Pith. sign in

REVIEW 4 major objections 5 minor 54 references

Executable Blender code as a process-level chain of thought enforces physical consistency in text-to-video generation.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 08:23 UTC pith:7IKT5FY5

load-bearing objection Clever code-as-CoT video pipeline with solid benchmark gains, but the central decomposition claim is not actually proven by the ablation, and missing code/data makes results unverifiable. the 4 major comments →

arxiv 2607.27380 v1 pith:7IKT5FY5 submitted 2026-07-29 cs.CV cs.AI

VideoCoCo: Code-as-CoT for Physically-Consistent Video Generation via an Agentic Dual-Engine System

Haodong Li , Tianfei Ren , Xiaoxiao Ma , Chunmei Qing , Zhen Fang , Sipeng He , Ziyu Guo , Haoyu Wu
show 20 more authors
Juanxi Tian Yihang Zou Ruichuan An Dongzhi Jiang Boxue Yang Ji Xie Xu Huang Wenhao Yan Jialv Zou Zhengrong Yue Yaxin Luo Xiaotong Li Yuzhu Wang Junyan Ye Jinjing Zhao Zehui Chen Lin Chen Renye Yan Feng Zhao Pheng-Ann Heng
This is my paper
classification cs.CV cs.AI
keywords video generationphysical consistencychain-of-thoughtexecutable codeBlender simulationdraft-conditioned editingtext-to-videoworld modeling
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tries to establish that physically consistent video generation can be achieved by separating the physics from the pixels: have an agent write an executable Blender program specifying the scene and its temporal evolution, run it in a sandbox to get a deterministic low-fidelity draft, then condition a video editor on that draft to add photorealistic appearance. The claim is that this executable, frame-dense intermediate—a 'process-level chain of thought'—carries the physical dynamics, so the generative model only needs to restyle, not reason about how objects move. The authors report that this raises the average physical-consistency score from 0.475 to 0.558 on PhyGenBench and from 52.18% to 77.88% on VBench-2.0, the best average on both benchmarks, with the largest gains in thermal and material dynamics where appearance-driven priors are weakest. A sympathetic reader would care because it offers a controllable, inspectable way to inject physics into video generation without hand-authored rules per phenomenon.

Core claim

The central claim is that executable code, not textual plans or keyframes, should be the intermediate reasoning representation for video generation. VideoCoCo externalizes the process hidden in a text prompt as a self-contained Blender Python program; running that program in a sandbox yields a deterministic, temporally dense draft of the event. A second engine, a video editor fine-tuned on the VideoCoCo-3K dataset of draft–instruction–target triplets, takes this draft as a structural condition and produces a photorealistic video. The paper argues that this decomposition lets the simulation engine commit to what happens and when, while the editor decides only how it looks, and presents benchm

What carries the argument

The central object is the 'process-level chain of thought': an executable Blender program that explicitly declares objects, physical properties, and temporal evolution, rendered into a deterministic spatiotemporal draft. Named in the paper as Code-as-CoT, this representation is explicit (every object and motion is declared), executable (the program actually runs and produces every frame), and inspectable (the code can be read, edited, and re-run). Its work in the argument is to commit the system to a concrete physical realization before any pixel is synthesized, so that the downstream generative video engine is asked only to restyle an already-instantiated process; the VideoCoCo-3K dataset s

Load-bearing premise

The load-bearing premise is that the LLM judges used by the benchmarks score genuine physical consistency, so the reported gains really reflect better physics rather than a bias toward the edited videos' appearance; the paper offers no human comparison to confirm this.

