Pith. sign in

REVIEW 2 major objections 5 minor 135 references

AlayaWorld: Interactive Long-Horizon World Modeling - Full Technical Report (v1.1)

T0 review · 2 major / 5 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 AlayaWorld claims that replacing depth-warped spatial memory with a streaming 3D point-cache renderer, and feeding all visual conditions through the same causal-VAE latent space as the generated video, yields top long-horizon consistency…

desk verdict A clearly described engineering iteration on AlayaWorld with strong WBench consistency numbers, but the paper never shows that the six conditioning changes cause the improvement. read the letter →

arxiv 2608.13492 v1 pith:ASOMPE5V submitted 2026-08-13 cs.AI

AlayaWorld Team: Kaipeng Zhang , Chuanhao Li , Yifan Zhan , Yongtao Ge , Yuanyang Yin , Jiaming Tan , Kang He , Liaoyuan Fan
show 9 more authors
Mingliang Zhai Ruicong Liu Xiaojie Xu Xuangeng Chu Zhen Li Zhengyuan Lin Zhixiang Wang Zian Meng Zihui Gao
This is my paper · ORCID
classification cs.AI
keywords AlayaWorldlong-horizonmodelinginteractivevideogeneration3Dpoint-cacherenderercausalVAEconditioningvisualconsistencycameracontrolWBenchnavigation
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 the previous version of AlayaWorld kept generated video from staying consistent because its conditioning signals did not match how the video itself was represented: static images lacked temporal context, spatial memory was encoded outside the causal latent structure, and camera control arrived through a separate branch. The new version rewires conditioning so that every visual input, including image conditions, spatial memory, and temporal memory, is encoded with the same causal video autoencoder and aligned temporal windows used to encode the generated video. The authors test this on the WBench navigation benchmark and report the best overall Consistency score of 89.5, with the highest Background, Perspective, Subject, and Geometric consistency among the evaluated methods. The broader point is that long-horizon interactive world models may need the conditions they see to match the latent and temporal structure of what they generate, not just more capacity or more data.

What carries the argument

The load-bearing mechanism is a streaming 3D point-cache renderer paired with a nine-frame causal-encoding window. After each generated chunk, per-pixel 3D points are accumulated in a persistent cache; before the next chunk, the cache is re-rendered from the planned camera viewpoint, giving the model a geometry-aligned spatial condition. That rendered view is then encoded causally along with a real prefix frame, instead of being encoded as an isolated image, so the resulting spatial latents share the same temporal context and VAE structure as the target video latents. The same nine-frame window is used for image conditioning and chunk-to-chunk handoff, and invalid rendered regions are removed rather than masked. This mechanism is what carries the reported consistency improvement by making all visual conditions and the generated video live in a common latent and temporal representation.

What would settle it

Re-run the 158 WBench navigation cases several times with different evaluator seeds and compute confidence intervals for each consistency metric. If the average consistency scores of AlayaWorld and the second-best method overlap within noise, then the best-consistency claim is not established. Alternatively, ablate the hard memory dropout alone and look at Background and Segment consistency; if scores do not change, the sequence-length mechanism is not load-bearing.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that a conditioning pipeline built around a streaming 3D point-cache renderer, plus six specific design changes, produces the strongest long-horizon visual-persistence results in the comparison. The six changes are: motion-aware latent conditioning instead of static-frame conditioning; causal encoding of re-rendered spatial memory as a continuous sequence; pixel-aligned temporal memory; hard memory dropout that removes tokens rather than zeroing them; a unified VAE protocol across training and inference; and removal of the separate camera modulation branch so that viewpoint control is carried entirely by geometry-aligned visual conditions. The numbers carrying the claim are the consistency averages: 89.5 overall, 94.1 background, 93.4 subject, 86.6 perspective, and 94.1 geometric, each best in the reported table. The paper treats these results as validating the principle that conditioning should match generated content in both latent representation and temporal structure.

Load-bearing premise

The entire best-consistency claim rests on WBench's Consistency scores being reliable, repeatable measurements of visual and geometric consistency; the report gives single point estimates with no confidence intervals or evaluator-variance analysis, so if those scores are noisy the 89.5 result could change.

