Pith. sign in

REVIEW 4 major objections 5 minor 6 cited by

PhyT2V: LLM-Guided Iterative Self-Refinement for Physics-Grounded Text-to-Video Generation

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read PhyT2V claims that an iterative LLM refinement loop, which reads a caption of the current video and rewrites the prompt to close semantic and physical gaps, makes existing text-to-video models obey real-world physical rules up to 2.3x…

desk verdict Genuinely new inference-time prompt refinement loop for physics-grounded T2V, but the headline numbers rest on a circular metric; treat the 2.3x/35% as unverified until an independent evaluation lands. read the letter →

arxiv 2412.00596 v2 pith:WF6WLDAW submitted 2024-11-30 cs.CV cs.AI

Qiyao Xue , Xiangyu Yin , Boyuan Yang , Wei Gao This is my paper · ORCID
classification cs.CVcs.AI
keywords text-to-videogenerationdiffusiontransformerschain-of-thoughtpromptingstep-backphysicalcommonsensepromptrefinementvideocaptioningfeedbackLLM-guided
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 physical violations in AI-generated video can be reduced without retraining the generator or feeding it extra conditions such as depth maps or 3D geometry. The method, PhyT2V, runs an iterative loop in which an LLM first names the physical rules the scene should obey, a video captioning model converts the current video into text, the LLM identifies the mismatch between that caption and the user's prompt, and the LLM then rewrites the prompt to close the gap. The claim is that this text-only feedback loop improves adherence to real-world physical rules by up to 2.3x on existing text-to-video models and outperforms prompt enhancers by at least 35%. If true, the result would make text prompts a sufficient control surface for physical realism, offering a general, data-independent alternative to training on larger physics datasets.

What carries the argument

The central mechanism is the iterative self-refinement loop, which decomposes prompt rewriting into two parallel subproblems and one synthesis subproblem. Local chain-of-thought reasoning is used inside Step 1 (physical rules) and Step 2 (mismatch detection), while global step-back reasoning is used in Step 3 to combine those abstractions into a refined prompt; the video captioning model is the load-bearing bridge that keeps all reasoning in text. An external physics evaluator (VideoCon-Physics, a 0-to-1 score for physical commonsense and semantic adherence) supplies the feedback that decides whether another round is needed.

What would settle it

Take the evaluated prompts, generate videos with known physical errors (for example, an apple that splits into multiple bouncing pieces), caption those videos with Tarsier, and test whether the captions mention the violation; if captions are frequently silent or wrong, the mismatch step either corrects a non-existent problem or leaves the real one untouched. A direct rerun with a different video captioning model would also settle whether the reported gains come from the refinement loop or from idiosyncrasies of Tarsier's descriptions.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that chain-of-thought and step-back reasoning can be applied to text-to-video generation entirely in the text domain, using a caption of the generated video as the bridge. In each refinement round, Step 1 extracts the main object and the physical laws it should obey, Step 2 compares the video caption against the user prompt to find semantic and physical mismatches, and Step 3 produces a refined prompt that resolves the mismatch under the stated physical laws. The paper reports that this loop raises physical-commonsense and semantic-adherence scores across CogVideoX-5B, CogVideoX-2B, OpenSora, and VideoCrafter on the VideoPhy and PhyGenBench benchmarks, with gains up to 2.3x over the unrefined model and at least 35% over ChatGPT-4 and Promptist prompt enhancers, typically converging within three to four rounds.

Load-bearing premise

The load-bearing premise is that the video captioning model describes the generated video faithfully and completely enough for the LLM to see the actual physical violation; the paper does not measure caption accuracy on the evaluated videos.

Editorial extensions

If this is right

  • Existing T2V models should regain physical plausibility on out-of-distribution prompts through pure prompt rewriting, with most of the gain appearing in the first two refinement rounds.
  • The method transfers across model families with different architectures and input formats, since it only changes the text the model receives.
  • Prompt enhancers that do not inspect the generated video are expected to lag behind PhyT2V by at least 35% on physical-commonsense benchmarks.
  • The refinement loop needs only a few rounds before improvements converge, so the added cost is a handful of extra generations plus LLM calls per prompt.
  • VBench video-condition-consistency dimensions should improve alongside the physical-commonsense metrics, as reported in the paper's VBench results.

Reading between the lines

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

  • A natural extension, not tested in the paper, is to use the same generate-caption-compare-rewrite loop for other generative modalities whose outputs can be verbalized by a captioner or describer, such as text-to-image or text-to-3D.
  • The paper's stated bound is the captioning model's fidelity; a fair comparison would measure Tarsier's caption accuracy on the evaluated videos, since a hallucinated or over-literal caption would send the loop after a problem that is not in the video.
  • A testable follow-up is to replace the binary mismatch description with the evaluator's continuous score as the refinement signal, and check whether the loop converges faster or reaches higher final adherence.
  • Because the method requires no retraining and no model-internal access, it could plausibly be applied to closed, API-only T2V systems, which the paper does not demonstrate.
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

4 major / 5 minor

Summary. The paper proposes PhyT2V, a training-free, data-independent method that refines text-to-video prompts through an LLM-driven loop. In each round, Step 1 asks the LLM to identify the main object and relevant physical rules; Step 2 uses a video captioning model (Tarsier) to summarize the generated video and an LLM to find semantic mismatches between the caption and the user prompt; Step 3 uses step-back reasoning, together with a VideoCon-Physics score [S] of the previous round's video, to produce a refined prompt. The loop continues until the VideoCon-Physics score is satisfactory or improvement converges. The authors apply PhyT2V to CogVideoX-5B, CogVideoX-2B, OpenSora, and VideoCrafter on the VideoPhy and PhyGenBench benchmarks, reporting up to 2.3x improvement in physical-rule adherence over the unrefined model and at least 35% improvement over two prompt-enhancer baselines. The central claim is that prompt-only, LLM-guided iterative refinement can push existing T2V models toward more physically plausible videos without retraining or extra inputs.

