only from the triple (context video, prompt, starting frame), a material subset of benchmark samples may not satisfy this property; if those samples are concentrated in low-certainty cases, the reported headline gaps could be inflated by exactly the cases where the evaluation judge's rubric is least reliable. Please stratify the main results by recorded quality score (e.g., score 5 versus score 4) and report results after excluding samples flagged by the audit protocol, and report judge-human agreement conditional on quality score.
  2. [Sec. 5.1] The claim that all 18 sub-metric differences are statistically significant under paired randomization tests after Holm correction (p<0.01) is not supported by any details of the test procedure. The reader cannot tell whether the pairing is by sample, how many permutations were run, what test statistic was used, or whether the deterministic VLM judge's scores were treated as fixed. Please provide the full test protocol, including the test statistic and the number of resamples, in the appendix; without this information, the significance claim is not checkable.
  3. [Sec. 5.1 / C.2 / Table 9] The evaluation judge is validated on only 150 cases (30 per relation type), with Spearman's rho of 0.68 for Logical Correctness and 0.59 for Visual Consistency, and it exhibits a systematic positive offset relative to humans (Table 9). Because the same VLM family is used for dataset filtering (A.1) and for evaluation, there is a residual risk that the criterion 'target event inferable from the triple' is enforced and then measured by models sharing the same bias. The shuffled-cue ablation in Table 10 rules out the dense cues being a generic signal, but it does not test whether inference difficulty is comparable across samples. As a concrete check, please measure human-judge agreement separately on score-5 and score-4 subsets and report the human-verified inferability rate specifically for the 800 evaluation samples rather than for the whole dataset.
minor comments (4)
  1. [Table 6] The first row of Table 6 is malformed ('0.936 0.7520.7160.71'); the values for Rubric A, Rubric B, composite rate, and Fleiss' kappa need to be separated and clearly labeled.
  2. [Appendix B.1] The implementation details omit several quantities needed for reproducibility, including the classifier-free guidance scale, number of training steps, batch size per GPU, and total training compute; please add these.
  3. [Fig. 6] The per-relation-type gap figure annotates each bar with the strongest baseline, which can be confusing because the baseline varies across bars; please provide a full table of per-relation-type scores for all methods.
  4. [Sec. 3 / Appendix A.1] Please state explicitly which VLM is used in the data-construction pipeline (the main text only names Qwen2.5-VL in the baseline adaptations), and clarify whether it is the same model family as the evaluation judge.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: LogiShot's derivation is self-contained, and the reported gains are not reduced to the paper's own inputs.

full rationale

The paper's central derivation is not circular. LogiShot is trained with a rectified-flow objective (Eq. 4) to reconstruct the ground-truth target video from the context video, prompt, and starting frame, with no parameter fitted to the Logical Correctness or Visual Consistency metrics; the metrics are applied equally to all methods after generation. The prompt-verification judge used in data construction defines the dataset property that target events be inferable from the three inputs jointly, but this is a dataset filter, not an input to the generator's objective, and the same filter equally constrains all baselines. The MCG and VM contributions are tested by ablations (Tab. 2) that retain the decoded description and starting frame, and by a shuffled-cue control (Tab. 10) showing the dense cues carry context-specific information rather than generic signal. The evaluation is supported by judge-independent human ratings (Tab. 7) and a blind pairwise user study (Fig. 8), so the headline ordering is not forced by the judge. Self-citations in related work and the VBench diagnostics are not load-bearing; no uniqueness theorem or prior result by the same authors is invoked to exclude alternatives. The reliance of the evaluation on a VLM judge of the same family as the data filter is a statistical validity concern, not a circularity: no equation or fitted parameter reduces a reported result to its own input. Therefore no circular step is exhibited.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical or mathematical entities; it introduces a new dataset, a benchmark, and two learned conditioning mechanisms. The main free parameters are the context-frame count K, the flow shift, and the projector designs. The axioms are domain assumptions about the validity of VLM-based judging and the decomposition of logical coherence into the six metrics. The most load-bearing assumption is that the VLM judge and the human audits correctly enforce inferability from all three inputs jointly.