What would settle it

Run the same PhyGenBench and VBench-2.0 prompt suites through VideoCoCo and the base generator and have human annotators with physics training rate which video better matches the requested dynamics; if human ratings do not reproduce a substantial gap favoring VideoCoCo, the central claim that the draft carries the physics is not established.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If the claim holds, any prompt whose physics can be expressed in a simulator can be made physically consistent by drafting first, independent of the video model's own physical priors.
  • The largest gains in thermal and material dynamics suggest that appearance-driven generators are weakest exactly where an explicit simulator adds the most, pointing to a division of labor for future systems.
  • VideoCoCo-3K-style aligned triplets become a general training resource: any video editor can be adapted to consume simulated drafts, so improvements in the editor transfer to the whole pipeline.
  • Because the intermediate artifacts are code and rendered video, failures can be debugged by editing the program, not by re-prompting the generator; this makes physical errors inspectable and correctable.
  • LoRA adaptation outperforming full fine-tuning indicates that the needed skill—restyling drafts while preserving motion—is narrow and can be learned with modest data, lowering the cost of adapting the pipeline to new editors.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same code-as-draft idea could be pushed one step further: instead of a video editor conditioned on a draft, a video model could be trained with synthetic intermediate targets so that at inference the simulation is not needed, as the authors themselves suggest for future work—but the paper does not demonstrate this.
  • Because the staged construction is automatic, a natural extension is to close the loop with a verifier that checks whether the rendered draft matches the prompt before editing; the paper does not implement such a rejection loop.
  • The benchmark gains are only as trustworthy as the judges; a human rating study on the same prompt set would test whether the gap reflects genuine physical plausibility rather than a stylistic preference for the editor's outputs.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes VideoCoCo, an agentic dual-engine framework for physically consistent text-to-video generation. A coding agent writes a Blender program from a text prompt; the program is executed in a sandbox to render a deterministic, low-fidelity 'white-clay' spatiotemporal draft. An instruction agent composes an appearance-focused editing instruction, and a video editor is conditioned jointly on the draft and instruction to produce the final photorealistic video. To make off-the-shelf editors consume simulated drafts, the authors construct VideoCoCo-3K, a dataset of draft–instruction–target triplets generated by a teacher editor (Seedance 2.0), and fine-tune the editor with a conditional denoising objective (Eq. 7). Experiments on PhyGenBench and VBench-2.0 report substantial gains over the OmniWeaving base (0.475 to 0.558 and 52.18% to 77.88%, respectively), with ablations separating tuning-free and fine-tuned variants. The central claim is that executable code serves as a process-level chain of thought, decoupling physical dynamics from appearance synthesis.

Significance. If the claims hold, VideoCoCo is a meaningful step toward controllable and inspectable physical video generation: unlike textual plans or keyframes, executable code can be run, inspected, and revised, and the dense simulated draft provides explicit temporal grounding. The paper includes a useful ablation isolating the draft's contribution from editor adaptation, reports per-category scores on two physics benchmarks, and identifies clear failure cases (e.g., turbulent fluids). However, the strength of the central decomposition claim depends on untested assumptions about the LLM judges used for evaluation and on the attribution of gains in Table 3, which currently appears to contradict the text. The absence of released code, data, and evaluation details further limits verification. The idea is promising and the empirical framework is coherent, but the evidence as presented does not yet fully support the paper's strongest claims.

major comments (4)
  1. [§4.3, Table 3; §4.2 'Takeaway'] The text states that 'Drafting supplies most of the physical gain,' but the Table 3 numbers show the opposite attribution. OmniWeaving→Tune-Free improves the average by 0.506−0.475 = 0.031, while Tune-Free→LoRA-Tune improves by 0.558−0.506 = 0.052; thus editor adaptation contributes roughly 63% of the total gain. This is a load-bearing inconsistency for the central decomposition claim. Please correct the claim, or provide a controlled edit-only baseline (e.g., omitting the draft but keeping the instruction and LoRA adaptation) that would isolate the draft's contribution.
  2. [§4.1 'Datasets and Metrics'] PhyGenBench scores rely entirely on GPT-4o as the MLLM judge, and VBench-2.0 uses a 'pre-registered subset' that is never defined. No human agreement, judge discriminability analysis, or complete benchmark results are reported. Since the editor restyles appearance (lighting, texture, camera), the observed gains could reflect judge preference for cleaner restyled videos rather than physical fidelity. Please report judge–human correlation (or at least a human-rated subset), specify the exact VBench-2.0 prompts/suite used, and justify why the subset is representative.
  3. [§3.4, Eq. (7)] The training setup is underspecified. Eq. (7) is a generic conditional denoising loss, but the architecture changes needed to condition the editor on the draft video d are not described: is d injected as concatenated frames, via cross-attention, or through an adapter? What LoRA rank, learning rate, training steps, and compute were used? Without these details the 'LoRA-Tune outperforms Full-Tune' result cannot be reproduced or interpreted. Additionally, the VideoCoCo-3K targets are generated by Seedance 2.0 using the same draft-and-instruction pipeline; there is no verification that the teacher outputs actually preserve the draft's spatiotemporal trajectories, which is critical because the editor learns to imitate the teacher.
  4. [Reproducibility] The project page (https://github.com/micky-li-hd/VideoCoCo) is listed but no code, data, or checkpoints are provided in the manuscript. Given the paper's reliance on a new dataset and a tuned editor, and its claim of reproducibility ('fully automatic, inspectable, and reproducible'), releasing at least the dataset metadata, Blender programs, and evaluation scripts is necessary for the claims to be verifiable. Please clarify the release plan.
minor comments (5)
  1. [Table 1 vs. §4.2] The abstract and text use 0.475/0.558 while Table 1 rounds to 0.48/0.56; please standardize to avoid apparent inconsistency.
  2. [Figure 2] The figure contains placeholder symbols ('??') in the draft rendering chain, which make the pipeline unclear. Use actual rendered draft frames or remove the placeholders.
  3. [§2.3] 'VideoRL VR' reads as a typo; clarify the method name and citation.
  4. [§4.1] The term 'pre-registered subset' appears only here; define it in the text or in an appendix, including how it was selected from VBench-2.0.
  5. [§3.4, Eq. (7)] The loss is written as plain L2 on the noise residual; specify whether any reweighting (e.g., SNR weighting) is used, and whether the model is trained with classifier-free guidance on the draft condition.