Significance. If the reported gains hold up under independent measurement, PhyT2V would be a useful and low-cost contribution: it is model-agnostic, requires no retraining, and its three-step prompt templates are clearly specified and easy to reproduce. The paper also includes extensive qualitative comparisons, ablations of both reasoning steps, an honest appendix of failure cases, and a public code link. The main weakness is that the headline metric, VideoCon-Physics PC/SA, is also the feedback signal used inside the refinement loop, so the measured gains may reflect optimization of the evaluator rather than genuine physical adherence. The paper would be substantially strengthened by a human evaluation or a held-out metric not used in the loop, plus caption-fidelity checks, and by extending the baseline comparison to all four T2V models.

major comments (4)
  1. [§3.2, §3.3, §4] The headline PC/SA gains in Tables 1 and 2 are measured by the VideoCon-Physics evaluator, and the same evaluator's score [S] is injected into the Step 3 prompt (§3.2) and used as the stopping criterion (§3.3). The refinement loop is therefore explicitly optimizing the metric on which the paper's central claim rests: the LLM sees whether the previous prompt scored above or below 0.5, and the process terminates when the same evaluator reports convergence. This does not make the method internally inconsistent, but it means the reported 2.3x and 35% improvements are not independent evidence of improved physical adherence. I ask the authors to (i) report PC/SA with a held-out evaluator that was not used in the loop, (ii) run a blinded human evaluation on a stratified sample of generated videos, and (iii) report the agreement between VideoCon-Physics and human judgments on the generated videos. In addition, Tables 1–6 report only averages over binary outputs, with no confidence intervals or significance tests, so the reliability of the Round 2–4 increments is not established.
  2. [§3.2, Step 2] The mismatch signal in Step 2 depends on Tarsier's captions being faithful summaries of the generated videos. If the caption hallucinates content or omits a physical violation, the refined prompt will correct a problem that does not exist in the video or will leave the real problem uncorrected. The paper does not measure caption accuracy on the evaluated videos, and Appendix D's failure cases show that the T2V models can diverge substantially from prompts. Please add a caption-fidelity check, such as human evaluation of a random sample of captions or a caption-video consistency score, and report how often the Step 2 mismatch was judged correct on that sample.
  3. [§4.1, Tables 3 and 4] The claim of 'at least 35% improvement' over prompt enhancers is supported only for CogVideoX-5B and OpenSora; CogVideoX-2B and VideoCrafter are absent from Tables 3 and 4. Since the paper claims PhyT2V is generic across all four tested models, the baseline comparison should be extended to the remaining two models, or the claim should be restricted to the evaluated subset.
  4. [§1, §4.1, Figure 3] The out-of-distribution generality claim, stated in the abstract and introduction, is not quantified. No OOD split, distribution-shift measure, or separate OOD metric is provided; Figure 3 shows only selected anecdotal examples. I recommend either defining an OOD evaluation split and reporting the relevant metrics separately, or softening the claim to reflect that out-of-distribution behavior is demonstrated qualitatively.
minor comments (5)
  1. [§3, opening paragraph] The phrase 'Physcial-ground Text-to-Video' contains a typo and should read 'Physics-grounded Text-to-Video'.
  2. [§4, bullet list] 'PhyT2V is high generic' should be 'PhyT2V is highly generic'.
  3. [Figure 11 caption] The phrase 'which throne momentarily in the ambient light' appears to contain a typo; 'throne' should probably be 'hover' or 'hang'.
  4. [Figure 7 and surrounding text] The VBench results are presented only as a radar-style figure without numeric values; please report the numerical scores in a table or appendix so the improvements are independently readable.
  5. [Appendix A] The sentence 'The diffusion model generated video length is setted as 6 second and 8 frames per second' should be revised to 'set to 6 seconds and 8 frames per second'.

Circularity Check

1 steps flagged · score 7.0 of 10

Reported 2.3x/35% gains are measured with the same VideoCon-Physics evaluator used as the loop's feedback and stopping signal; the central claim is metric-optimization rather than independent evidence of physical grounding.

  1. self definitional [Section 3.2 'Prompting in the Final Subproblem for Global Step-Back Reasoning'; Section 3.3 'The Stopping Condition'; Section 4 'Evaluation metric' and Tables 1-4]
    "Furthermore, to ensure the correct reasoning path, we also provide quantitative feedback to the LLM about the effectiveness of previous round’s prompt refinement. Such effectiveness could be measured by the existing T2V evaluators, which judge the semantic alignment and quality of physical common sense of the currently generated video. For example, the VideoCon-Physics evaluator [2] gives a score ([S]) between 0 and 1. ... We use VideoCon-Physics evaluator provided with VideoPhy [2], to measure how the generated video adheres to physical common sense (PC) and achieves semantic adherence (SA)."