free parameters (3)
  • K (number of context frames in VM prefix) = 4
    Chosen by the authors as 'K=4 frames uniformly sampled from V' (Appendix B.1). No ablation over K is reported, so it is a hand-chosen design parameter that affects the visual memory capacity.
  • Rectified-flow shift factor = 5
    Set in Appendix B.1 as the shift factor for the flow-matching schedule. This is a standard hyperparameter but still a hand-set value that affects training dynamics.
  • Projector MLP architectures (MCG projector and CLIP projector) = two-layer MLP with GELU
    The projection layers are learned, and their capacity and hidden sizes are not fully specified; they are free design choices that influence how much VLM information reaches the DiT.
assumptions (4)
  • domain assumption The VLM judge's five-point rubric reliably measures whether a target event is inferable from the three inputs jointly and not from any subset.
    The dataset quality gate and the evaluation metrics both depend on this judge. It is validated against human consensus on 500 and 150 cases, with agreement 91.6% and kappa=0.84/0.76, so it is partially supported but not guaranteed.
  • domain assumption Logical coherence between two video clips can be decomposed into the six sub-metrics (Event Match, Relation Match, Completion, Participant Match, Role Match, State Match).
    The evaluation claims to measure logical coherence through these sub-metrics, but the paper does not validate that this decomposition captures all aspects of narrative logic. This is a modeling assumption about the evaluation target.
  • domain assumption The frozen VLM (Qwen2.5-VL family implied) jointly processes video, prompt, and starting frame accurately enough that its hidden states contain context-specific visual-semantic evidence.
    The MCG mechanism assumes the VLM's final-layer hidden states retain the relevant information. The shuffled-cue ablation supports this, but the VLM itself is taken as a reliable reasoner.
  • domain assumption The Wan-family 14B DiT and video VAE are adequate backbones such that freezing them and training only the denoiser and projections is sufficient.
    The whole method is built on a specific pretrained backbone; the paper assumes its inductive biases and that fine-tuning the denoiser is enough to learn cross-shot reasoning.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LogiShot: Logically Coherent Cross-Shot Video Generation." pith.science (2026). https://pith.science/paper/W3IFDF3T

@misc{pith2026260808820,
  author       = {Pith},
  title        = {Pith review of: LogiShot: Logically Coherent Cross-Shot Video Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W3IFDF3T}},
  note         = {Machine review of arXiv:2608.08820}
}
read the original abstract

Generating cross-shot videos that are logically connected is essential for content creation. Currently, most cross-shot video-generation workflows, such as short-drama production, still rely on isolated textual scripts or explicit reference images to specify the generated content. Consequently, when user instructions are underspecified or ambiguous, a generated clip may appear visually plausible on its own but fail to align with the overall narrative, leading to disjointed content. We argue that achieving cross-shot logical coherence in video generation requires establishing logical connections across shots and maintaining visual consistency. To this end, we propose LogiShot, which incorporates information through two complementary paths: 1) LogiShot jointly encodes the context video and other conditioning signals, yielding dense multimodal cues that provide visual-semantic evidence for cross-shot generation; 2) the model maintains a visual memory of the context video throughout generation to preserve visual consistency across shots. Additionally, we construct a dataset with 110K samples and a dedicated benchmark for evaluating cross-shot logical coherence. Experiments demonstrate that LogiShot consistently outperforms existing baselines in terms of logical coherence across multiple shots. Model and data will be made publicly available.

Figures

Figures reproduced from arXiv: 2608.08820 by the authors.

Figure 1
Figure 1. Logically coherent cross-shot generation. Given the context video, the input prompt, and a starting frame, our model [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Data construction pipeline. Left: The three-stage pipeline that constructs the dataset from raw videos: Shot Extraction [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. LogiShot architecture. Multimodal Cue Guidance (MCG; Sec. 4.1) augments a target-event description with dense [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison against baselines. Red boxes mark baseline failures. On the left, the baselines send the wrong [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Qualitative ablation of MCG and VM. The W/O MCG variant assigns the unlock action to the wrong actor (left), while the W/O VM variant fails to preserve the teddy bear’s appearance (right). Mechanism Logical Correctness ↑ Visual Consistency ↑ MCG VM EvMt RelMt Comp Part…
Figure 7
Figure 7. Figure 7: Attention over the context-video prefix. Pink over [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 6
Figure 6. Figure 6: Per-relation-type gap. The gap between LogiShot [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 4
Figure 4. Figure 4: In these examples, the baselines assign actions to [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 8
Figure 8. Figure 8: User study win rates. Pairwise A/B win rates for [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Failure case. The prompt relates the room before [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

