VideoCoCo, an agentic dual-engine framework in which executable Blender code serves as a process-level chain of thought, demonstrates that executable code provides an effective, controllable, and inspectable intermediate representation for physically consistent video generation.
Abstract
Text-to-video models have achieved remarkable visual quality, yet they still struggle to generate physically consistent dynamics because the temporal evolution of a scene must be inferred implicitly from a highly compressed text prompt. Existing chain-of-thought approaches introduce intermediate plans or visual states, but these representations are typically non-executable or temporally sparse, limiting their ability to instantiate and control the complete spatiotemporal process. To address this limitation, we introduce VideoCoCo, an agentic dual-engine framework in which executable Blender code serves as a process-level chain of thought. Given a text prompt, a coding agent synthesizes a Blender program that explicitly specifies the scene and its temporal evolution. The executable simulation engine runs the program to produce a deterministic spatiotemporal draft, which is subsequently transformed into a photorealistic video by a generative video engine through draft-conditioned editing. This decomposition separates process-level reasoning from high-fidelity visual realization. To adapt the video editor to simulated drafts, we construct VideoCoCo-3K, a curated dataset of draft-instruction-target triplets. VideoCoCo improves the OmniWeaving baseline from 0.475 to 0.558 on PhyGenBench and from 52.18 to 77.88 on VBench-2.0, achieving the best average score on both benchmarks. These results demonstrate that executable code provides an effective, controllable, and inspectable intermediate representation for physically consistent video generation.
Experiments on an unseen validation set show that VIPER achieves stronger reference-video physical similarity and higher human preference than representative video generation and video-as-prompt baselines, while maintaining competitive general video quality.
Recent text-to-video systems can generate visually appealing clips from natural language prompts, yet narrative prompts often contain multiple implicit temporal stages that require the generator to infer scene decomposition, subject persistence, action ordering, and visual continuity from a single unstructured input. This frequently leads to temporally inconsistent or structurally ambiguous outputs. In this work, we investigate whether introducing an explicit scene-planning layer can improve multi-stage video generation. We compare three generation paradigms: direct single-prompt generation, naive prompt decomposition, and a structured scene-planning pipeline. The proposed approach first converts a narrative prompt into a lightweight structured scene representation containing global subject information, visual style constraints, and scene-level descriptions, which is then compiled into scene-conditioned prompts for sequential video generation. A reference-guided continuation mechanism conditions the second scene on the final frame of the first scene to improve cross-scene identity and visual continuity. The framework is generator-agnostic and can operate on modern text-to-video backends without modifying the underlying models. To evaluate generation quality, we adopt a vision-language model (VLM) as an automatic judge that assesses prompt relevance, temporal continuity, aesthetic quality, and narrative clarity across candidate videos. This study provides an empirical investigation of how structured intermediate planning influences narrative video generation and offers a lightweight framework for improving temporal coherence in AI-generated videos.
Jing Chen· 2026 International Conferenc...· 0 citations
Action-conditioned video models require large-scale visual data paired with control signals that are temporally aligned with the resulting scene transitions. Such supervision is difficult to obtain from ordinary real-world video because the actions that caused each visual change are typically unknown. We present a large-scale synthetic data production pipeline built on Unreal Engine for generating action-conditioned, multi-view video. To accommodate the different execution requirements of real-time physics and high-quality offline rendering, the pipeline executes trajectory generation and final rendering in two stages: Stage I runs real physics in PIE and records per-frame character states, control inputs, and camera states into an intermediate trajectory representation; Stage II replays those trajectories in a new engine process and renders them offline with Movie Render Queue (MRQ). Around this core, we develop a distributed production system with cache-aware task partitioning, node-local slot scheduling, automated scene screening, aesthetic and luminance filtering, partial-output recovery, asynchronous upload, and continuous cluster health monitoring. The production cluster contains 25 servers with eight NVIDIA RTX 5090 GPUs per server. From 2,384 asset packs, 429 levels were retained for production together with a pool of 40 humanoid characters. The pipeline has produced 2,691 hours of 1080p video and 6,076 hours of 720p video. We describe the system architecture, the implementation decisions that emerged from production failures, and the limitations of using perceptual quality proxies for world-model data curation. The pipeline described in this report constitutes the Unreal Engine synthetic-data production component used in EchoWM.
Hao-Yu Wang, Song-Chun Zhang, Hao-Ran Li et al.· 0 citations
Long-form video understanding requires locating sparse, question-relevant evidence in long, multimodal videos. Real-world video distributions differ in modality-specific information density, content structure, and evidence patterns, causing fixed video-agent designs to incur redundant processing or fail when mismatched. Extending automated agent evolution from text to video is challenging because full long-video execution makes candidate validation expensive, failures propagate across coupled evidence-processing stages, and complex preprocessing, perception tools, and localization strategies make code-level updates difficult to implement reliably. We introduce MetaVideoAgent, a framework that automatically evolves a video agent for a target distribution. It profiles information density and evidence requirements from sparsely sampled frames and associated queries to guide initial design, then compresses localized failures into independently executable minimal validation tasks. It constructs evidence-grounded Gold Paths, audits Student trajectories, aggregates recurring failures across samples, and attributes them to responsible modules. A modular agent representation constrains each update to the primary responsible module and its necessary dependencies. We further introduce VA-EvoBench, covering eight video distributions with separate evolution and held-out splits. With four evolution iterations per distribution, MetaVideoAgent improves every initial agent and raises macro-average accuracy from 38.44% to 51.47%, at an average evolution cost of 3.54M tokens per distribution. The evolved agents outperform the strongest prior fixed-design video agent by 6.39 percentage points while using the fewest tokens and video frames per question among the compared video agents. We will release all code and data to support reproducible research.
Ben-Lei Cui, Ruize Wang, Junjie Li et al.· 1 citation
AlayaWorld is presented, an interactive long-horizon video world model that generates 24-fps video at 540p and 720p and introduces a discrete autoregressive distillation formulation that combines distribution-matching distillation, self-forcing++, and consistency distillation, reducing inference from approximately 30 sampling steps to four steps per chunk.
AlayaWorld Team Kaipeng Zhang, Chuanhao Li, Y. Zhan et al.· arXiv.org· 4 citations
Text-driven instruction-based video editing in complex scenes remains challenging: purely textual prompts often fail to capture precise spatial relationships and physical constraints, resulting in target ambiguity and physically implausible outcomes. To address this, we propose a plan--guide--edit framework that explicitly bridges semantic intent and spatial execution. In our framework, a Chain-of-Thought (CoT)-enhanced multimodal large language model (MLLM) serves as a planner, performing structured reasoning over the video and instructions to derive a precise sequence of bounding boxes and attribute-enriched editing directives. These spatial priors then guide a box-conditioned mask generator, transforming ambiguous global retrieval into localized, context-aware refinement and producing masks that more accurately capture object scale, contact relationships, and placement. Building on these spatial and semantic signals, a diffusion-based editor integrates the masks, enriched instructions, and frame features to render high-fidelity edits that remain temporally coherent and spatially well aligned. Trained first in a modular manner and then jointly, our framework achieves superior performance with reduced data requirements, delivering precise localization in scenes with multiple similar objects and physically consistent object additions, and extensive experiments demonstrate state-of-the-art performance over multiple strong baseline methods. More details are available at: https://github.com/flying-sky999/CoT-Edit
Sen Liang, Fengbin Guan, Youliang Zhang et al.· 5 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