    The same VideoCon-Physics evaluator supplies [S] to the Step 3 prompt-refinement instruction and to the stopping rule, and then provides the PC/SA numbers from which the paper's headline 2.3x/35% improvements are read. The loop is therefore optimizing the exact metric used to validate it: an increase in PC/SA is an increase in the feedback signal the LLM was told to chase. No human evaluation, held-out video-language evaluator, or independent physics metric is reported, so the central claim 'improves adherence to real-world physical rules' is not separable from 'raised the VideoCon-Physics score it was given as input.' This is a partial rather than full equivalence because the frozen T2V generator sits between the refined prompt and the score, so the LLM does not directly write the score.

full rationale

PhyT2V is otherwise self-contained: it does not fit parameters to the output data, its LLM and video captioner are external tools, and there is no load-bearing self-citation chain or imported uniqueness theorem. The T2V model is frozen and refinement operates only on prompts. The circularity is confined to the evaluation loop: the metric used to measure the claim is the same metric injected into Step 3 and used as the stopping criterion. The appendix's failure cases and ablations do not provide an independent measure of physical correctness; they only illustrate the same VideoCon-Physics-based comparisons. Thus the paper's central quantitative claims are not independently established, though the method could still be genuinely useful if confirmed by a blinded human study or a held-out physics-grounded evaluator.

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

The system introduces no fitted numerical parameters in the physics sense; the hand-chosen controls are the prompt-quality threshold (0.5), the 120-word prompt cap, and the stopping round. The load-bearing assumptions are that prompt text can steer frozen T2V dynamics, that the video caption is faithful, and that VideoCon-Physics is a valid proxy for physical realism.

free parameters (3)
  • Score threshold for prompt quality = 0.5
    In Step 3, the LLM is told that a VideoCon-Physics score [S] above 0.5 means a good prompt and below means bad; this hand-chosen binary threshold steers which refinement path the LLM takes and affects all reported rounds.
  • Maximum refined prompt length = 120 words
    The LLM is instructed to output at most 120 words to fit the T2V model token limit; this constraint shapes all refined prompts and may limit expressiveness for complex physical scenes.
  • Number of refinement rounds = 4 rounds
    Experiments report rounds 1-4 and the paper concludes 3-4 rounds suffice based on observed convergence; the stopping rule is not a fixed formula but an observed convergence criterion.
assumptions (4)
  • domain assumption T2V model behavior can be steered by textual prompt content alone toward more physically realistic dynamics.
    PhyT2V changes only text prompts and assumes frozen T2V models will translate added physical context into improved motion and interactions; Section 3 and the failure cases show this does not hold for temporal flickering or rare scenes.
  • domain assumption The video captioning model (Tarsier) accurately and completely describes the generated video's semantics and physical content.
    Step 2 feeds the caption to the LLM to compute mismatches; any caption hallucination or omission changes the refinement target. Section 3.1 and Figure 6.
  • domain assumption VideoCon-Physics PC/SA scores are a valid proxy for real-world physical adherence.
    The same evaluator provides the [S] feedback signal in Step 3 and the reported PC/SA metrics; its validity is assumed, and no human study is included. Sections 3.2 and 4.
  • domain assumption LLM chain-of-thought and step-back reasoning improves prompt quality for T2V.
    The paper assumes CoT and step-back reasoning produce better physical descriptions; no controlled comparison against simple verbose prompting is shown except ablations of each step.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PhyT2V: LLM-Guided Iterative Self-Refinement for Physics-Grounded Text-to-Video Generation." pith.science (2026). https://pith.science/paper/WF6WLDAW

@misc{pith2026241200596,
  author       = {Pith},
  title        = {Pith review of: PhyT2V: LLM-Guided Iterative Self-Refinement for Physics-Grounded Text-to-Video Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WF6WLDAW}},
  note         = {Machine review of arXiv:2412.00596}
}
read the original abstract

Text-to-video (T2V) generation has been recently enabled by transformer-based diffusion models, but current T2V models lack capabilities in adhering to the real-world common knowledge and physical rules, due to their limited understanding of physical realism and deficiency in temporal modeling. Existing solutions are either data-driven or require extra model inputs, but cannot be generalizable to out-of-distribution domains. In this paper, we present PhyT2V, a new data-independent T2V technique that expands the current T2V model's capability of video generation to out-of-distribution domains, by enabling chain-of-thought and step-back reasoning in T2V prompting. Our experiments show that PhyT2V improves existing T2V models' adherence to real-world physical rules by 2.3x, and achieves 35% improvement compared to T2V prompt enhancers. The source codes are available at: https://github.com/pittisl/PhyT2V.

Figures

Figures reproduced from arXiv: 2412.00596 by the authors.