Editorial extensions

If this is right

  • If the reported consistency scores hold, representing every conditioning signal in the same causal-VAE latent space as the generated video is a direct lever for long-horizon visual stability, separate from model capacity or data scale.
  • Camera control without a dedicated modulation branch becomes viable: a planned viewpoint expressed by re-rendering 3D memory is enough to convey scale, visibility, and parallax to the generator.
  • Hard memory dropout that removes tokens rather than zeroing them should reduce the training-inference gap in autoregressive video models, because inference-time sequence lengths then match the memory-free steps seen during training.
  • Because the gains concentrate in Background, Perspective, Subject, and Geometric consistency, improving scene semantics and physical plausibility is a separate problem from memory-driven persistence.

Reading between the lines

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

  • A direct ablation of the six design changes would show whether the consistency gain comes mostly from the 3D point-cache renderer, the matched causal encoding, or the hard dropout; the report does not isolate them, so that causal story is my reading rather than an established result.
  • The same latent-matching principle could be tested in other interactive settings, such as object manipulation or embodied navigation, where keeping the same identity and layout stable over time is the bottleneck.
  • Because the point-cache geometry is scale-aligned by a pairwise-median ratio over prefix displacements, a natural stress test is to corrupt or drop the estimated 3D points mid-rollout; if consistency degrades sharply, geometry is indeed the carrier of the effect.
  • The reported margin over the nearest competitor is only a few points, so independent re-scoring with evaluator variance would clarify whether the 89.5 ranking is stable or within noise.
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

2 major / 5 minor

Summary. The manuscript describes version 1.1 of AlayaWorld, an interactive long-horizon video world model. The backbone, chunk-wise autoregressive generation scheme, and training data are unchanged from a previous release, but the conditioning pipeline is substantially redesigned. Specifically, the previous depth-warping-based spatial memory is replaced by a streaming 3D point-cache renderer, and six conditioning modifications are introduced: motion-aware latent conditioning, causal encoding of re-rendered spatial memory, pixel-aligned temporal memory, hard memory dropout, a unified VAE protocol, and removal of the camera AdaLN branch in favor of geometry-based camera control. The report evaluates AlayaWorld on the WBench navigation split (158 cases) and claims the best overall Consistency score of 89.5, with leading scores in Background, Perspective, Subject, and Geometric Consistency, alongside a competitive Video Quality average of 79.1. The paper attributes this result to the redesigned conditioning pipeline and concludes that the new spatial and temporal memory mechanisms are effective for long-horizon interactive generation.

Significance. If the reported result is taken at face value, the paper demonstrates a practically meaningful improvement in long-horizon visual and geometric consistency for interactive world models, and it does so using an external benchmark (WBench) and an external geometry estimator (ViGeo), which is a strength relative to self-constructed evaluations. The design principle that conditioning signals should match generated content in latent representation and temporal structure is coherent and potentially transferable. However, the quantitative support is thin: there are no error bars, no ablations of the six proposed changes, and no comparison against the previous AlayaWorld version, so the causal attribution of the Consistency improvement to the new conditioning pipeline is not established. The contribution is therefore currently a plausible engineering claim rather than a verified one.

major comments (2)
  1. [§2.1, Table 1] The central claim that the six stated conditioning modifications cause the best Consistency score of 89.5 is not supported by the reported experiments. Table 1 contains no row for the previous AlayaWorld version and no ablation that disables any of the six changes; the baselines are unrelated third-party systems that differ in backbone, training data, and evaluation protocol. Consequently, the Consistency advantage cannot be attributed to the redesigned conditioning pipeline, contrary to the wording in §2.1 that the results are 'validating the effectiveness of the proposed spatial and temporal memory mechanisms.' A controlled comparison (previous-version row, plus ablations of at least the point-cache renderer, motion-aware conditioning, hard memory dropout, and unified VAE protocol) is required to establish the paper's central claim.
  2. [§2.1, Table 1] Every WBench score is reported as a single point estimate with no confidence intervals, repeated trials, evaluator-variance analysis, or description of how the benchmark handles stochasticity in generative video models. Because the headline 'best Consistency score of 89.5' is the paper's main quantitative result, the authors should report uncertainty (for example, bootstrapped intervals over the 158 navigation cases or multiple evaluation runs) or explain explicitly why the scores are deterministic.
minor comments (5)
  1. [§2.1, text vs. Table 1] There are numeric inconsistencies between the text and Table 1: the text states a Video Quality average of 79.1 while Table 1 reports 79.3, and it states a Navigation score of 79.9 while Table 1 reports 80.0. These should be reconciled.
  2. [§1, modification 1] The description 'stride + 1 = 9' implicitly assumes a stride of 8, but the stride is never defined; please state the frame stride and why a nine-frame window is used.
  3. [§1, modification 2] The abbreviation 'DA3' is used without definition; it first appears in 'DA3-based depth warping' and only later becomes clear from reference [1] that it refers to Depth Anything 3.
  4. [§1, modification 1] The sentence 'use the second latent as the image condition' is unclear: please specify why the second latent, rather than the first or another latent, corresponds to the conditioning frame and how this aligns with the decoder prefix.
  5. [Table 1] Several table entries appear run together (for example, '62.664.451.6' and '96.896.894.9'); the table should be reformatted so each cell is clearly separated.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the central claim is supported by an external benchmark, and no fitted parameter is folded back into the reported result.