Circularity Check

0 steps flagged

No significant circularity: the Code-as-CoT claim rests on external benchmarks and a tuning-free ablation; the self-citations and self-built dataset do not make the prediction equivalent to its inputs.

full rationale

The derivation chain is empirical: p -> c = A_code(p) (Eq. 1), d = B(c) (Eq. 2), e = A_edit(p,d) (Eq. 3), v = G_theta(d,e) (Eq. 4). The only fitted component, G_theta, is adapted on VideoCoCo-3K triplets whose targets are produced by an external teacher Seedance 2.0 (Eq. 5-6), and Sec. 3.3 excludes evaluation-benchmark prompts and near-duplicates, so benchmark scores are not obtained by feeding the training labels back through the model. The central claim that the draft supplies dynamics is isolated by the tuning-free ablation in Sec. 4.3 (0.475 -> 0.506 with no editor parameter update), which is an empirical comparison rather than a construction; no equation in the paper reduces the predicted physical consistency to the fitted objective by definition. The Code-as-CoT framing cites the authors' own CoCo/DraCo papers, but those self-citations are motivational, not load-bearing evidence, and the LoRA gain is normal teacher distillation from an external model rather than a parameter renamed as a prediction. Two concerns are noted for the correctness pass, not the circularity pass: the Sec. 4.3 takeaway "Drafting supplies most of the physical gain" overstates the table's deltas (LoRA adaptation, 0.506 -> 0.558, contributes more than the draft, 0.475 -> 0.506), and the LLM-judge validity on PhyGenBench/VBench-2.0 is not checked against human ratings. Section 5's stated limitations (latency, Blender expressiveness) similarly affect scope, not circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 2 invented entities

The paper's central claim does not rest on fitted numerical constants; the main burden is on unstated evaluation details, judge reliability, and the self-generated training set. No explicit free parameters are disclosed, which is itself a gap because dataset size/quality and judge configuration are implicit choices.

axioms (4)
  • domain assumption The sandboxed Blender execution is deterministic and the rendered draft faithfully captures the physical process intended by the prompt.
    The method relies on the code agent writing correct-enough Blender programs, and on the draft video preserving the target dynamics. The paper acknowledges simulator expressiveness limits (Sec. 5), but treats the drafts as trustworthy scaffolds (Sec. 3.1).
  • domain assumption LLM judge scores (GPT-4o for PhyGenBench; VBench-2.0 protocol) are a valid proxy for physical consistency.
    The headline results are completely mediated by these judges; no human correlation or error analysis is provided (Sec. 4.1, Tables 1-2).
  • domain assumption OmniWeaving is a suitable base editor, and its weights/training setup can be adapted with Eq. 7 as described.
    The model is initialized from OmniWeaving and the paper omits architectural specifics of how the draft conditions the editor; this assumption is load-bearing for the reported gains.
  • domain assumption VideoCoCo-3K triplets, generated by Seedance, provide supervision that transfers to new prompts.
    The dataset is created by the same pipeline (Sec. 3.3) and evaluation prompts are excluded, but no analysis of distribution/shift or contamination is given beyond the exclusion claim.
invented entities (2)
  • VideoCoCo-3K dataset no independent evidence
    purpose: Training data aligning Blender drafts with photorealistic targets to adapt the video editor.
    The dataset has no external validation; it is constructed with the authors' own pipeline and is not released in the manuscript.
  • Draft-conditioned editor adaptation paradigm (Code-as-CoT for video) independent evidence
    purpose: Intermediate representation for physically consistent video generation.
    The generated videos and benchmark scores provide falsifiable handles outside the paper, but no artifact release is provided.