106 extracted references · 49 canonical work pages

  1. [1]

    International Conference on Learning Representations , volume=

    Cogvideox: Text-to-video diffusion models with an expert transformer , author=. International Conference on Learning Representations , volume=

  2. [4]

    2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=

    Phantom: Subject-consistent video generation via cross-modal alignment , author=. 2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=. 2025 , organization=

    2025

  3. [5]

    International Conference on Learning Representations , volume=

    Univideo: Unified understanding, generation, and editing for videos , author=. International Conference on Learning Representations , volume=

  4. [8]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Video-as-answer: Predict and generate next video event with joint-grpo , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  5. [9]

    2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Vbench: Comprehensive benchmark suite for video generative models , author=. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=. 2024 , organization=

    2024

  6. [10]

    2021 IEEE/CVF international conference on computer vision (ICCV) , pages=

    Emerging properties in self-supervised vision transformers , author=. 2021 IEEE/CVF international conference on computer vision (ICCV) , pages=. 2021 , organization=

    2021

  7. [11]

    2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Amt: All-pairs multi-field transforms for efficient frame interpolation , author=. 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=. 2023 , organization=

    2023

  8. [13]

    2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Next-qa: Next phase of question-answering to explaining temporal actions , author=. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=. 2021 , organization=

    2021

Show all 106 references
  1. [14]

    European Conference on Computer Vision , pages=

    Videostudio: Generating consistent-content and multi-scene videos , author=. European Conference on Computer Vision , pages=. 2024 , organization=

    2024

  2. [15]

    Proceedings of the 2020 conference on empirical methods in natural language processing (EMNLP) , pages=

    What is more likely to happen next? video-and-language future event prediction , author=. Proceedings of the 2020 conference on empirical methods in natural language processing (EMNLP) , pages=

    2020

  3. [17]

    2023 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=

    Intentqa: Context-aware video intent reasoning , author=. 2023 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=. 2023 , organization=

    2023

  4. [18]

    2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Morevqa: Exploring modular reasoning models for video question answering , author=. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=. 2024 , organization=

    2024

  5. [20]

    2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Identity-preserving text-to-video generation by frequency decomposition , author=. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=. 2025 , organization=

    2025

  6. [21]

    2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Vlogger: Make your dream a vlog , author=. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=. 2024 , organization=

    2024

  7. [24]

    Advances in Neural Information Processing Systems , volume=

    Video-r1: Reinforcing video reasoning in mllms , author=. Advances in Neural Information Processing Systems , volume=

  8. [26]

    International Conference on Learning Representations , volume=

    Fostering video reasoning via next-event prediction , author=. International Conference on Learning Representations , volume=

  9. [28]

    Advances in neural information processing systems , volume=

    MECD: Unlocking multi-event causal discovery in video reasoning , author=. Advances in neural information processing systems , volume=

  10. [29]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    VidEvent: A Large Dataset for Understanding Dynamic Evolution of Events in Videos , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  11. [32]

    2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=

    Videoauteur: Towards long narrative video generation , author=. 2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=. 2025 , organization=

    2025

  12. [33]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Holocine: Holistic generation of cinematic multi-shot long video narratives , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  13. [34]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Long context tuning for video generation , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  14. [35]

    Proceedings of the SIGGRAPH Asia 2025 Conference Papers , pages=

    Cut2next: Generating next shot via in-context tuning , author=. Proceedings of the SIGGRAPH Asia 2025 Conference Papers , pages=

    2025

  15. [36]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Onestory: Coherent multi-shot video generation with adaptive memory , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  16. [38]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Stage: Storyboard-anchored generation for cinematic multi-shot narrative , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  17. [39]

    2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Videobooth: Diffusion-based video generation with image prompts , author=. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=. 2024 , organization=

    2024

  18. [40]

    European Conference on Computer Vision , pages=

    Magic-me: Identity-specific video customized diffusion , author=. European Conference on Computer Vision , pages=. 2024 , organization=

    2024

  19. [41]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    What happens next? next scene prediction with a unified video model , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  20. [45]