full rationale

The report derives no result from its redesigns: Section 1 describes six conditioning changes as design choices, and Section 2.1 reports WBench scores measured on an external benchmark (Ref. [2]) against third-party baselines. No equation in the paper shows that any reported consistency score is equivalent to an input by construction. ViGeo (Ref. [3]) and Depth Anything 3 (Ref. [1]) are cited external estimators, and their outputs are inputs to the generation pipeline rather than predictions that are then relabeled as WBench scores. The one least-squares fit in the paper, for camera intrinsics, is an internal calibration step and is not presented as a predicted benchmark result. The only self-comparison is to 'the previous version' of AlayaWorld, which is descriptive rather than a load-bearing citation chain; no previous-version row or ablation is given, so the attribution of the 89.5 Consistency score to the six specific redesigns is under-supported, but missing ablative evidence is a correctness gap, not circularity. No uniqueness theorem, ansatz, or fitted parameter is imported from the authors' own prior work, and the cited references are external. Therefore the circularity score is 0.

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

The central claim rests on benchmark measurements rather than on a mathematical derivation. The main unstated assumptions are the reliability of the WBench evaluator, the accuracy of ViGeo geometry, and the comparability of external baseline scores. The model's trained weights are not enumerated as free parameters because the training data and training procedure are not disclosed.

free parameters (3)
  • Temporal memory length N = 4
    Reduced from six latents to four by design; no ablation or derivation justifies this specific choice.
  • Causal encoding window stride = 8 (nine-frame window)
    Chosen to match the VAE temporal stride and used for image conditioning, spatial memory, and chunk handoff; it is a structural hyperparameter.
  • Camera intrinsics = least-squares pinhole fit per clip
    Estimated per clip from the ViGeo point map and kept fixed; this is a fitted calibration, not a target-derived constant.
assumptions (4)
  • domain assumption ViGeo per-pixel 3D geometry is accurate enough for point-cache registration and re-rendering.
    Used in Section 1, item 2; if geometry estimates drift, geometric consistency would degrade.
  • domain assumption The causal VAE's frame-to-latent mapping is consistent across training and inference after the protocol unification.
    Assumed in items 1, 3, and 5; no formal proof or inspection is provided.
  • domain assumption The WBench navigation split is a valid and unbiased measure of interactive world-model consistency.
    Table 1 relies on it; no error bars or protocol details are given in the report.
  • domain assumption The external baseline scores in Table 1 were obtained under comparable evaluation conditions.
    Required for the 'best' claim; not verifiable from the report.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AlayaWorld: Interactive Long-Horizon World Modeling - Full Technical Report (v1.1)." pith.science (2026). https://pith.science/paper/ASOMPE5V

@misc{pith2026260813492,
  author       = {Pith},
  title        = {Pith review of: AlayaWorld: Interactive Long-Horizon World Modeling - Full Technical Report (v1.1)},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ASOMPE5V}},
  note         = {Machine review of arXiv:2608.13492}
}
read the original abstract

This report presents an improved version of AlayaWorld. While the backbone architecture, chunk-wise autoregressive generation scheme, and training data remain unchanged from the previous release, we substantially revise how conditioning signals are represented and integrated into the model. The new design is guided by a simple principle: conditioning signals should match the generated content as closely as possible in both latent representation and temporal structure. To this end, we make two major changes. First, we replace the previous depth-warping-based spatial memory with a streaming 3D point-cache renderer. Second, we redesign the conditioning pipeline so that visual conditions are encoded in the same causal-VAE latent space, with temporal statistics consistent with those of the generated video. Concretely, the new version introduces six modifications: (1) replacing static-frame image conditioning with motion-aware latent conditioning; (2) causally encoding re-rendered spatial memory as a continuous sequence; (3) aligning the temporal-memory window in pixel space; (4) adopting hard memory dropout that removes memory tokens rather than zeroing them; (5) unifying the VAE encoding and decoding protocol across training and inference; and (6) removing the camera AdaLN branch, such that viewpoint control is provided entirely through the re-rendered spatial condition.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