Figure 1
Figure 1. Left: videos generated by the current text-to-video generation model (CogVideoX-5B [50]) cannot adhere to the real-world physical rules (described in brackets following the user prompt). Right: our method PhyT2V, when applied to the same model, better reflects the real-world physical knowledge. Abstract Text-to-video (T2V) generation has been recently enabled by transformer-based diffusion models, but current T2V mo… view at source ↗
Figure 2
Figure 2. One iteration of video and prompt self-refinement in [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Examples of videos generated from in-distribution and [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (25 more)
Figure 5
Figure 5. Figure 5: Examples of CoT and step-back reasoning 2.2. Using LLM in T2V Generation LLMs with strong capabilities in natural language process￾ing (NLP) have been a natural choice for prompt refine￾ment in text-to-image and text-to-video generation, and ex￾isting work utilized LLM…
Figure 4
Figure 4. Figure 4: A video generated by enhancing the out-of-distribution [PITH_FULL_IMAGE:figures/full_fig_p003_4.png]
Figure 6
Figure 6. Figure 6: Our design of PhyT2V, illustrated by one round of video refinement consisting of three steps. Texts in brown are inputs from [PITH_FULL_IMAGE:figures/full_fig_p004_6.png]
Figure 8
Figure 8. Figure 8: Examples of videos generated using different categories [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Examples of videos generated using different categories [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 7
Figure 7. Figure 7: PhyT2V VBench evaluation results with CogVideoX-5B [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 11
Figure 11. Figure 11: Ablation study on Step 2 of mismatch reasoning [PITH_FULL_IMAGE:figures/full_fig_p008_11.png]
Figure 10
Figure 10. Figure 10: Ablation study on Step 1 of physical rule reasoning [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 12
Figure 12. Figure 12: Video generation example on fluid to fluid specific prompt in VideoPhy dataset [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]
Figure 13
Figure 13. Figure 13: Video generation example on solid to fluid specific prompt in VideoPhy dataset [PITH_FULL_IMAGE:figures/full_fig_p015_13.png]
Figure 14
Figure 14. Figure 14: Video generation example on solid to solid specific prompt in VideoPhy dataset [PITH_FULL_IMAGE:figures/full_fig_p016_14.png]
Figure 15
Figure 15. Figure 15: Video generation example on force specific prompt in PhyGenBench dataset [PITH_FULL_IMAGE:figures/full_fig_p017_15.png]
Figure 16
Figure 16. Figure 16: Video generation example on optics specific prompt in PhyGenBench [PITH_FULL_IMAGE:figures/full_fig_p018_16.png]
Figure 17
Figure 17. Figure 17: Video generation example on thermal specific prompt in PhyGenBench [PITH_FULL_IMAGE:figures/full_fig_p019_17.png]
Figure 18
Figure 18. Figure 18: Refinement detail example on solid to solid specific prompt in VideoPhy dataset [PITH_FULL_IMAGE:figures/full_fig_p020_18.png]
Figure 19
Figure 19. Figure 19: Refinement detail example on solid to fluid specific prompt in VideoPhy dataset [PITH_FULL_IMAGE:figures/full_fig_p021_19.png]
Figure 20
Figure 20. Figure 20: Refinement detail example on fluid to fluid specific prompt in VideoPhy dataset [PITH_FULL_IMAGE:figures/full_fig_p022_20.png]
Figure 21
Figure 21. Figure 21: Refinement detail example on force specific prompt in PhyGenBench dataset [PITH_FULL_IMAGE:figures/full_fig_p023_21.png]
Figure 22
Figure 22. Figure 22: Refinement detail example on optics specific prompt in PhyGenBench dataset [PITH_FULL_IMAGE:figures/full_fig_p024_22.png]
Figure 23
Figure 23. Figure 23: Refinement detail example on thermal specific prompt in PhyGenBench dataset [PITH_FULL_IMAGE:figures/full_fig_p025_23.png]
Figure 24
Figure 24. Figure 24: Ablation study on model size [PITH_FULL_IMAGE:figures/full_fig_p026_24.png]
Figure 25
Figure 25. Figure 25: The original prompt and output ### Instruction Your task is to identify the main object in the given user prompt and provide the physical laws in reality the main object should obey with as much detail as possible in a descriptive way without giving formulas. Some in-…
Figure 26
Figure 26. Figure 26: The prompt without the role indicator and output [PITH_FULL_IMAGE:figures/full_fig_p027_26.png]
Figure 27
Figure 27. Figure 27: The prompt without the in-context example and output [PITH_FULL_IMAGE:figures/full_fig_p028_27.png]
Figure 28
Figure 28. Figure 28: Failure case example on human hand related video generation [PITH_FULL_IMAGE:figures/full_fig_p028_28.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 6 Pith papers

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

  1. RDPO: Real Data Preference Optimization for Physics Consistency Video Generation

    cs.CV 2025-06 conditional novelty 8.0 of 10

    RDPO builds preference pairs by reverse-sampling real video latents with a pre-trained generator, then fine-tunes with Flow-DPO, improving physics consistency metrics on two video models.

  2. VideoREPA: Learning Physics for Video Generation through Relational Alignment with Foundation Models

    cs.CV 2025-05 conditional novelty 6.0 of 10

    VideoREPA adds a token-relation distillation loss that aligns a text-to-video diffusion model's internal features with VideoMAEv2, boosting physical commonsense scores on VideoPhy and VideoPhy2.

  3. Generative Physical AI in Vision: A Survey

    cs.CV 2025-01 conditional novelty 6.0 of 10

    A structured review that categorizes physics-aware generative models in vision into explicit-simulation and implicit-learning families and proposes six integration paradigms.

