Pith. sign in

REVIEW 5 major objections 7 minor 26 references

HALLELUAI makes a closed loop of source-image-conditioned moderation and targeted regeneration the gate for production-grade image-to-video output.

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 04:00 UTC pith:Z2UXWO3H

load-bearing objection Coherent industrial system paper whose closed-loop moderation-plus-regeneration design is genuinely useful, but the hallucination-aware claim rests on an unvalidated GPT-4o component and thin evidence. the 5 major comments →

arxiv 2607.22959 v1 pith:Z2UXWO3H submitted 2026-07-25 cs.CV cs.AI

HALLELUAI: A Hallucination-Aware AI System for Ultra-Realistic Image-to-Video Generation at Scale

Aniket Sakpal , Yang Jiang , Rouzbeh Davoudi , Shayan Hassantabar , Mani Najmabadi This is my paper
classification cs.CV cs.AI
keywords image-to-video generationhallucination detectionvideo moderationsource-image fidelityagentic regenerationquality of experiencegenerative video quality controlmultimodal LLM evaluation
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 image-to-video generation can be made production-safe by pairing an automated moderation gate with an agentic regeneration loop. The moderation module scores each video relative to its source image on frame-level quality, camera motion, and hallucination, producing structured failure codes and a PASS/FAIL decision. The regeneration module turns those codes into targeted fixes — prompt refinement, camera adjustment, base-image substitution, or model switching — and iterates until the video passes or resources run out. In human-in-the-loop tests the system reached 86.9% PASS/FAIL agreement with experts in shadow mode and 97% precision on 1,158 system-approved clips, which the authors present as evidence of production-grade gating. The pith is that per-asset, source-image-conditioned moderation plus targeted regeneration can replace manual review at scale in high-trust domains like travel and real estate.

Core claim

The central claim is that hallucination-aware moderation, keyed to the source image, acts as both gate and diagnostic engine for image-to-video output. The paper separates object hallucination (drift, fusion, deformation, disappearance, splitting, text corruption) from new-structure hallucination (natural, camera-induced, or abrupt entry), and asks a multimodal language model to score each category 0–10 and to make a per-frame binary call on new structures. These scores, with frame-signal and motion checks, drive a PASS/FAIL decision and action codes. The regeneration module converts codes into targeted interventions — prompt refinement, camera adjustment, base-image substitution, model swit

What carries the argument

The load-bearing object is the structured moderation report: a tuple of failure codes, category-wise severity scores (0–10), risk levels, and rationale text that serves both as a gate and as input to a planning agent. Hallucination detection rests on a taxonomy of seven object-hallucination categories and three new-structure entry types, applied by a multimodal language model that tracks source-image objects across sampled frames. Around it sit classical signal checks (Laplacian variance for blur and noise, grayscale mean and standard deviation for exposure and contrast), dense optical flow for motion intensity and smoothness, and point-trajectory alignment against prompt intent. The regener

Load-bearing premise

The system's quality gate rests on judgments made by a multimodal language model — the per-category severity scores and per-frame new-structure flags — and those judgments were tuned on 58 calibration clips and never checked against independent, per-category hallucination labels.

What would settle it

Have independent annotators score the same 146 shadow-mode clips on the paper's seven object-hallucination and three new-structure categories, then compute per-category agreement with the system's 0–10 severity scores and binary flags. If per-category agreement is low, the reported 86.9% PASS/FAIL agreement and 97% approved-clip precision may be driven by non-hallucination failures, leaving the hallucination gate itself unvalidated.

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

If this is right

  • Per-asset gating becomes feasible: each video is judged against its own source image, so a single failing asset can be blocked or fixed before release instead of only scoring a model's average output.
  • Failures become programmable: structured codes map to regeneration levers, so the loop replaces blind retries with targeted corrective actions.
  • The system can serve high-trust domains: at 97% precision on 1,158 approved clips, the authors position it as a practical gate for travel, real-estate, and e-commerce placements.
  • The taxonomy supplies a shared vocabulary for creative guidelines, letting expert standards be encoded as thresholds, risk levels, and remediation policies.

Where Pith is reading between the lines

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

  • An extension left implicit: the same hallucination taxonomy could gate text-to-video output if the source image were replaced by a parsed scene graph from the prompt.
  • Per-category severity scores could be published as calibration curves per failure type, letting clients set their own risk thresholds; the paper reports only aggregate precision.
  • The 70,000+ videos produced in deployment, if audited per failure code, would reveal which regeneration levers actually fix each failure class — a benchmark no current evaluation suite offers.
  • The architecture is not limited to generative media; pointed at live-action footage, the same source-fidelity gate could screen for deepfakes or brand-safety violations.