135 extracted references · 14 canonical work pages

  1. [1]

    arXiv preprint arXiv:2406.10126 , year=

    Training-free camera control for video generation , author=. arXiv preprint arXiv:2406.10126 , year=

    arXiv

  2. [2]

    Latent-reframe: Enabling camera control for video diffusion models without training , author=

  3. [3]

    ACM SIGGRAPH 2024 Conference Papers , pages=

    Motionctrl: A unified and flexible motion controller for video generation , author=. ACM SIGGRAPH 2024 Conference Papers , pages=

    2024

  4. [4]

    arXiv preprint arXiv:2404.02101 , year=

    Cameractrl: Enabling camera control for text-to-video generation , author=. arXiv preprint arXiv:2404.02101 , year=

    arXiv

  5. [5]

    arXiv preprint arXiv:2410.15957 , year=

    Cami2v: Camera-controlled image-to-video diffusion model , author=. arXiv preprint arXiv:2410.15957 , year=

    arXiv

  6. [6]

    arXiv preprint arXiv:2406.02509 , year=

    Camco: Camera-controllable 3d-consistent image-to-video generation , author=. arXiv preprint arXiv:2406.02509 , year=

    arXiv

  7. [7]

    Cameractrl II: Dynamic scene exploration via camera-controlled video diffusion models , author=

  8. [8]

    arXiv preprint arXiv:2507.17744 , year=

    Yume: An interactive world generation model , author=. arXiv preprint arXiv:2507.17744 , year=

    arXiv

Show all 135 references
  1. [9]

    Cameras as relative positional encoding , author=

  2. [10]

    Unified camera positional encoding for controlled video generation , author=

  3. [11]

    Bullettime: Decoupled control of time and camera pose for video generation , author=

  4. [12]

    Gen3c: 3d-informed world-consistent video generation with precise camera control , author=

  5. [13]

    Trajectorycrafter: Redirecting camera trajectory for monocular videos via diffusion models , author=

  6. [14]

    Recammaster: Camera-controlled generative rendering from a single video , author=

  7. [15]

    Recapture: Generative video camera controls for user-provided videos using masked video fine-tuning , author=

  8. [16]

    SIGGRAPH Asia 2025 Conference Papers , pages=

    Camclonemaster: Enabling reference-based camera control for video generation , author=. SIGGRAPH Asia 2025 Conference Papers , pages=

    2025

  9. [17]

    2025 , eprint=

    HunyuanVideo 1.5 Technical Report , author=. 2025 , eprint=

    2025

  10. [18]

    arXiv preprint , year=

    HY-World 1.5: A Systematic Framework for Interactive World Modeling with Real-Time Latency and Geometric Consistency , author=. arXiv preprint , year=

  11. [19]

    Frame context packing and drift prevention in next-frame-prediction video diffusion models , author=

  12. [20]

    arXiv preprint arXiv:2512.23851 , year=

    Pretraining Frame Preservation in Autoregressive Video Memory Compression , author=. arXiv preprint arXiv:2512.23851 , year=

    arXiv

  13. [21]

    SIGGRAPH Asia 2025 Conference Papers , pages=

    Context as memory: Scene-consistent interactive long video generation with memory retrieval , author=. SIGGRAPH Asia 2025 Conference Papers , pages=

    2025

  14. [22]

    Worldmem: Long-term consistent world simulation with memory , author=

  15. [23]

    Video world models with long-term spatial memory , author=

  16. [24]

    arXiv preprint arXiv:2510.01183 , year=

    Evoworld: Evolving panoramic world generation with explicit 3d memory , author=. arXiv preprint arXiv:2510.01183 , year=

  17. [25]

    Spatia: Video generation with updatable spatial memory , author=

  18. [26]

    arXiv preprint arXiv:2604.13036 , year=

    Lyra 2.0: Explorable generative 3d worlds , author=. arXiv preprint arXiv:2604.13036 , year=

    arXiv

  19. [27]

    arXiv e-prints , year=

    HY-World 2.0: A Multi-Modal World Model for Reconstructing, Generating, and Simulating 3D Worlds , author=. arXiv e-prints , year=