  4. From Black Box to Transparency: Enhancing Automated Interpreting Assessment with Explainable AI in College Classrooms

    cs.CL 2025-08 unverdicted novelty 5.0 of 10

    An explainable model using BLEURT, CometKiwi, pause features, and Chinese phraseological diversity predicts human-rated quality dimensions in English-Chinese consecutive interpreting, with SHAP identifying the stronge...

  5. Reasoning Physical Video Generation with Diffusion Timestep Tokens via Reinforcement Learning

    cs.CV 2025-04 conditional novelty 5.0 of 10

    Phys-AR, an autoregressive LLM trained on diffusion-timestep tokens with RL rewards for velocity and radius, generates physically consistent ball trajectories that extrapolate beyond its training distribution on PhyWorld.

  6. Hierarchical Fine-grained Preference Optimization for Physically Plausible Video Generation

    cs.CV 2025-08 conditional novelty 4.0 of 10

    A hierarchical direct preference optimization with four alignment levels plus automated data selection improves physical plausibility of text-to-video models.

Reference graph

Works this paper leans on

63 extracted references · 37 canonical work pages · cited by 6 Pith papers

  1. [1]

    Ai video generation expert discusses the tech- nology’s rapid advances—and its current limitations, 2024

    Luke Auburn. Ai video generation expert discusses the tech- nology’s rapid advances—and its current limitations, 2024. 1

    work page 2024

  2. [2]

    Videophy: Evaluating physical commonsense for video generation

    Hritik Bansal, Zongyu Lin, Tianyi Xie, Zeshun Zong, Michal Yarom, Yonatan Bitton, Chenfanfu Jiang, Yizhou Sun, Kai- Wei Chang, and Aditya Grover. Videophy: Evaluating physical commonsense for video generation. arXiv preprint arXiv:2406.03520, 2024. 5, 6

    arXiv 2024

  3. [3]

    Video generation models as world simulators, 2024

    Tim Brooks, Bill Peebles, Connor Holmes, Will DePue, Yufei Guo, Li Jing, David Schnurr, Joe Taylor, Troy Luh- man, Eric Luhman, et al. Video generation models as world simulators, 2024. 1, 6

    work page 2024

  4. [4]

    Videocrafter2: Overcoming data limitations for high-quality video diffu- sion models

    Haoxin Chen, Yong Zhang, Xiaodong Cun, Menghan Xia, Xintao Wang, Chao Weng, and Ying Shan. Videocrafter2: Overcoming data limitations for high-quality video diffu- sion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 7310– 7320, 2024. 3, 6, 11

    work page 2024

  5. [5]

    A survey on in-context learning

    Qingxiu Dong, Lei Li, Damai Dai, Ce Zheng, Jingyuan Ma, Rui Li, Heming Xia, Jingjing Xu, Zhiyong Wu, Tianyu Liu, et al. A survey on in-context learning. arXiv preprint arXiv:2301.00234, 2022. 5

    arXiv 2022

  6. [6]

    A survey on in-context learning

    Qingxiu Dong, Lei Li, Damai Dai, Ce Zheng, Jingyuan Ma, Rui Li, Heming Xia, Jingjing Xu, Zhiyong Wu, Baobao Chang, et al. A survey on in-context learning. In Proceed- ings of the 2024 Conference on Empirical Methods in Natu- ral Language Processing, pages 1107–1128, 2024. 5

    work page 2024

  7. [7]

    Blender, an open source design tool: Advances and integration in the architectural production pipeline

    Theodoros Dounas and Alexandros Sigalas. Blender, an open source design tool: Advances and integration in the architectural production pipeline. Aristoteleio University of Thessaloniki, 21:737–744, 2009. 2, 6

    work page 2009

  8. [8]

    Upbge: an open-source, 3d game en- gine forked from the old blender game engine, 2024

    Blender Foundation. Upbge: an open-source, 3d game en- gine forked from the old blender game engine, 2024. 2

    work page 2024

Show all 63 references
  1. [9]

    A systematic survey of prompt engineering on vision-language foundation models, 2023

    Jindong Gu, Zhen Han, Shuo Chen, Ahmad Beirami, Bailan He, Gengyuan Zhang, Ruotong Liao, Yao Qin, V olker Tresp, and Philip Torr. A systematic survey of prompt engineering on vision-language foundation models, 2023. 2

    2023

  2. [10]

    Photorealistic video generation with diffusion models

    Agrim Gupta, Lijun Yu, Kihyuk Sohn, Xiuye Gu, Meera Hahn, Fei-Fei Li, Irfan Essa, Lu Jiang, and Jos ´e Lezama. Photorealistic video generation with diffusion models. In European Conference on Computer Vision, pages 393–411. Springer, 2025. 2

    2025

  3. [11]

    Cogvideo: Large-scale pretraining for text-to-video generation via transformers

    Wenyi Hong, Ming Ding, Wendi Zheng, Xinghan Liu, and Jie Tang. Cogvideo: Large-scale pretraining for text-to-video generation via transformers. arXiv preprint arXiv:2205.15868, 2022. 2, 3

    2022 arXiv

  4. [12]

    Freezeasguard: Mitigating illegal adaptation of diffusion models via selec- tive tensor freezing

    Kai Huang, Haoming Wang, and Wei Gao. Freezeasguard: Mitigating illegal adaptation of diffusion models via selec- tive tensor freezing. arXiv preprint arXiv:2405.17472, 2024. 2