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

5 major / 7 minor

Summary. The manuscript proposes HALLELUAI, a closed-loop moderation-and-regeneration system for image-to-video (I2V) generation. The moderation module is a three-part gate: deterministic frame-level quality checks (blur, contrast, brightness, noise; Algorithm 1), deterministic motion checks based on RAFT and TAPIR (Algorithm 2), and an MLLM-based hallucination detector that rates seven object-hallucination categories and three new-structure cases (Section 3.3). The regeneration module translates structured diagnostics into prompt, camera, image, or model changes (Section 4). Evaluation is human-in-the-loop with creative experts: 86.9% PASS/FAIL agreement and 88% precision on 146 shadow clips, 97% expert precision on 1,158 pseudo-production clips, and a production deployment claimed at 70,000+ videos. The paper claims this is the first integrated system that provides per-asset, source-image-conditioned, machine-actionable quality control for I2V at scale.

Significance. The problem is real and the modular design is sensible. If the results hold, HALLELUAI would fill a practical gap by producing structured diagnostics that directly drive remediation rather than a scalar quality score. The deterministic algorithms are specified with explicit thresholds, and the use of external expert review in the loop is a strength. What is not yet demonstrated is the hallucination-aware core: the GPT-4o category scores and new-structure labels are never validated against independent human annotations, so the central claim that the system detects fine-grained hallucinations is unsupported. The evaluation also lacks confidence intervals, inter-annotator agreement, a regeneration-control comparison, and released data/code. These are remediable, but until they are supplied the paper should be treated as a system description with promising pilot validation rather than a validated production claim.

major comments (5)
  1. [§3.3.2, §5.1] The hallucination-detection layer is the advertised differentiator, but its intermediate outputs are never validated. In §3.3.2, GPT-4o emits per-category severity scores S_o^i and per-frame labels d_new; §3.4 thresholds these to PASS/FAIL and §4 uses them to select regeneration actions. The only check is aggregate PASS/FAIL agreement (Table 1), which also includes the deterministic frame/motion modules; the 86.9%/88% figures could be produced by those modules while the MLLM adds noise. Calibration on 58 clips does not establish generalization for per-category hallucination judgments. Please report per-category human annotations, confusion matrices for object/new-structure categories, and an ablation with the hallucination module disabled.
  2. [§5.1, Tables 1 and 3] The shadow evaluation is statistically thin. It uses 146 clips with no confidence intervals, no significance tests against baselines, and no inter-annotator reliability for the expert labels; expert PASS/FAIL is treated as ground truth. At this sample size the 86.9% agreement and 88% precision have wide intervals, and Table 3 reports no significance tests. Report confidence intervals, Cohen's kappa (or equivalent), the number of experts per clip, and the distribution of base I2V models and prompts so the reader can judge generalization.
  3. [§5.2] The 97% pseudo-production precision cannot be attributed to regeneration. Because only system-approved clips are reviewed, the experiment measures gate precision, not the causal effect of the regeneration loop. The sentence in §5.2 that the system is 'actively improving clips' is not supported without a control arm (e.g., regeneration disabled or random retries) and without reporting iteration counts, budget consumption, and per-action outcomes. Please supply a controlled comparison and the distribution of accepted clips by number of regeneration rounds.
  4. [§5.1.1, Table 2] Table 2 and its surrounding text are internally ambiguous. The caption says 'contribution to Precision and False Positives' but the column is labeled 'Precision'; entries are 97–100% for hallucination categories, while the text says the majority of precision-impacting errors arise from object-level hallucinations. If these are per-signal precision values, the hallucination module is not the dominant error source; if they are error proportions, the column is mislabeled. Because diagnostics in Section 4 depend on failure codes, report per-signal error counts and conditional rates with confidence intervals.
  5. [§3.3.1] Section 3.3.1 includes 'natural entry' and 'camera-induced entry' as new-structure hallucinations. In a camera pan, tilt, or dolly, previously off-screen content entering the frame is physically expected and is not an error relative to a coherent scene. Since d_new is an OR of all three cases and triggers FAIL, the system may systematically over-reject the camera-motion videos it explicitly claims to support. Please justify this definition with expert evidence or revise it.
minor comments (7)
  1. [Abstract] There is a stray semicolon in 'automated; high-precision quality control'; use a comma or em dash.
  2. [Fig. 2] The caption uses 'VLLM' while the text uses 'MLLM'; unify the terminology.
  3. [Algorithm 2 (B5)] The alignment score s_align is defined as a 0/1 indicator, making thresholding by τ_align meaningless unless the classifier emits a probability. Provide a continuous alignment score or remove the threshold.
  4. [§5.1] Shadow Mode is described as 'calibration' although calibration was performed on a separate 58-clip set; rephrase to avoid confusion.
  5. [§5.3] The 70,000+ video production claim is not accompanied by sampled precision numbers or monitoring details; add the protocol and results.
  6. [References] References [12] and [13] are incomplete author lists ('Liu et al.'); provide full citations.
  7. [Abstract/§1] The abstract promises end-user QoE and ultra-realism, but the evaluation measures expert pass/fail only; qualify this claim or add a user/QoE study.