  20. [28]

    arXiv preprint arXiv:2511.12940 , year=

    Recurrent autoregressive diffusion: Global memory meets local attention , author=. arXiv preprint arXiv:2511.12940 , year=

    arXiv

  21. [29]

    Long-context state-space video world models , author=

  22. [30]

    arXiv preprint arXiv:2512.04519 , year=

    Videossm: Autoregressive long video generation with hybrid state-space memory , author=. arXiv preprint arXiv:2512.04519 , year=

  23. [31]

    arXiv preprint arXiv:2602.06028 , year=

    Context Forcing: Consistent Autoregressive Video Generation with Long Context , author=. arXiv preprint arXiv:2602.06028 , year=

  24. [32]

    arXiv preprint arXiv:2602.02393 , year=

    Infinite-World: Scaling Interactive World Models to 1000-Frame Horizons via Pose-Free Hierarchical Memory , author=. arXiv preprint arXiv:2602.02393 , year=

  25. [33]

    arXiv preprint arXiv:2603.21366 , year=

    Relax forcing: Relaxed kv-memory for consistent long video generation , author=. arXiv preprint arXiv:2603.21366 , year=

  26. [34]

    Self forcing: Bridging the train-test gap in autoregressive video diffusion , author=

  27. [35]

    arXiv preprint arXiv:2510.09212 , year=

    Stable video infinity: Infinite-length video generation with error recycling , author=. arXiv preprint arXiv:2510.09212 , year=

  28. [36]

    arXiv preprint arXiv:2603.04379 , year=

    Helios: Real Real-Time Long Video Generation Model , author=. arXiv preprint arXiv:2603.04379 , year=

  29. [37]

    arXiv preprint arXiv:2509.25161 , year=

    Rolling forcing: Autoregressive long video diffusion in real time , author=. arXiv preprint arXiv:2509.25161 , year=

    arXiv

  30. [38]

    arXiv preprint arXiv:2605.20476 , year=

    Goodbye Drift: Anchored Tree Sampling for Long-Horizon Video-to-Video Generation , author=. arXiv preprint arXiv:2605.20476 , year=

    arXiv

  31. [39]

    Worldweaver: Generating long-horizon video worlds via rich perception , author=

  32. [40]

    Endless World: Real-Time 3D-Aware Long Video Generation , author=

  33. [41]

    arXiv preprint arXiv:2202.00512 , year=

    Progressive distillation for fast sampling of diffusion models , author=. arXiv preprint arXiv:2202.00512 , year=

    arXiv

  34. [42]

    Consistency Models , author=

  35. [43]

    arXiv preprint arXiv:2310.04378 , year=

    Latent consistency models: Synthesizing high-resolution images with few-step inference , author=. arXiv preprint arXiv:2310.04378 , year=

    arXiv

  36. [44]

    One-step diffusion with distribution matching distillation , author=

  37. [45]

    Improved distribution matching distillation for fast image synthesis , author=

  38. [46]

    arXiv preprint arXiv:2312.09109 , year=

    Videolcm: Video latent consistency model , author=. arXiv preprint arXiv:2312.09109 , year=

    arXiv

  39. [47]

    SIGGRAPH Asia 2024 Technical Communications , year=

    AnimateLCM: Accelerating the Animation of Personalized Diffusion Models and Adapters with Decoupled Consistency Learning , author=. SIGGRAPH Asia 2024 Technical Communications , year=

    2024

  40. [48]

    T2v-turbo: Breaking the quality bottleneck of video consistency model with mixed reward feedback , author=

  41. [49]

    arXiv preprint arXiv:2309.03549 , year=

    Reuse and diffuse: Iterative denoising for text-to-video generation , author=. arXiv preprint arXiv:2309.03549 , year=

    arXiv

  42. [50]

    Fifo-diffusion: Generating infinite videos from text without training , author=

  43. [51]

    Streamingt2v: Consistent, dynamic, and extendable long video generation from text , author=

  44. [52]

    Streamdit: Real-time streaming text-to-video generation , author=

  45. [53]

    From slow bidirectional to fast causal video generators , author=

  46. [54]

    arXiv preprint arXiv:2511.01266 , year=

    Motionstream: Real-time video generation with interactive motion controls , author=. arXiv preprint arXiv:2511.01266 , year=

  47. [55]

    arXiv preprint arXiv:2509.22622 , year=

    Longlive: Real-time interactive long video generation , author=. arXiv preprint arXiv:2509.22622 , year=

    arXiv

  48. [56]