    2023 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=

    Scalable diffusion models with transformers , author=. 2023 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=. 2023 , organization=

    2023

  21. [46]

    International conference on machine learning , pages=

    Learning transferable visual models from natural language supervision , author=. International conference on machine learning , pages=. 2021 , organization=

    2021

  22. [47]

    Shuai Bai and Keqin Chen and Xuejing Liu and Jialin Wang and Wenbin Ge and Sibo Song and Kai Dang and Peng Wang and Shijie Wang and Jun Tang and others , journal =

  23. [48]

    2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=

    Hpsv3: Towards wide-spectrum human preference score , author=. 2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=. 2025 , organization=

    2025

  24. [55]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Gloria: Consistent Character Video Generation via Content Anchors , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  25. [57]

    2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=

    Sigman: Scaling 3d human gaussian generation with millions of assets , author=. 2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=. 2025 , organization=

    2025

  26. [59]

    2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=

    Hero: Human reaction generation from videos , author=. 2025 IEEE/CVF International Conference on Computer Vision (ICCV) , pages=. 2025 , organization=

    2025

  27. [61]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Grounding 3d object affordance from 2d interactions in images , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  28. [62]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Lemon: Learning 3d human-object interaction relation from 2d images , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  29. [65]

    Advances in Neural Information Processing Systems , volume=

    Egochoir: Capturing 3d human-object interaction regions from egocentric views , author=. Advances in Neural Information Processing Systems , volume=

  30. [66]

    International Conference on Learning Representations , volume=

    Touch: Text-guided controllable generation of free-form hand-object interactions , author=. International Conference on Learning Representations , volume=

  31. [67]

    2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=

    Great: Geometry-intention collaborative inference for open-vocabulary 3d object affordance grounding , author=. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages=. 2025 , organization=

    2025

  32. [68]

    An, Z.; Jia, M.; Qiu, H.; Zhou, Z.; Huang, X.; Liu, Z.; Ren, W.; Kahatapitiya, K.; Liu, D.; He, S.; et al. 2026. Onestory: Coherent multi-shot video generation with adaptive memory. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 16173--16184

    2026

  33. [69]

    Ayyubi, H.; Liu, J.; Asgarov, A.; Hakim, Z. I. A.; Sarker, N. H.; Wang, Z.; Tang, C.-W.; Alomari, H.; Atabuzzaman, M.; Lin, X.; et al. 2025. Enter: Event based interpretable reasoning for videoqa. arXiv preprint arXiv:2501.14194

    2025 arXiv

  34. [70]

    Bai, S.; Chen, K.; Liu, X.; Wang, J.; Ge, W.; Song, S.; Dang, K.; Wang, P.; Wang, S.; Tang, J.; et al. 2025. Qwen2.5-VL Technical Report. arXiv preprint arXiv:2502.13923

    2025 arXiv

  35. [71]

    Caron, M.; Touvron, H.; Misra, I.; J \'e gou, H.; Mairal, J.; Bojanowski, P.; and Joulin, A. 2021. Emerging properties in self-supervised vision transformers. In 2021 IEEE/CVF international conference on computer vision (ICCV), 9630--9640. IEEE

    2021

  36. [72]

    Chen, J.; Chen, M.; Xu, J.; Li, X.; Dong, J.; Sun, M.; Jiang, P.; Li, H.; Yang, Y.; Zhao, H.; et al. 2025. Dancetogether! identity-preserving multi-person interactive video generation. arXiv preprint arXiv:2505.18078

    2025 arXiv

  37. [73]

    Chen, T.; Liu, H.; He, T.; Chen, Y.; Gan, C.; Ma, X.; Zhong, C.; Zhang, Y.; Wang, Y.; Lin, H.; et al. 2024. MECD: Unlocking multi-event causal discovery in video reasoning. Advances in neural information processing systems, 37: 92554--92580

    2024

  38. [74]

    Cheng, J.; Hou, L.; Tao, X.; and Liao, J. 2026. Video-as-answer: Predict and generate next video event with joint-grpo. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 38915--38925

    2026

  39. [75]

    Fei, H.; Wu, S.; Ji, W.; Zhang, H.; Zhang, M.; Lee, M.-L.; and Hsu, W. 2024. Video-of-thought: Step-by-step video reasoning from perception to cognition. arXiv preprint arXiv:2501.03230