Circularity Check

0 steps flagged

No significant circularity: the core alignment claim is anchored to external expert judgments on held-out clips.

full rationale

The paper's central result is the moderation system's agreement with creative-expert PASS/FAIL decisions on 146 unseen Shadow Mode clips (86.9% agreement, 88% precision) and 97% expert precision on 1,158 system-approved clips in Pseudo Production. These are external human judgments, not quantities derived from the system's own fitted inputs by construction. Thresholds and prompts were calibrated on a separate 58-clip set, and the evaluation is performed on disjoint clips, so the reported agreement is a genuine held-out measurement rather than a renamed fit. The hallucination taxonomy and GPT-4o prompt are authored by the same team, which raises a validity/bias concern about the intermediate per-category severity scores, but the paper does not define its final prediction in terms of those intermediate labels in a way that forces agreement; there is no equation in which a reported output reduces to its own input. The paper contains no load-bearing self-citations: cited references are external prior work, and the 'first integrated framework' claim is not justified by a uniqueness theorem from the authors. The closest issue—unvalidated per-category MLLM hallucination judgments—is an evidence gap about measurement trustworthiness, not a circular derivation. Consequently, no specific circular step can be quoted, and the appropriate score is 0.

Axiom & Free-Parameter Ledger

5 free parameters · 7 axioms · 0 invented entities

The central claim rests on several unvalidated tooling assumptions: RAFT/TAPIR motion estimates on synthetic video, GPT-4o as a ground-truth hallucination detector, and the completeness of the hand-authored taxonomy. Combined with unreported thresholds calibrated on 58 clips, these assumptions mean the reported agreement numbers cannot be independently checked.

free parameters (5)
  • Frame-level thresholds (τ_blur, τ↓, τ↑, τB↓, τB↑, τnoise, τp) = not reported; calibrated on 58 clips
    Algorithm 1's PASS/FAIL gates for blur, contrast, exposure, and noise all depend on these tuned values; they are not derived from theory.
  • Motion thresholds (τ_low, τ_high, τ_jit, τ_align) and top-pixel fraction q=0.15 = not reported (q=0.15 stated)
    Algorithm 2's Low/High Motion, Jitter, and Prompt Alignment decisions depend on thresholds calibrated on 58 clips.
  • Hallucination category thresholds τ_i_object and severity aggregation weights = not reported; 'optimized to maximize sensitivity' on 58 clips
    Section 3.3.2 thresholds per-category severity scores from GPT-4o; Section 3.4 aggregates them via domain-calibrated weights.
  • Unified verdict scoring weights = not reported
    Section 3.4 combines frame, temporal, and hallucination scores via 'domain-calibrated weights' with no values or fitting procedure given.
  • Hallucination-detection prompt template = GPT-4o prompt shown in Fig. 2; iteratively refined on 58 clips
    The prompt determines the categories and severity scores; it was selected by maximizing recall on the calibration set, making it a tuned input to the system.
axioms (7)
  • domain assumption RAFT optical flow gives reliable motion estimates on AI-generated video.
    Section 3.2.1 uses RAFT to estimate per-pixel displacement without validating it on synthetic artifacts typical of diffusion outputs.
  • domain assumption TAPIR point tracking trajectories reflect camera motion rather than object motion confounds.
    Section 3.2.2 relies on edge-biased TAPIR points to infer pan/tilt/zoom; the separation of camera and object motion is assumed, not demonstrated.
  • domain assumption GPT-4o hallucination judgments are an adequate proxy for ground-truth hallucination annotations.
    Section 3.3.2 uses MLLM severity scores and binary new-structure decisions to gate videos, with no independent validation of these labels.
  • domain assumption Grayscale variance, mean, and standard deviation correspond to perceived blur, exposure, and contrast.
    Section 3.1 defines quality signals via simple pixel statistics without perceptual calibration against human ratings.
  • ad hoc to paper The hand-authored hallucination taxonomy covers all relevant production failure modes.
    The taxonomy in Section 3.3.1 is introduced for this system; its completeness is assumed and never tested against an independent failure taxonomy.
  • domain assumption The 58-clip calibration set and 146-clip shadow set are representative of production I2V outputs.
    Section 5 gives no distributional description of these clips, so representativeness is assumed.
  • domain assumption Creative expert reviewers agree consistently with each other.
    Expert judgments are used as ground truth in Section 5.1, but no inter-annotator agreement is reported.