    Real-time video generation with pyramid attention broadcast , author=

  49. [57]

    Fastercache: Training-free video diffusion model acceleration with high quality , author=

  50. [58]

    arXiv preprint arXiv:2208.01626 , year=

    Prompt-to-Prompt Image Editing with Cross Attention Control , author=. arXiv preprint arXiv:2208.01626 , year=

    arXiv

  51. [59]

    Video-P2P: Video Editing with Cross-attention Control , author=

  52. [60]

    arXiv preprint arXiv:2508.13009 , year=

    Matrix-game 2.0: An open-source real-time and streaming interactive world model , author=. arXiv preprint arXiv:2508.13009 , year=

    arXiv

  53. [61]

    Nature , volume=

    World and human action models towards gameplay ideation , author=. Nature , volume=. 2025 , publisher=

    2025

  54. [62]

    arXiv preprint arXiv:2601.20540 , year=

    Advancing Open-source World Models , author=. arXiv preprint arXiv:2601.20540 , year=

    arXiv

  55. [63]

    arXiv preprint arXiv:2511.23429 , year=

    Hunyuan-gamecraft-2: Instruction-following interactive game world model , author=. arXiv preprint arXiv:2511.23429 , year=

  56. [64]

    arXiv preprint arXiv:2604.08995 , year=

    Matrix-game 3.0: Real-time and streaming interactive world model with long-horizon memory , author=. arXiv preprint arXiv:2604.08995 , year=

    arXiv

  57. [65]

    arXiv preprint arXiv:2506.17201 , year=

    Hunyuan-gamecraft: High-dynamic interactive game video generation with hybrid history condition , author=. arXiv preprint arXiv:2506.17201 , year=

    arXiv

  58. [66]

    Yume1.5: A Text-Controlled Interactive World Generation Model , author=

  59. [67]

    arXiv preprint arXiv:2504.08388 , year=

    Mineworld: a real-time and open-source interactive world model on minecraft , author=. arXiv preprint arXiv:2504.08388 , year=

    arXiv

  60. [68]

    Gamefactory: Creating new games with generative interactive videos , author=

  61. [69]

    The matrix: Infinite-horizon world generation with real-time moving control , author=

  62. [70]

    Gamegen-x: Interactive open-world game video generation , author=

  63. [71]

    arXiv preprint arXiv:2412.00887 , year=

    Playable game generation , author=. arXiv preprint arXiv:2412.00887 , year=

    arXiv

  64. [72]

    ACM Transactions on Graphics , volume=

    Promptable game models: Text-guided game simulation via masked diffusion models , author=. ACM Transactions on Graphics , volume=. 2024 , publisher=

    2024

  65. [73]

    2024 , howpublished=

    Oasis: A universe in a transformer , author=. 2024 , howpublished=

    2024

  66. [74]

    Diffusion for world modeling: Visual details matter in atari , author=

  67. [75]

    Diffusion models are real-time game engines , author=

  68. [76]

    2024 , howpublished=

    Genie 2: A large-scale foundation world model , author=. 2024 , howpublished=

    2024

  69. [77]

    Genie: Generative interactive environments , author=

  70. [78]

    World model on million-length video and language with blockwise ringattention , author=

  71. [79]

    arXiv preprint arXiv:2501.03575 , year=

    Cosmos world foundation model platform for physical ai , author=. arXiv preprint arXiv:2501.03575 , year=

    arXiv

  72. [80]

    arXiv preprint arXiv:2410.13720 , year=

    Movie gen: A cast of media foundation models , author=. arXiv preprint arXiv:2410.13720 , year=

    arXiv

  73. [81]

    arXiv preprint arXiv:2505.13211 , year=

    Magi-1: Autoregressive video generation at scale , author=. arXiv preprint arXiv:2505.13211 , year=

    arXiv

  74. [82]

    arXiv preprint arXiv:2502.10248 , year=

    Step-video-t2v technical report: The practice, challenges, and future of video foundation model , author=. arXiv preprint arXiv:2502.10248 , year=

    arXiv

  75. [83]

    arXiv preprint arXiv:2501.00103 , year=

    Ltx-video: Realtime video latent diffusion , author=. arXiv preprint arXiv:2501.00103 , year=

    arXiv

  76. [84]

    arXiv preprint arXiv:2503.20314 , year=

    Wan: Open and advanced large-scale video generative models , author=. arXiv preprint arXiv:2503.20314 , year=

    arXiv

  77. [85]