    2024 arXiv

  40. [76]

    Feng, K.; Gong, K.; Li, B.; Guo, Z.; Wang, Y.; Peng, T.; Wu, J.; Zhang, X.; Wang, B.; and Yue, X. 2026. Video-r1: Reinforcing video reasoning in mllms. Advances in Neural Information Processing Systems, 38: 99114--99137

    2026

  41. [77]

    Guo, Y.; Yang, C.; Yang, Z.; Ma, Z.; Lin, Z.; Yang, Z.; Lin, D.; and Jiang, L. 2025. Long context tuning for video generation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 17281--17291

    2025

  42. [78]

    Han, G.; Zhai, W.; Yang, Y.; Cao, Y.; and Zha, Z.-J. 2026. Touch: Text-guided controllable generation of free-form hand-object interactions. In International Conference on Learning Representations, volume 2026, 79185--79210

    2026

  43. [79]

    He, J.; Liu, H.; Li, J.; Huang, Z.; Yu, Q.; Ouyang, W.; and Liu, Z. 2025. Cut2next: Generating next shot via in-context tuning. In Proceedings of the SIGGRAPH Asia 2025 Conference Papers, 1--11

    2025

  44. [80]

    Hendrycks, D.; and Gimpel, K. 2016. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415

    2016 arXiv

  45. [81]

    Huang, Z.; He, Y.; Yu, J.; Zhang, F.; Si, C.; Jiang, Y.; Zhang, Y.; Wu, T.; Jin, Q.; Chanpaisit, N.; et al. 2024. Vbench: Comprehensive benchmark suite for video generative models. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 21807--21818. IEEE

    2024

  46. [82]

    Huang, Z.; Yu, N.; Chen, G.; Qiu, H.; Debevec, P.; and Liu, Z. 2025. Vchain: Chain-of-visual-thought for reasoning in video generation. arXiv preprint arXiv:2510.05094

    2025 arXiv

  47. [83]

    C.; and Liu, Z

    Jiang, Y.; Wu, T.; Yang, S.; Si, C.; Lin, D.; Qiao, Y.; Loy, C. C.; and Liu, Z. 2024. Videobooth: Diffusion-based video generation with image prompts. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 6689--6700. IEEE

    2024

  48. [84]

    P.; and Ba, J

    Kingma, D. P.; and Ba, J. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

    2014 arXiv

  49. [85]

    Kong, W.; Tian, Q.; Zhang, Z.; Min, R.; Dai, Z.; Zhou, J.; Xiong, J.; Li, X.; Wu, B.; Zhang, J.; et al. 2024. Hunyuanvideo: A systematic framework for large video generative models. arXiv preprint arXiv:2412.03603

    2024 arXiv

  50. [86]

    Lei, J.; Yu, L.; Berg, T.; and Bansal, M. 2020. What is more likely to happen next? video-and-language future event prediction. In Proceedings of the 2020 conference on empirical methods in natural language processing (EMNLP), 8769--8784

    2020

  51. [87]

    Li, H.; Li, Y.; Yang, Y.; Cao, J.; Zhu, Z.; Cheng, X.; and Chen, L. 2024. Dispose: Disentangling pose guidance for controllable human image animation. arXiv preprint arXiv:2412.09349

    2024 arXiv

  52. [88]

    Li, J.; Wei, P.; Han, W.; and Fan, L. 2023 a . Intentqa: Context-aware video intent reasoning. In 2023 IEEE/CVF International Conference on Computer Vision (ICCV), 11929--11940. IEEE

    2023

  53. [89]

    Li, X.; Chen, Z.; Zhao, R.; Schiffers, F.; Liao, Z.; and Bhat, V. 2026. What happens next? next scene prediction with a unified video model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 5693--5703

    2026

  54. [90]

    Li, Z.; Zhu, Z.-L.; Han, L.-H.; Hou, Q.; Guo, C.-L.; and Cheng, M.-M. 2023 b . Amt: All-pairs multi-field transforms for efficient frame interpolation. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 9801--9810. IEEE

    2023

  55. [91]