pith-pipeline@v1.3.0-daily-deepseek · 12708 in / 6955 out tokens · 50281 ms · 2026-08-01T08:23:34.002878+00:00 · methodology

0 comments
read the original abstract

Text-to-video models have achieved remarkable visual quality, yet they still struggle to generate physically consistent dynamics because the temporal evolution of a scene must be inferred implicitly from a highly compressed text prompt. Existing chain-of-thought approaches introduce intermediate plans or visual states, but these representations are typically non-executable or temporally sparse, limiting their ability to instantiate and control the complete spatiotemporal process. To address this limitation, we introduce VideoCoCo, an agentic dual-engine framework in which executable Blender code serves as a process-level chain of thought. Given a text prompt, a coding agent synthesizes a Blender program that explicitly specifies the scene and its temporal evolution. The executable simulation engine runs the program to produce a deterministic spatiotemporal draft, which is subsequently transformed into a photorealistic video by a generative video engine through draft-conditioned editing. This decomposition separates process-level reasoning from high-fidelity visual realization. To adapt the video editor to simulated drafts, we construct VideoCoCo-3K, a curated dataset of draft-instruction-target triplets. VideoCoCo improves the OmniWeaving baseline from 0.475 to 0.558 on PhyGenBench and from 52.18 to 77.88 on VBench-2.0, achieving the best average score on both benchmarks. These results demonstrate that executable code provides an effective, controllable, and inspectable intermediate representation for physically consistent video generation.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

54 extracted references · 26 linked inside Pith

  1. [1]

    Vchain: Chain-of-visual- thought for reasoning in video generation

    Ziqi Huang, Ning Yu, Gordon Chen, Haonan Qiu, Paul Debevec, and Ziwei Liu. Vchain: Chain-of-visual- thought for reasoning in video generation. In Findings of the Association for Computational Linguistics (ACL Findings), 2026

    2026

  2. [2]

    Videodirectorgpt: Consistent multi-scene video generation via llm-guided planning

    Han Lin, Abhay Zala, Jaemin Cho, and Mohit Bansal. Videodirectorgpt: Consistent multi-scene video generation via llm-guided planning. In Conference on Language Modeling (COLM), 2024

    2024

  3. [3]

    Huang, Selena Song, Xiaofei Wen, Yuankai Li, Hoifung Poon, and Muhao Chen

    Tinghui Zhu, Sheng Zhang, James Y. Huang, Selena Song, Xiaofei Wen, Yuankai Li, Hoifung Poon, and Muhao Chen. Video models can reason with verifiable rewards. arXiv preprint arXiv:2605.15458, 2026

    Pith/arXiv arXiv 2026

  4. [4]

    Video models reason early: Exploiting plan commit- ment for maze solving

    Kaleb Newman, Tyler Zhu, and Olga Russakovsky. Video models reason early: Exploiting plan commit- ment for maze solving. arXiv preprint arXiv:2603.30043, 2026

    arXiv 2026

  5. [5]

    Video-t1: Test-time scaling for video generation

    Fangfu Liu, Hanyang Wang, Yimo Cai, Kaiyan Zhang, Xiaohang Zhan, and Yueqi Duan. Video-t1: Test-time scaling for video generation. In IEEE/CVF International Conference on Computer Vision (ICCV), 2025

    2025

  6. [6]

    Temporal backtracking search for test-time generative video reasoning

    Sejoon Jun, Zheng Ding, Huangyuan Su, Weirui Ye, and Yilun Du. Temporal backtracking search for test-time generative video reasoning. arXiv preprint arXiv:2606.13861, 2026

    arXiv 2026

  7. [7]

    Video diffusion models

    Jonathan Ho, Tim Salimans, Alexey Gritsenko, William Chan, Mohammad Norouzi, and David J Fleet. Video diffusion models. Advances in neural information processing systems, 35:8633–8646, 2022

    2022

  8. [8]