    2024 arXiv

  5. [13]

    To- wards green ai in fine-tuning large language models via adap- tive backpropagation

    Kai Huang, Hanyun Yin, Heng Huang, and Wei Gao. To- wards green ai in fine-tuning large language models via adap- tive backpropagation. ICLR, 2024. 3

    2024

  6. [14]

    Modality plug-and-play: Runtime modality adaptation in LLM-driven autonomous mobile systems

    Kai Huang, Xiangyu Yin, Heng Huang, and Wei Gao. Modality plug-and-play: Runtime modality adaptation in LLM-driven autonomous mobile systems. In ACM Mobi- Com, 2025. 3

    2025

  7. [15]

    Vbench: Comprehensive bench- mark suite for video generative models

    Ziqi Huang, Yinan He, Jiashuo Yu, Fan Zhang, Chenyang Si, Yuming Jiang, Yuanhan Zhang, Tianxing Wu, Qingyang Jin, Nattapol Chanpaisit, et al. Vbench: Comprehensive bench- mark suite for video generative models. In Proceedings of the IEEE/CVF Conference on Computer Vision and P...

    2024

  8. [16]

    Unreal engine: The most powerful real- time 3d creation tool, 2024

    Epic Games Inc. Unreal engine: The most powerful real- time 3d creation tool, 2024. 2

    2024

  9. [17]

    Pika labs, 2023

    Mellis Inc. Pika labs, 2023. 1

    2023

  10. [18]

    Introducing openai o1-preview, 2024

    OpenAI Inc. Introducing openai o1-preview, 2024. 2, 6

    2024

  11. [19]

    Kuhlen, and Tim Gerrits

    Marcel Kr ¨uger, David Gilbert, Torsten W. Kuhlen, and Tim Gerrits. Game engines for immersive visualization: Using unreal engine beyond entertainment. PRESENCE: Virtual and Augmented Reality, 33:31–55, 2024. 2, 6

    2024

  12. [20]

    A survey on long video generation: Challenges, methods, and prospects, 2024

    Chengxuan Li, Di Huang, Zeyu Lu, Yang Xiao, Qingqi Pei, and Lei Bai. A survey on long video generation: Challenges, methods, and prospects, 2024. 2

    2024

  13. [21]

    Promptist: Auto- mated prompt optimization for text-to-image synthesis

    WeiJie Li, Jin Wang, and Xuejie Zhang. Promptist: Auto- mated prompt optimization for text-to-image synthesis. In CCF International Conference on Natural Language Pro- cessing and Chinese Computing , pages 295–306. Springer,

  14. [22]

    Dissecting chain-of- thought: Compositionality through in-context filtering and learning

    Yingcong Li, Kartik Sreenivasan, Angeliki Giannou, Dim- itris Papailiopoulos, and Samet Oymak. Dissecting chain-of- thought: Compositionality through in-context filtering and learning. Advances in Neural Information Processing Sys- tems, 36, 2024. 5

    2024

  15. [23]

    Llm- grounded diffusion: Enhancing prompt understanding of text-to-image diffusion models with large language models

    Long Lian, Boyi Li, Adam Yala, and Trevor Darrell. Llm- grounded diffusion: Enhancing prompt understanding of text-to-image diffusion models with large language models. arXiv preprint arXiv:2305.13655, 2023. 3

    2023 arXiv

  16. [24]

    Llm-grounded video diffusion models

    Long Lian, Baifeng Shi, Adam Yala, Trevor Darrell, and Boyi Li. Llm-grounded video diffusion models. arXiv preprint arXiv:2309.17444, 2023

    2023 arXiv

  17. [25]

    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. arXiv preprint arXiv:2309.15091,

    arXiv

  18. [26]

    Physgen: Rigid-body physics-grounded image- to-video generation

    Shaowei Liu, Zhongzheng Ren, Saurabh Gupta, and Shen- long Wang. Physgen: Rigid-body physics-grounded image- to-video generation. In European Conference on Computer Vision, pages 360–378. Springer, 2025. 2, 6

    2025

  19. [27]

    Gpt4motion: Scripting physical motions in text-to-video generation via blender-oriented gpt planning

    Jiaxi Lv, Yi Huang, Mingfu Yan, Jiancheng Huang, Jianzhuang Liu, Yifan Liu, Yafei Wen, Xiaoxin Chen, and Shifeng Chen. Gpt4motion: Scripting physical motions in text-to-video generation via blender-oriented gpt planning. In Proceedings of the IEEE/CVF Conference on Computer Vi...