    Liang, B.; Su, Q.; Zhu, S.; Liang, Y.; and Tong, C. 2025. VidEvent: A Large Dataset for Understanding Dynamic Evolution of Events in Videos. Proceedings of the AAAI Conference on Artificial Intelligence, 39(5): 5128--5136

    2025

  56. [92]

    Liang, J.; Jiang, S.; Wang, Z.; Pan, H.; Chen, Z.; Chu, Z.; Liu, M.; Fu, R.; Wang, Z.; and Qin, B. 2024. GUIDE: a guideline-guided dataset for instructional video comprehension. arXiv preprint arXiv:2406.18227

    2024 arXiv

  57. [93]

    Lin, H.; Zala, A.; Cho, J.; and Bansal, M. 2023. Videodirectorgpt: Consistent multi-scene video generation via llm-guided planning. arXiv preprint arXiv:2309.15091

    2023 arXiv

  58. [94]

    T.; Ben-Hamu, H.; Nickel, M.; and Le, M

    Lipman, Y.; Chen, R. T.; Ben-Hamu, H.; Nickel, M.; and Le, M. 2022. Flow matching for generative modeling. arXiv preprint arXiv:2210.02747

    2022 arXiv

  59. [95]

    Liu, L.; Ma, T.; Li, B.; Chen, Z.; Liu, J.; Li, G.; Zhou, S.; He, Q.; and Wu, X. 2025. Phantom: Subject-consistent video generation via cross-modal alignment. In 2025 IEEE/CVF International Conference on Computer Vision (ICCV), 14951--14961. IEEE

    2025

  60. [96]

    Liu, X.; Gong, C.; and Liu, Q. 2022. Flow straight and fast: Learning to generate and transfer data with rectified flow. arXiv preprint arXiv:2209.03003

    2022 arXiv

  61. [97]

    Long, F.; Qiu, Z.; Yao, T.; and Mei, T. 2024. Videostudio: Generating consistent-content and multi-scene videos. In European Conference on Computer Vision, 468--485. Springer

    2024

  62. [98]

    Luo, J.; Lin, J.; Zhang, Z.; Wu, B.; Fang, M.; Chen, L.; and Tang, H. 2025. Univid: The open-source unified video model. arXiv preprint arXiv:2509.24200

    2025

  63. [99]

    Luo, Y.; Shi, X.; Zhuang, J.; Chen, Y.; Liu, Q.; Wang, X.; Wan, P.; and Xue, T. 2026. Shotstream: Streaming multi-shot video generation for interactive storytelling. arXiv preprint arXiv:2603.25746

    2026

  64. [100]

    Ma, Y.; Wu, X.; Sun, K.; and Li, H. 2025. Hpsv3: Towards wide-spectrum human preference score. In 2025 IEEE/CVF International Conference on Computer Vision (ICCV), 15086--15095. IEEE

    2025

  65. [101]

    Ma, Z.; Zhou, D.; Wang, X.-S.; Yeh, C.-H.; Li, X.; Yang, H.; Dong, Z.; Keutzer, K.; and Feng, J. 2024. Magic-me: Identity-specific video customized diffusion. In European Conference on Computer Vision, 19--37. Springer

    2024

  66. [102]

    L.; Wang, H.; Ma, S.; Li, Y.; Chen, C.; et al

    Meng, Y.; Ouyang, H.; Yu, Y.; Wang, Q.; Wang, W.; Cheng, K. L.; Wang, H.; Ma, S.; Li, Y.; Chen, C.; et al. 2026. Holocine: Holistic generation of cinematic multi-shot long video narratives. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 461--471

    2026

  67. [103]

    Min, J.; Buch, S.; Nagrani, A.; Cho, M.; and Schmid, C. 2024. Morevqa: Exploring modular reasoning models for video question answering. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 13235--13245. IEEE

    2024

  68. [104]

    Peebles, W.; and Xie, S. 2023. Scalable diffusion models with transformers. In 2023 IEEE/CVF International Conference on Computer Vision (ICCV), 4172--4182. IEEE

    2023

  69. [105]

    Polyak, A.; Zohar, A.; Brown, A.; Tjandra, A.; Sinha, A.; Lee, A.; Vyas, A.; Shi, B.; Ma, C.-Y.; Chuang, C.-Y.; et al. 2024. Movie gen: A cast of media foundation models. arXiv preprint arXiv:2410.13720