    Make-a-video: Text-to-video generation without text-video data

    Uriel Singer, Adam Polyak, Thomas Hayes, Xi Yin, Jie An, Songyang Zhang, Qiyuan Hu, Harry Yang, Oron Ashual, Oran Gafni, et al. Make-a-video: Text-to-video generation without text-video data. arXiv preprint arXiv:2209.14792, 2022

    Pith/arXiv arXiv 2022

  9. [9]

    Imagen video: High definition video generation with diffusion models

    Jonathan Ho, William Chan, Chitwan Saharia, Jay Whang, Ruiqi Gao, Alexey Gritsenko, Diederik P Kingma, Ben Poole, Mohammad Norouzi, David J Fleet, et al. Imagen video: High definition video generation with diffusion models. arXiv preprint arXiv:2210.02303, 2022

    Pith/arXiv arXiv 2022

  10. [10]

    Align your latents: High-resolution video synthesis with latent diffusion models

    Andreas Blattmann, Robin Rombach, Huan Ling, Tim Dockhorn, Seung Wook Kim, Sanja Fidler, and Karsten Kreis. Align your latents: High-resolution video synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 22563– 22575, 2023

    2023

  11. [11]

    Videocrafter2: Overcoming data limitations for high-quality video diffusion models

    Haoxin Chen, Yong Zhang, Xiaodong Cun, Menghan Xia, Xintao Wang, Chao Weng, and Ying Shan. Videocrafter2: Overcoming data limitations for high-quality video diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7310–7320, 2024

    2024

  12. [12]

    Lavie: High-quality video generation with cascaded latent diffusion models

    Yaohui Wang, Xinyuan Chen, Xin Ma, Shangchen Zhou, Ziqi Huang, Yi Wang, Ceyuan Yang, Yinan He, Jiashuo Yu, Peiqing Yang, et al. Lavie: High-quality video generation with cascaded latent diffusion models. International Journal of Computer Vision, 133(5):3059–3078, 2025

    2025

  13. [13]

    Tune-a-video: One-shot tuning of image diffusion models for text-to-video generation

    Jay Zhangjie Wu, Yixiao Ge, Xintao Wang, Stan Weixian Lei, Yuchao Gu, Yufei Shi, Wynne Hsu, Ying Shan, Xiaohu Qie, and Mike Zheng Shou. Tune-a-video: One-shot tuning of image diffusion models for text-to-video generation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 7623–7633, October 2023. 12

    2023

  14. [14]

    Open-sora: Democratizing efficient video production for all

    Zangwei Zheng, Xiangyu Peng, Tianji Yang, Chenhui Shen, Shenggui Li, Hongxin Liu, Yukun Zhou, Tianyi Li, and Yang You. Open-sora: Democratizing efficient video production for all. arXiv preprint arXiv:2412.20404, 2024

    Pith/arXiv arXiv 2024

  15. [15]

    Hunyuanvideo: A systematic framework for large video generative models

    Weijie Kong, Qi Tian, Zijian Zhang, Rox Min, Zuozhuo Dai, Jin Zhou, Jiangfeng Xiong, Xin Li, Bo Wu, Jianwei Zhang, et al. Hunyuanvideo: A systematic framework for large video generative models. arXiv preprint arXiv:2412.03603, 2024

    Pith/arXiv arXiv 2024

  16. [17]

    Cogvideox: Text-to-video diffusion models with an expert transformer

    Zhuoyi Yang, Jiayan Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, et al. Cogvideox: Text-to-video diffusion models with an expert transformer. arXiv preprint arXiv:2408.06072, 2024

    Pith/arXiv arXiv 2024

  17. [18]

    Step-video-t2v technical report: The practice, challenges, and future of video foundation model, 2025

    Guoqing Ma, Haoyang Huang, Kun Yan, Liangyu Chen, Nan Duan, Shengming Yin, Changyi Wan, Ranchen Ming, Xiaoniu Song, Xing Chen, Yu Zhou, Deshan Sun, Deyu Zhou, Jian Zhou, Kaijun Tan, Kang An, Mei Chen, Wei Ji, Qiling Wu, Wen Sun, Xin Han, Yanan Wei, Zheng Ge, Aojie Li, Bin Wang, Bizhu Huang, Bo Wang, Brian Li, Changxing Miao, Chen Xu, Chenfei Wu, Chenguang...