    2024

  20. [28]

    Gpt4motion: Scripting physical motions in text-to-video generation via blender-oriented gpt planning,

    Jiaxi Lv, Yi Huang, Mingfu Yan, Jiancheng Huang, Jianzhuang Liu, Yifan Liu, Yafei Wen, Xiaoxin Chen, and Shifeng Chen. Gpt4motion: Scripting physical motions in text-to-video generation via blender-oriented gpt planning,

  21. [29]

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

    Fanqing Meng, Jiaqi Liao, Xinyu Tan, Wenqi Shao, Quan- feng Lu, Kaipeng Zhang, Yu Cheng, Dianqi Li, Yu Qiao, and Ping Luo. Towards world simulator: Crafting physical commonsense-based benchmark for video generation. arXiv preprint arXiv:2410.05363, 2024. 6

    arXiv

  22. [30]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF Inter- national Conference on Computer Vision, pages 4195–4205,

  23. [31]

    Unrealcv: Connecting com- puter vision to unreal engine

    Weichao Qiu and Alan Yuille. Unrealcv: Connecting com- puter vision to unreal engine. In Computer Vision–ECCV 2016 Workshops: Amsterdam, The Netherlands, October 8- 10 and 15-16, 2016, Proceedings, Part III 14 , pages 909–

    2016

  24. [32]

    A systematic sur- vey of prompt engineering in large language models: Tech- niques and applications, 2024

    Pranab Sahoo, Ayush Kumar Singh, Sriparna Saha, Vinija Jain, Samrat Mondal, and Aman Chadha. A systematic sur- vey of prompt engineering in large language models: Tech- niques and applications, 2024. 2

    2024

  25. [33]

    Rogers, Inna Goncearenco, Giuseppe Sarli, Igor Galynker, Denis Peskoff, Marine Carpuat, Jules White, Shyamal Anad- kat, Alexander Hoyle, and Philip Resnik

    Sander Schulhoff, Michael Ilie, Nishant Balepur, Konstan- tine Kahadze, Amanda Liu, Chenglei Si, Yinheng Li, Aayush Gupta, HyoJung Han, Sevien Schulhoff, Pranav Sandeep Dulepet, Saurav Vidyadhara, Dayeon Ki, Sweta Agrawal, Chau Pham, Gerson Kroiz, Feileen Li, Hudson Tao, Ashay...

    2024

  26. [34]

    The prompt report: A systematic survey of prompting techniques

    Sander Schulhoff, Michael Ilie, Nishant Balepur, Konstan- tine Kahadze, Amanda Liu, Chenglei Si, Yinheng Li, Aayush Gupta, HyoJung Han, Sevien Schulhoff, et al. The prompt report: A systematic survey of prompting techniques. arXiv preprint arXiv:2406.06608, 2024. 5

    2024 arXiv

  27. [35]

    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 ,

    arXiv

  28. [36]

    Unity real-time development platform,

    Unity Technologies. Unity real-time development platform,

  29. [37]

    Thompson

    Stuart A. Thompson. A.i. can now create lifelike videos. can you tell what’s real?, 2024. 1

    2024

  30. [38]

    Towards understand- ing chain-of-thought prompting: An empirical study of what matters

    Boshi Wang, Sewon Min, Xiang Deng, Jiaming Shen, You Wu, Luke Zettlemoyer, and Huan Sun. Towards understand- ing chain-of-thought prompting: An empirical study of what matters. arXiv preprint arXiv:2212.10001, 2022. 4

    2022 arXiv

  31. [39]

    Tarsier: Recipes for training and evaluating large video description models

    Jiawei Wang, Liping Yuan, Yuchen Zhang, and Haomiao Sun. Tarsier: Recipes for training and evaluating large video description models. arXiv preprint arXiv:2407.00634, 2024. 2, 6, 11

    2024 arXiv

  32. [40]

    Knowledge-driven cot: Exploring faithful reasoning in llms for knowledge-intensive question answering

    Keheng Wang, Feiyu Duan, Sirui Wang, Peiguang Li, Yun- sen Xian, Chuantao Yin, Wenge Rong, and Zhang Xiong. Knowledge-driven cot: Exploring faithful reasoning in llms for knowledge-intensive question answering. arXiv preprint arXiv:2308.13259, 2023. 5

    2023 arXiv

  33. [41]

    Worlddreamer: Towards general world models for video generation via predicting masked to- kens, 2024

    Xiaofeng Wang, Zheng Zhu, Guan Huang, Boyuan Wang, Xinze Chen, and Jiwen Lu. Worlddreamer: Towards general world models for video generation via predicting masked to- kens, 2024. 2

    2024

  34. [42]

    Lavie: High-quality video gener- ation 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 gener- ation with cascaded latent diffusion models. arXiv preprint arXiv:2309.15103, 2023. 2, 3

    2023 arXiv

  35. [43]

    Internvid: A large-scale video-text dataset for multimodal understanding and generation

    Yi Wang, Yinan He, Yizhuo Li, Kunchang Li, Jiashuo Yu, Xin Ma, Xinhao Li, Guo Chen, Xinyuan Chen, Yaohui Wang, et al. Internvid: A large-scale video-text dataset for multimodal understanding and generation. arXiv preprint arXiv:2307.06942, 2023. 3

    2023 arXiv

  36. [44]

    In- context learning unlocked for diffusion models

    Zhendong Wang, Yifan Jiang, Yadong Lu, Pengcheng He, Weizhu Chen, Zhangyang Wang, Mingyuan Zhou, et al. In- context learning unlocked for diffusion models. Advances in Neural Information Processing Systems , 36:8542–8562,

  37. [45]

    Customvideo: Customizing text-to- video generation with multiple subjects

    Zhao Wang, Aoxue Li, Lingting Zhu, Yong Guo, Qi Dou, and Zhenguo Li. Customvideo: Customizing text-to- video generation with multiple subjects. arXiv preprint arXiv:2401.09962, 2024. 3

    2024

  38. [46]