pith-pipeline@v1.3.0-alltime-deepseek · 9692 in / 14687 out tokens · 129685 ms · 2026-08-01T04:00:47.963466+00:00 · methodology

0 comments
read the original abstract

AI-generated video is increasingly used across marketing, product storytelling, and creative workflows, yet automated; high-precision quality control remains a major constraint to scaling production. We present HALLELUAI, an end-to-end system that moderates and regenerates image-to-video outputs to meet expert-level creative standards and deliver ultra-realistic videos with consistent end-user quality of experience (QoE) at scale. The system integrates a video moderation module that evaluates frame-level aesthetics, temporal motion fidelity, and fine-grained hallucination risks relative to the source image, with an agentic regeneration module that iteratively fixes failures through prompt refinement, controlled camera adjustments, targeted model or image switching, and structured retry strategies. The moderation logic is aligned with domain-specific creative guidelines and produces granular, machine-actionable feedback that directly drives regeneration. In human-in-the-loop evaluations with creative experts, HALLELUAI shows strong alignment and reliably outputs ultra-realistic, production-grade videos suitable for product and marketing placements at scale. This framework advances trustworthy AI generated video content by enforcing visual realism, brand safety, and strict input-image fidelity while enabling image-to-video generation at scale.

Figures

Figures reproduced from arXiv: 2607.22959 by Aniket Sakpal, Mani Najmabadi, Rouzbeh Davoudi, Shayan Hassantabar, Yang Jiang.

Figure 1
Figure 1. Figure 1: Architectural overview of the proposed closed-loop system for image-to-video generation. The framework [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Prompt template provided to the VLLM for hal [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Hallucination detection module for Image2Video generation. Uniformly sampled frames are analyzed [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Evaluation Prompt Designed for MLLMs to Evaluate Frame-level Quality,Temporal Motion Quality and Hallucination in Image2Video Generation 5.2 Stage 2: Pseudo Production (Gating Performance) In Pseudo Production mode, we evaluate the combined moderation + regeneration system end-to-end by sending only system-approved clips for human creative review. Of 1,158 system-approved clips, 1,126 were also accepted by… view at source ↗

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

26 extracted references · 7 linked inside Pith

  1. [1]

    Understanding hallucina- tionsindiffusionmodelsthroughmodeinterpolation

    Sumukh K Aithal, Pratyush Maini, Zachary C Lip- ton, and J Zico Kolter. Understanding hallucina- tionsindiffusionmodelsthroughmodeinterpolation. Advances in neural information processing systems, 37:134614–134644, 2024

    2024

  2. [2]

    A note on the inception score.arXiv preprint arXiv:1801.01973, 2018

    Shane Barratt and Rishi Sharma. A note on the inception score.arXiv preprint arXiv:1801.01973, 2018. 8 Table 2: Stage-level contribution to Precision and False Positives (FP). Percentages reflect the proportion of total observed errors attributable to each signal. Module Sub-Category Precision Frame-Level Quality Blur 100% Contrast 97% Brightness 100% Noi...