    2024 arXiv

  70. [106]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learning, 8748--8763. PmLR

    2021

  71. [107]

    Shao, Y.; Zhai, W.; Yang, Y.; Luo, H.; Cao, Y.; and Zha, Z.-J. 2025. Great: Geometry-intention collaborative inference for open-vocabulary 3d object affordance grounding. In 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 17326--17336. IEEE

    2025

  72. [108]

    Shen, Y.; Maksutova, A.; Li, C.; and Unberath, M. 2025. Counterfactual world models via digital twin-conditioned video diffusion. arXiv preprint arXiv:2511.17481

    2025

  73. [109]

    Spyrou, N.; Vlontzos, A.; Pegios, P.; Melistas, T.; Gkouti, N.; Panagakis, Y.; Papanastasiou, G.; and Tsaftaris, S. A. 2025. Causally steered diffusion for automated video counterfactual generation. arXiv preprint arXiv:2506.14404

    2025 arXiv

  74. [110]

    Tan, Z.; Yang, H.; Qin, L.; Gong, J.; Yang, M.; and Li, H. 2025. Omni-video: Democratizing unified video understanding and generation. arXiv preprint arXiv:2507.06119

    2025

  75. [111]

    Teng, H.; Jia, H.; Sun, L.; Li, L.; Li, M.; Tang, M.; Han, S.; Zhang, T.; Zhang, W.; Luo, W.; et al. 2025. Magi-1: Autoregressive video generation at scale. arXiv preprint arXiv:2505.13211

    2025 arXiv

  76. [112]

    Wan, T.; Wang, A.; Ai, B.; Wen, B.; Mao, C.; Xie, C.-W.; Chen, D.; Yu, F.; Zhao, H.; Yang, J.; et al. 2025. Wan: Open and advanced large-scale video generative models. arXiv preprint arXiv:2503.20314

    2025 arXiv

  77. [113]

    Wang, H.; Liu, H.; Liu, X.; Du, C.; Kawaguchi, K.; Wang, Y.; and Pang, T. 2026. Fostering video reasoning via next-event prediction. In International Conference on Learning Representations, volume 2026, 31524--31570

    2026

  78. [114]

    Wei, C.; Liu, Q.; Ye, Z.; Wang, Q.; Wang, X.; Wan, P.; Gai, K.; and Chen, W. 2026. Univideo: Unified understanding, generation, and editing for videos. In International Conference on Learning Representations, volume 2026, 113905--113933

    2026

  79. [115]

    B.; and Gan, C

    Wu, B.; Yu, S.; Chen, Z.; Tenenbaum, J. B.; and Gan, C. 2024. Star: A benchmark for situated reasoning in real-world videos. arXiv preprint arXiv:2405.09711

    2024 arXiv

  80. [116]

    Wu, W.; Zhu, Z.; and Shou, M. Z. 2025. Automated movie generation via multi-agent cot planning. arXiv preprint arXiv:2503.07314

    2025 arXiv

  81. [117]

    Xiao, J.; Cheng, F.; Qi, L.; Gui, L.; Zhao, Y.; Lin, S.; Cen, J.; Ma, Z.; Yuille, A.; and Jiang, L. 2025. Videoauteur: Towards long narrative video generation. In 2025 IEEE/CVF International Conference on Computer Vision (ICCV), 19163--19173. IEEE

    2025

  82. [118]

    Xiao, J.; Shang, X.; Yao, A.; and Chua, T.-S. 2021. Next-qa: Next phase of question-answering to explaining temporal actions. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 9772--9781. IEEE

    2021

  83. [119]

    Yang, Y.; Fan, K.; Sun, S.; Li, H.; Zeng, A.; Han, F.; Zhai, W.; Liu, W.; Cao, Y.; and Zha, Z.-J. 2025 a . Videogen-eval: Agent-based system for video generation evaluation. arXiv preprint arXiv:2503.23452

    2025 arXiv

  84. [120]

    Yang, Y.; Liu, F.; Lu, Y.; Zhao, Q.; Wu, P.; Zhai, W.; Yi, R.; Cao, Y.; Ma, L.; Zha, Z.-J.; et al. 2025 b . Sigman: Scaling 3d human gaussian generation with millions of assets. In 2025 IEEE/CVF International Conference on Computer Vision (ICCV), 5122--5133. IEEE