    Pith/arXiv arXiv 2025

  18. [19]

    Diffusion model as a generalist segmentation learner

    Haoxiao Wang, Antao Xiang, Haiyang Sun, Peilin Sun, Changhao Pan, Yifu Chen, Minjie Hong, Weijie Wang, Shuang Chen, Yue Chen, et al. Diffusion model as a generalist segmentation learner. arXiv preprint arXiv:2604.24575, 2026

    Pith/arXiv arXiv 2026

  19. [20]

    Scope: Simulating cross-game operations in playable environments for fps world models

    Zizhao Tong, Hongfeng Lai, Zeqing Wang, Zhaohu Xing, Kexu Cheng, Haoran Xu, Zhao Pu, Shangwen Zhu, Ruili Feng, Jian Zhao, et al. Scope: Simulating cross-game operations in playable environments for fps world models. arXiv preprint arXiv:2605.23345, 2026

    Pith/arXiv arXiv 2026

  20. [21]

    Mind: Benchmarking memory consistency and action control in world models

    Yixuan Ye, Xuanyu Lu, Yuxin Jiang, Yuchao Gu, Rui Zhao, Qiwei Liang, Jiachun Pan, Fengda Zhang, Weijia Wu, and Alex Jinpeng Wang. Mind: Benchmarking memory consistency and action control in world models. arXiv preprint arXiv:2602.08025, 2026

    arXiv 2026

  21. [22]

    Reconstruction alignment improves unified multimodal models

    Ji Xie, Trevor Darrell, Luke Zettlemoyer, and XuDong Wang. Reconstruction alignment improves unified multimodal models. arXiv preprint arXiv:2509.07295, 2025

    Pith/arXiv arXiv 2025

  22. [23]

    Geometry forcing: Marrying video diffusion and 3D representation for consistent world modeling

    Haoyu Wu, Diankun Wu, Tianyu He, Junliang Guo, Yang Ye, Yueqi Duan, and Jiang Bian. Geometry forcing: Marrying video diffusion and 3D representation for consistent world modeling. In The Fourteenth International Conference on Learning Representations, 2026. URL https://openreview.net/forum? id=ULXYZCms41

    2026

  23. [24]

    MultiWorld: Scalable multi-agent multi-view video world models

    Haoyu Wu, Jiwen Yu, Yingtian Zou, and Xihui Liu. MultiWorld: Scalable multi-agent multi-view video world models. arXiv preprint arXiv:2604.18564, 2026

    Pith/arXiv arXiv 2026

  24. [25]

    Video generation models as world simulators

    Tim Brooks, Bill Peebles, Connor Holmes, Will DePue, Yufei Guo, Leo Jing, David Schnurr, Joe Taylor, 13 Troy Luhman, Eric Luhman, et al. Video generation models as world simulators. OpenAI Blog, 1(8):1, 2024

    2024

  25. [27]

    Videophy-2: A challenging action-centric physical commonsense evaluation in video generation

    Hritik Bansal, Clark Peng, Yonatan Bitton, Roman Goldenberg, Aditya Grover, and Kai-Wei Chang. Videophy-2: A challenging action-centric physical commonsense evaluation in video generation. arXiv preprint arXiv:2503.06800, 2025

    Pith/arXiv arXiv 2025

  26. [28]

    Think before you diffuse: Infusing physical rules into video diffusion

    Ke Zhang, Cihan Xiao, Jiacong Xu, Yiqun Mei, and Vishal M Patel. Think before you diffuse: Infusing physical rules into video diffusion. arXiv preprint arXiv:2505.21653, 2025

    arXiv 2025

  27. [29]

    Wisa: World simulator assistant for physics-aware text-to-video generation

    Jing Wang, Ao Ma, Ke Cao, Jun Zheng, Zhanjie Zhang, Jiasong Feng, Shanyuan Liu, Yuhang Ma, Bo Cheng, Dawei Leng, et al. Wisa: World simulator assistant for physics-aware text-to-video generation. arXiv preprint arXiv:2503.08153, 2025

    Pith/arXiv arXiv 2025

  28. [30]

    Phyt2v: Llm-guided iterative self-refinement for physics-grounded text-to-video generation

    Qiyao Xue, Xiangyu Yin, Boyuan Yang, and Wei Gao. Phyt2v: Llm-guided iterative self-refinement for physics-grounded text-to-video generation. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 18826–18836, 2025

    2025

  29. [31]