    Pith/arXiv arXiv 2018

  3. [3]

    Sora detec- tor: A unified hallucination detection for large text- to-video models.arXiv preprint arXiv:2405.04180, 2024

    Zhixuan Chu, Lei Zhang, Yichen Sun, Siqiao Xue, Zhibo Wang, Zhan Qin, and Kui Ren. Sora detec- tor: A unified hallucination detection for large text- to-video models.arXiv preprint arXiv:2405.04180, 2024

    Pith/arXiv arXiv 2024

  4. [4]

    Cover: A compre- hensive video quality evaluator

    Chenlong He, Qi Zheng, Ruoxi Zhu, Xiaoyang Zeng, Yibo Fan, and Zhengzhong Tu. Cover: A compre- hensive video quality evaluator. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5799–5809, 2024

    2024

  5. [5]

    Videoscore: Building automatic metrics to simulate fine-grained human feedback for video generation

    Xuan He, Dongfu Jiang, Ge Zhang, Max Ku, Achint Soni, Sherman Siu, Haonan Chen, Abhranil Chan- dra, Ziyan Jiang, Aaran Arulraj, et al. Videoscore: Building automatic metrics to simulate fine-grained human feedback for video generation. InProceedings of the 2024 Conference on Empirical Methods in Nat- ural Language Processing, pages 2105–2123, 2024

    2024

  6. [6]

    Gans trained by a two time-scale update rule con- verge to a local nash equilibrium.Advances in Neural Information Processing Systems, 30, 2017

    Martin Heusel, Hubert Ramsauer, Thomas Un- terthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule con- verge to a local nash equilibrium.Advances in Neural Information Processing Systems, 30, 2017

    2017

  7. [7]

    Vbench: Comprehensive benchmark suite for video generative models

    Ziqi Huang, Yinan He, Jiashuo Yu, Fan Zhang, Chenyang Si, Yuming Jiang, Yuanhan Zhang, Tianx- ing Wu, Qingyang Jin, Nattapol Chanpaisit, et al. Vbench: Comprehensive benchmark suite for video generative models. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition, pages 21807–21818, 2024

    2024

  8. [8]

    Language models (mostly) know what they know.arXiv preprint arXiv:2207.05221, 2022

    Saurabh Kadavath et al. Language models (mostly) know what they know.arXiv preprint arXiv:2207.05221, 2022

    Pith/arXiv arXiv 2022

  9. [9]

    Think-then-generate: Reasoning- aware text-to-image diffusion with llm encoders

    Siqi Kou, Jiachun Jin, Zetong Zhou, Ye Ma, Yu- gang Wang, Quan Chen, Peng Jiang, Xiao Yang, Jun Zhu, Kai Yu, et al. Think-then-generate: Reasoning- aware text-to-image diffusion with llm encoders. arXiv preprint arXiv:2601.10332, 2026

    arXiv 2026

  10. [10]

    Multilevel semantic-aware model for ai-generated video quality assessment

    Jiaze Li, Haoran Xu, Shiding Zhu, Junwei He, and Haozhao Wang. Multilevel semantic-aware model for ai-generated video quality assessment. InICASSP 2025-2025 IEEE International Conference on Acous- tics, Speech and Signal Processing (ICASSP), pages 1–5. IEEE, 2025

    2025

  11. [11]

    Evaluating text-to-visual generation with image-to-text generation

    Zhiqiu Lin, Deepak Pathak, Baiqi Li, Jiayao Li, Xide Xia, Graham Neubig, Pengchuan Zhang, and Deva Ramanan. Evaluating text-to-visual generation with image-to-text generation. InEuropean Conference on Computer Vision, pages 366–384. Springer, 2024

    2024

  12. [12]

    Fréchetvideomotiondistance: Ametricfor evaluating motion consistency in video generation

    Liuetal. Fréchetvideomotiondistance: Ametricfor evaluating motion consistency in video generation. arXiv preprint arXiv:2407.16124, 2024

    Pith/arXiv arXiv 2024

  13. [13]

    Genai-bench: Evaluating and improv- ing compositional text-to-visual generation.arXiv preprint arXiv:2406.13743, 2024

    Liu et al. Genai-bench: Evaluating and improv- ing compositional text-to-visual generation.arXiv preprint arXiv:2406.13743, 2024

    Pith/arXiv arXiv 2024

  14. [14]

    Evalcrafter: Benchmarking and evaluating large video generation models

    Yaofang Liu, Xiaodong Cun, Xuebo Liu, Xintao Wang, YongZhang, HaoxinChen, YangLiu, Tieyong Zeng, Raymond Chan, and Ying Shan. Evalcrafter: Benchmarking and evaluating large video generation models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 22139–22149, 2024

    2024

  15. [15]

    Video diffusion gen- eration: comprehensive review and open problems

    Wenping Ma, Xiaoting Yang, Licheng Jiao, Lingling Li, Xu Liu, Fang Liu, Puhua Chen, Yuting Yang, 9 Mengru Ma, Long Sun, et al. Video diffusion gen- eration: comprehensive review and open problems. Artificial Intelligence Review, 58(11):338, 2025