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

  78. [86]

    arXiv preprint arXiv:2412.03603 , year=

    Hunyuanvideo: A systematic framework for large video generative models , author=. arXiv preprint arXiv:2412.03603 , year=

    arXiv

  79. [87]

    arXiv preprint arXiv:2412.00131 , year=

    Open-sora plan: Open-source large video generation model , author=. arXiv preprint arXiv:2412.00131 , year=

    arXiv

  80. [88]

    arXiv preprint arXiv:2412.20404 , year=

    Open-sora: Democratizing efficient video production for all , author=. arXiv preprint arXiv:2412.20404 , year=

    arXiv

  81. [89]

    OpenAI Blog , year=

    Video generation models as world simulators , author=. OpenAI Blog , year=

  82. [90]

    Scalable diffusion models with transformers , author=

  83. [91]

    High-resolution image synthesis with latent diffusion models , author=

  84. [92]

    2025 , howpublished=

    Genie 3: A new frontier for world models , author=. 2025 , howpublished=

    2025

  85. [93]

    2025 , howpublished=

    WHAMM! Real-time world modelling of interactive environments , author=. 2025 , howpublished=

    2025

  86. [94]

    2024 , howpublished=

    Veo , author=. 2024 , howpublished=

    2024

  87. [95]

    2024 , howpublished=

    Kling , author=. 2024 , howpublished=

    2024

  88. [96]

    2024 , howpublished=

    Introducing Gen-3 Alpha , author=. 2024 , howpublished=

    2024

  89. [97]

    arXiv preprint arXiv:2605.25874 , year=

    WBench: A Comprehensive Multi-turn Benchmark for Interactive Video World Model Evaluation , author=. arXiv preprint arXiv:2605.25874 , year=

    arXiv

  90. [98]

    arXiv preprint arXiv:2604.21686 , year=

    WorldMark: A Unified Benchmark Suite for Interactive Video World Models , author=. arXiv preprint arXiv:2604.21686 , year=

    arXiv

  91. [99]

    2026 , howpublished=

    Happy Oyster , author=. 2026 , howpublished=

    2026

  92. [100]

    arXiv preprint arXiv:2606.04527 , year=

    Echo-Infinity: Learnable Evolving Memory for Real-Time Infinite Video Generation , author=. arXiv preprint arXiv:2606.04527 , year=

    arXiv

  93. [101]

    arXiv preprint arXiv:2605.15178 , year=

    Sana-wm: Efficient minute-scale world modeling with hybrid linear diffusion transformer , author=. arXiv preprint arXiv:2605.15178 , year=

    arXiv

  94. [102]

    2026 , month =

    JoyAI-Echo: Pushing the Frontier of Long Video Generation , author =. 2026 , month =

    2026

  95. [103]

    arXiv preprint arXiv:2606.02800 , year=

    Cosmos 3: Omnimodal world models for physical ai , author=. arXiv preprint arXiv:2606.02800 , year=

    arXiv

  96. [104]

    arXiv preprint arXiv:2604.02329 , year=

    Generative World Renderer , author=. arXiv preprint arXiv:2604.02329 , year=

  97. [105]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Diffusion renderer: Neural inverse and forward rendering with video diffusion models , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  98. [106]

    arXiv preprint arXiv:2512.16670 , year=

    FrameDiffuser: G-Buffer-Conditioned Diffusion for Neural Forward Frame Rendering , author=. arXiv preprint arXiv:2512.16670 , year=

  99. [107]

    arXiv preprint arXiv:2601.06928 , year=

    RenderFlow: Single-Step Neural Rendering via Flow Matching , author=. arXiv preprint arXiv:2601.06928 , year=

    arXiv

  100. [108]

    arXiv preprint arXiv:2604.09304 , year=

    GeRM: A Generative Rendering Model From Physically Realistic to Photorealistic , author=. arXiv preprint arXiv:2604.09304 , year=

    arXiv

  101. [109]

    arXiv preprint arXiv:2506.10600 , year=

    Embodiedgen: Towards a generative 3d world engine for embodied intelligence , author=. arXiv preprint arXiv:2506.10600 , year=

    arXiv

  102. [110]

    arXiv preprint arXiv:2411.10109 , volume=

    Generative agent simulations of 1,000 people , author=. arXiv preprint arXiv:2411.10109 , volume=

    arXiv

  103. [111]