    Vlipp: Towards physically plausible video generation with vision and language informed physical prior

    Xindi Yang, Baolu Li, Yiming Zhang, Zhenfei Yin, Lei Bai, Liqian Ma, Zhiyong Wang, Jianfei Cai, Tien-Tsin Wong, Huchuan Lu, et al. Vlipp: Towards physically plausible video generation with vision and language informed physical prior. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12360–12370, 2025

    2025

  30. [32]

    Physctrl: Generative physics for controllable and physics-grounded video generation

    Chen Wang, Chuhao Chen, Yiming Huang, Zhiyang Dou, Yuan Liu, Jiatao Gu, and Lingjie Liu. Physctrl: Generative physics for controllable and physics-grounded video generation. arXiv preprint arXiv:2509.20358, 2025

    arXiv 2025

  31. [33]

    Physchoreo: Physics-controllable video generation with part-aware semantic grounding

    Haoze Zhang, Tianyu Huang, Zichen Wan, Xiaowei Jin, Hongzhi Zhang, Hui Li, and Wangmeng Zuo. Physchoreo: Physics-controllable video generation with part-aware semantic grounding. arXiv preprint arXiv:2511.20562, 2025

    Pith/arXiv arXiv 2025

  32. [34]

    VideoDPO: Omni-preference alignment for video diffusion generation

    Runtao Liu, Haoyu Wu, Ziqiang Zheng, Chen Wei, Yingqing He, Renjie Pi, and Qifeng Chen. VideoDPO: Omni-preference alignment for video diffusion generation. arXiv preprint arXiv:2412.14167, 2024

    Pith/arXiv arXiv 2024

  33. [35]

    Phygdpo: Physics-aware groupwise direct preference optimiza- tion for physically consistent text-to-video generation

    Yuanhao Cai, Kunpeng Li, Menglin Jia, Jialiang Wang, Junzhe Sun, Feng Liang, Weifeng Chen, Felix Juefei-Xu, Chu Wang, Ali Thabet, et al. Phygdpo: Physics-aware groupwise direct preference optimiza- tion for physically consistent text-to-video generation. arXiv preprint arXiv:2512.24551, 2025

    Pith/arXiv arXiv 2025

  34. [36]

    Physcorr: Dual-reward dpo for physics-constrained text-to- video generation with automated preference selection

    Peiyao Wang, Weining Wang, and Qi Li. Physcorr: Dual-reward dpo for physics-constrained text-to- video generation with automated preference selection. arXiv preprint arXiv:2511.03997, 2025

    arXiv 2025

  35. [37]

    Pisa experiments: Exploring physics post-training for video diffusion models by watching stuff drop

    Chenyu Li, Oscar Michel, Xichen Pan, Sainan Liu, Mike Roberts, and Saining Xie. Pisa experiments: Exploring physics post-training for video diffusion models by watching stuff drop. arXiv preprint arXiv:2503.09595, 2025

    Pith/arXiv arXiv 2025

  36. [38]

    Physrvg: Physics-aware unified reinforcement learning for video generative models

    Qiyuan Zhang, Biao Gong, Shuai Tan, Zheng Zhang, Yujun Shen, Xing Zhu, Yuyuan Li, Kelu Yao, Chunhua Shen, and Changqing Zou. Physrvg: Physics-aware unified reinforcement learning for video generative models. arXiv preprint arXiv:2601.11087, 2026

    arXiv 2026

  37. [39]

    Auto-rubric as reward: From implicit preferences to explicit multimodal generative criteria

    Juanxi Tian, Fengyuan Liu, Jiaming Han, Yilei Jiang, Yongliang Wu, Yesheng Liu, Haodong Li, Furong Xu, and Wanhua Li. Auto-rubric as reward: From implicit preferences to explicit multimodal generative criteria. arXiv preprint arXiv:2605.08354, 2026

    Pith/arXiv arXiv 2026

  38. [40]

    Chi, Quoc V

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems (NeurIPS), 2022. 14

    2022

  39. [41]

    DraCo: Draft as CoT for text-to-image preview and rare concept generation

    Dongzhi Jiang, Renrui Zhang, Haodong Li, Zhuofan Zong, Ziyu Guo, Jun He, Claire Guo, Junyan Ye, Rongyao Fang, Weijia Li, et al. DraCo: Draft as CoT for text-to-image preview and rare concept generation. arXiv preprint arXiv:2512.05112, 2025

    arXiv 2025

  40. [42]