    2025

  16. [16]

    Controllable video gener- ation: A survey.arXiv preprint arXiv:2507.16869, 2025

    Yue Ma, Kunyu Feng, Zhongyuan Hu, Xinyu Wang, Yucheng Wang, Mingzhe Zheng, Xuanhua He, Chenyang Zhu, Hongyu Liu, Yingqing He, Zeyu Wang, Zhifeng Li, Xiu Li, Wei Liu, Dan Xu, Linfeng Zhang, and Qifeng Chen. Controllable video gener- ation: A survey.arXiv preprint arXiv:2507.16869, 2025

    arXiv 2025

  17. [17]

    Video- bench: A comprehensive benchmark and toolkit for evaluating video-based large language models.Com- putational Visual Media, 2025

    Munan Ning, Bin Zhu, Yujia Xie, Bin Lin, Jiaxi Cui, Lu Yuan, Dongdong Chen, and Li Yuan. Video- bench: A comprehensive benchmark and toolkit for evaluating video-based large language models.Com- putational Visual Media, 2025

    2025

  18. [18]

    Sora: Openai text-to-video model, 2026

    OpenAI. Sora: Openai text-to-video model, 2026. Product documentation and public reports, URL: https://openai.com/research/sora

    2026

  19. [19]

    Con- sisti2v: Enhancing visual consistency for image-to- video generation.arXiv preprint arXiv:2402.04324, 2024

    Weiming Ren, Harry Yang, Ge Zhang, Cong Wei, Xinrun Du, Stephen Huang, and Wenhu Chen. Con- sisti2v: Enhancing visual consistency for image-to- video generation.arXiv preprint arXiv:2402.04324, 2024

    Pith/arXiv arXiv 2024

  20. [20]

    Towards accurate gener- ative models of video: A new metric & challenges

    Thomas Unterthiner et al. Towards accurate gener- ative models of video: A new metric & challenges. InInternational Conference on Learning Represen- tations Workshop, 2018

    2018

  21. [21]

    Aigv-assessor: Bench- marking and evaluating the perceptual quality of text-to-video generation with lmm

    Jiarui Wang, Huiyu Duan, Guangtao Zhai, Juntong Wang, and Xiongkuo Min. Aigv-assessor: Bench- marking and evaluating the perceptual quality of text-to-video generation with lmm. InProceedings of the Computer Vision and Pattern Recognition Con- ference, pages 18869–18880, 2025

    2025

  22. [22]

    Self-consistency improves chain of thought reasoning in language models.arXiv preprint arXiv:2203.11171, 2022

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models.arXiv preprint arXiv:2203.11171, 2022

    Pith/arXiv arXiv 2022

  23. [23]

    Bovik, Hamid R

    Zhou Wang, Alan C. Bovik, Hamid R. Sheikh, and Eero P. Simoncelli. Image quality assessment: From error visibility to structural similarity.IEEE Trans- actions on Image Processing, 13(4):600–612, 2004

    2004

  24. [24]

    Exploring video quality assessment on user generated contents from aesthetic and technical perspectives

    Haoning Wu, Erli Zhang, Liang Liao, Chaofeng Chen, Jingwen Hou, Annan Wang, Wenxiu Sun, Qiong Yan, and Weisi Lin. Exploring video quality assessment on user generated contents from aesthetic and technical perspectives. InProceedings of the IEEE/CVF international conference on computer vi- sion, pages 20144–20154, 2023

    2023

  25. [25]

    Zhichao Zhang, Wei Sun, Li Xinyue, Jun Jia, Xiongkuo Min, Zicheng Zhang, Chunyi Li, Zijian Chen, Wang Puyi, Sun Fengyu, et al. Benchmark- ing multi-dimensional aigc video quality assessment: A dataset and unified model.ACM Transactions on Multimedia Computing, Communications and Appli- cations, 21(9):1–24, 2025

    2025

  26. [26]

    Identifying and solving conditional image leakage in image-to-video diffusion model.Advances in Neural Information Processing Systems, 37:30300–30326, 2024

    Min Zhao, Hongzhou Zhu, Chendong Xiang, Kaiwen Zheng, Chongxuan Li, and Jun Zhu. Identifying and solving conditional image leakage in image-to-video diffusion model.Advances in Neural Information Processing Systems, 37:30300–30326, 2024. 10

    2024