    Self-correcting llm-controlled diffu- sion models

    Tsung-Han Wu, Long Lian, Joseph E Gonzalez, Boyi Li, and Trevor Darrell. Self-correcting llm-controlled diffu- sion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 6327– 6336, 2024. 3

    2024

  39. [47]

    Worldgpt: a sora-inspired video ai agent as rich world models from text and image inputs

    Deshun Yang, Luhui Hu, Yu Tian, Zihao Li, Chris Kelly, Bang Yang, Cindy Yang, and Yuexian Zou. Worldgpt: a sora-inspired video ai agent as rich world models from text and image inputs. arXiv preprint arXiv:2403.07944, 2024. 6, 11

    2024 arXiv

  40. [48]

    Mastering text-to-image dif- fusion: Recaptioning, planning, and generating with multi- modal llms

    Ling Yang, Zhaochen Yu, Chenlin Meng, Minkai Xu, Ste- fano Ermon, and CUI Bin. Mastering text-to-image dif- fusion: Recaptioning, planning, and generating with multi- modal llms. In Forty-first International Conference on Ma- chine Learning, 2024. 3

    2024

  41. [49]

    Learn- ing interactive real-world simulators

    Mengjiao Yang, Yilun Du, Kamyar Ghasemipour, Jonathan Tompson, Dale Schuurmans, and Pieter Abbeel. Learn- ing interactive real-world simulators. arXiv preprint arXiv:2310.06114, 2023. 2

    2023 arXiv

  42. [50]

    Cogvideox: Text-to-video diffusion models with an expert transformer, 2024

    Zhuoyi Yang, Jiayan Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong, and Jie Tang. Cogvideox: Text-to-video diffusion models ...

    2024

  43. [51]

    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, Xiao- han Zhang, Guanyu Feng, et al. Cogvideox: Text-to-video diffusion models with an expert transformer. arXiv preprint arXiv:2408.06072, 2024. 1, 2, 3, 6

    2024 arXiv

  44. [52]

    Video in-context learning

    Wentao Zhang, Junliang Guo, Tianyu He, Li Zhao, Linli Xu, and Jiang Bian. Video in-context learning. arXiv preprint arXiv:2407.07356, 2024. 3

    2024 arXiv

  45. [53]

    Open-sora: Democratizing efficient video production for all, 2024

    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, 2024. 3, 6, 11

    2024

  46. [54]

    Compositional 3d-aware video gen- eration with llm director

    Hanxin Zhu, Tianyu He, Anni Tang, Junliang Guo, Zhibo Chen, and Jiang Bian. Compositional 3d-aware video gen- eration with llm director. arXiv preprint arXiv:2409.00558,

    arXiv

  47. [57]

    This is described by the phase transition theory, a subset of thermodynamics

    Phase Transition: Arsenic trioxide should experience phase transition under sufficient heat. This is described by the phase transition theory, a subset of thermodynamics . Initially, arsenic trioxide at room conditions is in solid form, and as the thermodynamic equilibrium cha...

  48. [58]

    Gravity pulls the apple downward, causing it to accelerate toward the Earth

    Gravitational Acceleration As the apple falls, it is influenced by the force of gravity. Gravity pulls the apple downward, causing it to accelerate toward the Earth. The apple's acceleration due to gravity is constant near the Earth's surface, regardless of its mass. Its veloc...

  49. [59]

    The magnitude of air resistance depends on the apple's shape, size, and speed, as well as the density of the air

    Air Resistance (Drag Force) While the apple is falling, it encounters air resistance, a force opposing its motion. The magnitude of air resistance depends on the apple's shape, size, and speed, as well as the density of the air. If the apple falls from a sufficient height, it ...

  50. [60]

    This interaction is governed by: Newton's Third Law of Motion: The apple exerts a force on the ground, and the ground exerts an equal and opposite force on the apple

    Impact with the Hard Ground When the apple hits the ground, it undergoes a collision. This interaction is governed by: Newton's Third Law of Motion: The apple exerts a force on the ground, and the ground exerts an equal and opposite force on the apple. Impulse and Momentum: Du...

  51. [61]

    As it falls, this potential energy is converted into kinetic energy

    Conservation of Energy Before the fall, the apple has gravitational potential energy due to its height above the ground. As it falls, this potential energy is converted into kinetic energy. At the moment of impact, some of the kinetic energy transforms into other forms: Elasti...

  52. [62]

    For the apple-ground system, the COR is less than 1, indicating an inelastic collision where energy is lost to deformation and other processes

    Coefficient of Restitution (COR) The apple’s ability to bounce depends on the COR, a measure of how elastic the collision is. For the apple-ground system, the COR is less than 1, indicating an inelastic collision where energy is lost to deformation and other processes

  53. [63]

    Its upward motion decelerates due to gravity until it momentarily stops at its maximum height

    Post-Bounce Motion After bouncing, the apple follows a parabolic trajectory dictated by gravitational forces. Its upward motion decelerates due to gravity until it momentarily stops at its maximum height. If the apple bounces multiple times, each subsequent bounce will be shor...

  54. [916]

    Springer, 2016. 2, 6

    2016

  55. [2024]

    3 A. Details of Evaluation Setup Since our proposed technique of PhyT2V does not involve any efforts of retraining the T2V model, in this section we describe details about our evaluation setup of the LLM in- ference for CoT and step-back reasoning. In our evaluations, we use 4...

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.