    2025

  85. [121]

    Yang, Y.; Zhai, W.; Luo, H.; Cao, Y.; Luo, J.; and Zha, Z.-J. 2023. Grounding 3d object affordance from 2d interactions in images. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 10905--10915

    2023

  86. [122]

    Yang, Y.; Zhai, W.; Luo, H.; Cao, Y.; and Zha, Z.-J. 2024 a . Lemon: Learning 3d human-object interaction relation from 2d images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 16284--16295

    2024

  87. [123]

    Yang, Y.; Zhai, W.; Wang, C.; Yu, C.; Cao, Y.; and Zha, Z.-J. 2024 b . Egochoir: Capturing 3d human-object interaction regions from egocentric views. Advances in Neural Information Processing Systems, 37: 54529--54557

    2024

  88. [124]

    Yang, Y.; Zhang, F.; Pi, H.; Zeng, A.; Guo, S.; Xu, G.; Zhai, W.; Cao, Y.; and Zha, Z.-J. 2026. Gloria: Consistent Character Video Generation via Content Anchors. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 36724--36735

    2026

  89. [125]

    Yang, Z.; Teng, J.; Zheng, W.; Ding, M.; Huang, S.; Xu, J.; Yang, Y.; Hong, W.; Zhang, X.; Feng, G.; et al. 2025 c . Cogvideox: Text-to-video diffusion models with an expert transformer. In International Conference on Learning Representations, volume 2025, 83048--83077

    2025

  90. [126]

    Yi, K.; Gan, C.; Li, Y.; Kohli, P.; Wu, J.; Torralba, A.; and Tenenbaum, J. B. 2019. Clevrer: Collision events for video representation and reasoning. arXiv preprint arXiv:1910.01442

    2019 arXiv

  91. [127]

    Yu, C.; Zhai, W.; Yang, Y.; Cao, Y.; and Zha, Z.-J. 2025. Hero: Human reaction generation from videos. In 2025 IEEE/CVF International Conference on Computer Vision (ICCV), 10262--10274. IEEE

    2025

  92. [128]

    Yuan, S.; Huang, J.; He, X.; Ge, Y.; Shi, Y.; Chen, L.; Luo, J.; and Yuan, L. 2025. Identity-preserving text-to-video generation by frequency decomposition. In 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 12978--12988. IEEE

    2025

  93. [129]

    Zeng, A.; Yang, C.; Ge, C.; Zhang, E.; Xu, G.; Lin, G.; Gu, G.; Pi, J.; Li, L.; Shi, M.; et al. 2026. Lpm 1.0: Video-based character performance model. arXiv preprint arXiv:2604.07823

    2026 arXiv

  94. [130]

    Zeng, A.; Yang, Y.; Chen, W.; and Liu, W. 2024. The dawn of video generation: Preliminary explorations with sora-like models. arXiv preprint arXiv:2410.05227

    2024 arXiv

  95. [131]

    Z.; Zhi, T.; Yan, Q.; Kang, H.; Lu, X.; and Pan, X

    Zhang, K.; Jiang, L.; Wang, A.; Fang, J. Z.; Zhi, T.; Yan, Q.; Kang, H.; Lu, X.; and Pan, X. 2025. Storymem: Multi-shot long video storytelling with memory. arXiv preprint arXiv:2512.19539

    2025

  96. [132]

    Zhang, P.; Jia, Z.; Liu, K.; Weng, S.; Li, S.; and Shi, B. 2026. Stage: Storyboard-anchored generation for cinematic multi-shot narrative. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 659--669

    2026

  97. [133]

    Zheng, M.; Xu, Y.; Huang, H.; Ma, X.; Liu, Y.; Shu, W.; Pang, Y.; Tang, F.; Chen, Q.; Yang, H.; et al. 2024. Videogen-of-thought: Step-by-step generating multi-shot video with minimal manual intervention. arXiv preprint arXiv:2412.02259

    2024

  98. [134]

    Zhuang, S.; Li, K.; Chen, X.; Wang, Y.; Liu, Z.; Qiao, Y.; and Wang, Y. 2024. Vlogger: Make your dream a vlog. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 8806--8817. IEEE

    2024

Pith tools

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