    Sage: Scalable agentic 3d scene generation for embodied ai , author=

  104. [112]

    arXiv preprint arXiv:2605.03941 , year=

    iWorld-Bench: A Benchmark for Interactive World Models with a Unified Action Generation Framework , author=. arXiv preprint arXiv:2605.03941 , year=

    arXiv

  105. [113]

    Advances in neural information processing systems , volume=

    Improved distribution matching distillation for fast image synthesis , author=. Advances in neural information processing systems , volume=

  106. [114]

    arXiv preprint arXiv:2510.02283 , year=

    Self-forcing++: Towards minute-scale high-quality video generation , author=. arXiv preprint arXiv:2510.02283 , year=

    arXiv

  107. [115]

    arXiv preprint arXiv:2510.08431 , year=

    Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency , author=. arXiv preprint arXiv:2510.08431 , year=

    arXiv

  108. [116]

    arXiv preprint arXiv:2606.25473 , year=

    Causal-rCM: A Unified Teacher-Forcing and Self-Forcing Open Recipe for Autoregressive Diffusion Distillation in Streaming Video Generation and Interactive World Models , author=. arXiv preprint arXiv:2606.25473 , year=

    arXiv

  109. [117]

    arXiv preprint arXiv:2606.11187 , year=

    Next Forcing: Causal World Modeling with Multi-Chunk Prediction , author=. arXiv preprint arXiv:2606.11187 , year=

    arXiv

  110. [118]

    arXiv preprint arXiv:2506.15675 , year=

    Sekai: A video dataset towards world exploration , author=. arXiv preprint arXiv:2506.15675 , year=

  111. [119]

    arXiv preprint arXiv:2509.09676 , year=

    Spatialvid: A large-scale video dataset with spatial annotations , author=. arXiv preprint arXiv:2509.09676 , year=

  112. [120]

    ACM Transactions on Graphics (SIGGRAPH) , volume=

    Stereo magnification: Learning view synthesis using multiplane images , author=. ACM Transactions on Graphics (SIGGRAPH) , volume=

  113. [121]

    Dl3dv-10k: A large-scale scene dataset for deep learning-based 3d vision , author=

  114. [122]

    arXiv preprint arXiv:2508.10934 , year=

    Vipe: Video pose engine for 3d geometric perception , author=. arXiv preprint arXiv:2508.10934 , year=

    arXiv

  115. [123]

    Raft: Recurrent all-pairs field transforms for optical flow , author=

  116. [124]

    arXiv preprint arXiv:2604.24762 , year=

    Omnishotcut: Holistic relational shot boundary detection with shot-query transformer , author=. arXiv preprint arXiv:2604.24762 , year=

    arXiv

  117. [125]

    EasyOCR: Ready-to-use OCR with 80+ supported languages , author=

  118. [126]

    arXiv preprint arXiv:2410.17725 , year=

    Yolov11: An overview of the key architectural enhancements , author=. arXiv preprint arXiv:2410.17725 , year=

    arXiv

  119. [127]

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

    Cover: A comprehensive video quality evaluator , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops , pages=

  120. [128]

    Vbench: Comprehensive benchmark suite for video generative models , author=

  121. [129]

    arXiv preprint arXiv:2502.14786 , year=

    Siglip 2: Multilingual vision-language encoders with improved semantic understanding, localization, and dense features , author=. arXiv preprint arXiv:2502.14786 , year=

    arXiv

  122. [130]

    Learning transferable visual models from natural language supervision , author=

  123. [131]

    arXiv preprint arXiv:2506.09985 , year=

    V-jepa 2: Self-supervised video models enable understanding, prediction and planning , author=. arXiv preprint arXiv:2506.09985 , year=

    arXiv

  124. [132]

    arXiv preprint arXiv:2507.06261 , year=

    Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities , author=. arXiv preprint arXiv:2507.06261 , year=

    arXiv

  125. [133]

    arXiv preprint arXiv:2503.19786 , year=

    Gemma 3 technical report , author=. arXiv preprint arXiv:2503.19786 , year=

    arXiv

  126. [134]

    arXiv:2605.30060 , year=

    Towards Consistent Video Geometry Estimation , author=. arXiv:2605.30060 , year=

    arXiv

  127. [135]

    arXiv preprint arXiv:2511.10647 , year=

    Depth Anything 3: Recovering the visual space from any views , author=. arXiv preprint arXiv:2511.10647 , year=

    arXiv

Pith tools

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