    CoCo: Code as CoT for text-to-image preview and rare concept generation, 2026

    Haodong Li, Chunmei Qing, Huanyu Zhang, Dongzhi Jiang, Yihang Zou, Hongbo Peng, Dingming Li, Yuhong Dai, ZePeng Lin, Juanxi Tian, Yi Zhou, Siqi Dai, and Jingwei Wu. CoCo: Code as CoT for text-to-image preview and rare concept generation, 2026. URL https://arxiv.org/abs/2603.08652

    arXiv 2026

  41. [43]

    MetaPoint: Unlocking precise spatial control in agentic visual generation

    Dewei Zhou, Xinyu Huang, Xun Wang, Ji Xie, Yabo Zhang, Liang Li, Kunchang Li, Zongxin Yang, and Yi Yang. MetaPoint: Unlocking precise spatial control in agentic visual generation. arXiv preprint arXiv:2606.05031, 2026

    Pith/arXiv arXiv 2026

  42. [44]

    VideoCoF: Unified video editing with temporal reasoner

    Xiangpeng Yang, Ji Xie, Yiyuan Yang, Yue Ma, Yan Huang, Min Xu, and Qiang Wu. VideoCoF: Unified video editing with temporal reasoner. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 37940–37949, June 2026

    2026

  43. [45]

    Pika Labs. Pika. https://pika.art, 2023. Accessed: 2026-07-20

    2023

  44. [46]

    Introducing gen-3 alpha: A new frontier for video generation

    Runway. Introducing gen-3 alpha: A new frontier for video generation. https://runwayml.com/ research/introducing-gen-3-alpha , 2024. Accessed: 2026-07-20

    2024

  45. [47]

    Kling ai: Text-to-video generation service

    Kuaishou Technology. Kling ai: Text-to-video generation service. https://klingai.com/, 2024. Ac- cessed: 2026-07-20

    2024

  46. [48]

    Vchitect-2.0: Parallel transformer for scaling up video diffusion models

    Weichen Fan, Chenyang Si, Junhao Song, Zhenyu Yang, Yinan He, Long Zhuo, Ziqi Huang, Ziyue Dong, Jingwen He, Dongwei Pan, et al. Vchitect-2.0: Parallel transformer for scaling up video diffusion models. arXiv preprint arXiv:2501.08453, 2025

    Pith/arXiv arXiv 2025

  47. [49]

    Wan: Open and advanced large-scale video generative models

    Wan Team. Wan: Open and advanced large-scale video generative models. arXiv preprint arXiv:2503.20314, 2025

    Pith/arXiv arXiv 2025

  48. [50]

    Cosmos world foundation model platform for physical ai

    NVIDIA. Cosmos world foundation model platform for physical ai. arXiv preprint arXiv:2501.03575, 2025

    Pith/arXiv arXiv 2025

  49. [51]

    LTX-Video: Realtime video latent diffusion

    Yoav HaCohen et al. LTX-Video: Realtime video latent diffusion. arXiv preprint arXiv:2501.00103, 2024

    Pith/arXiv arXiv 2024

  50. [52]

    Omniweaving: Towards unified video generation with free-form composition and reasoning

    Kaihang Pan et al. Omniweaving: Towards unified video generation with free-form composition and reasoning. arXiv preprint arXiv:2603.24458, 2026

    arXiv 2026

  51. [53]

    Kling ai 1.6: Text-to-video generation service

    Kuaishou Technology. Kling ai 1.6: Text-to-video generation service. https://klingai.com/, 2024. Accessed: 2026-07-20

    2024

  52. [54]

    Towards world simulator: Crafting physical commonsense-based benchmark for video generation

    Fanqing Meng et al. Towards world simulator: Crafting physical commonsense-based benchmark for video generation. arXiv preprint arXiv:2410.05363, 2024

    Pith/arXiv arXiv 2024

  53. [55]

    GPT-4o system card

    OpenAI. GPT-4o system card. arXiv preprint arXiv:2410.21276, 2024

    Pith/arXiv arXiv 2024

  54. [56]

    VBench-2.0: Advancing video generation benchmark suite for intrinsic faithfulness

    Dian Zheng et al. VBench-2.0: Advancing video generation benchmark suite for intrinsic faithfulness. arXiv preprint arXiv:2503.21755, 2025. 15

    Pith/arXiv arXiv 2025