arXiv subjects

Maneesh Agrawala

Publications and source records attributed to Maneesh Agrawala.

At least 19 recordsLinked to original sources

arXiv2026

RenderFormer-V2: Neural Rendering with Heterogeneous Scene Primitives

We present 'RenderFormer-V2', a unified learned transformer-based neural rendering model, complementary to modern physics-based rendering systems, that can handle diverse light-transport effects such as caustics, volumetric scattering, environment lighting, textured and displaced surfaces and out-of-distribution materials without per-scene training or specialized code. RenderFormer-V2 models global light transport as a sequence-to-sequence transformation. Following its predecessor, RenderFormer-V2 also employs a two stage process: a view-independent stage that resolves intra-scene primitive to primitive transport, and a view-dependent stage that transforms the internal neural scene representation into image pixels. Different from RenderFormer, our model employs a novel combined windowed-attention and rendering-informed attention sink in the view-independent stage to improve scalability while maintaining render accuracy. To further improve versatility, RenderFormerV2 supports heterogeneous scene primitives, including environment maps and participating media, and it employs a material encoding independent of the underlying surface reflectance model that encodes material appearance via a novel neural embedding. We demonstrate the versatility of RenderFormer-V2 on a variety of scenes and perform an extensive ablation of the improved attention mechanism.

cs.CV↗

arXiv2026

Ring Forcing: Towards Precise Long-Term Memory for Autoregressive Video Diffusion

Scaling video generation to long durations reveals a critical bottleneck: current models lack robust long-term memory. This deficiency can be studied along two critical aspects: object permanence, the ability to precisely reproduce the appearance of objects upon re-entry; and memory capacity, the ability to process ultra-long context and use information from distant history. Robust long-term memory requires both: object permanence without sufficient context handling limits the temporal scope, while long context length without permanence fails to maintain identity. To address this, we present Ring Forcing, an autoregressive video diffusion framework designed to robustly construct and precisely utilize long-term memory. Our ring-structured training strategy enforces retrieval from distant history, effectively reconciling the trade-off between strict historical adherence and generative diversity. To expand memory capacity, we introduce a compression and timestep composition strategy. Under fixed sequence length constraints, this method extends the effective historical span to minutes-long durations and achieves a comprehensive receptive field over the entire history. Furthermore, we present a sparse RoPE mechanism to enable flexible, scalable memory adaptation while fully exploiting pre-trained priors. Extensive experiments demonstrate that Ring Forcing achieves superior minutes-long coherence and object permanence, significantly outperforming state-of-the-art methods.

cs.CV↗

arXiv2026

MultiCube: Compositional 3D Generation With Part-Level Semantic and Spatial Control

Digital 3D objects used in games and animation are often required to be compositional; that is, decomposed into semantically meaningful parts. Recent 3D generation methods can produce high-quality compositional objects conditioned on image or text prompts. Yet, such global conditioning lacks the precise part-level controllability required for professional creative workflows. To address this, we introduce MultiCube, a novel compositional 3D generation method that provides explicit, independent control over both the semantics and spatial arrangement of each part. MultiCube takes as input a global text prompt, a text schema specifying the desired parts, and a spatial layout indicating the bounding boxes of the parts in the given schema. It outputs a 3D object composed of distinct meshes, one per specified part, that adhere to the given semantic and spatial conditions. Our approach employs a two-stage diffusion process, first generating a schema- and layout-aligned monolithic mesh, then decomposing the mesh into individual parts simultaneously. A novel Part Layout Adapter is used to encode per-part conditions independently of the other parts. Experiments demonstrate that our method can generate high-quality compositional 3D objects with precise part-level control, including those with unique layouts difficult to achieve with text or image prompting alone. Project page: https://multi-cube.github.io

cs.GR↗

arXiv2026

Collascope: Supporting Serendipitous Asset Exploration for Collage-Based Storytelling

Collage-based storytelling requires visual elements that support emerging narratives and inspire creative reinterpretation. Existing tools, however, rely largely on keyword- and image-based retrieval, offering limited support for serendipitous exploration beyond existing assets. We introduce Collascope, an interactive system that helps creators (1) concretize story intent with interactive element groups, (2) expand the exploration space based on concepts or cutouts towards conceptual and visual dimensions, and (3) develop grounded, traceable ideas in parallel with collage composition. Collascope's attribute-aware visual retrieval method, instantiated with collage-relevant visual dimensions, enables creators to retrieve cutouts through dimension-specific visual projections rather than holistic similarity. In a within-subject study (N=12) against a conventional search baseline, our participants used unexpected results and even gaps in the asset collection to redirect narratives, shift tone, and enrich compositions. Scene Parts helped organize exploration into manageable subtasks, while participants used association in distinct ways depending on whether exploration was guided by a clear goal, an evolving story, or visual intuition.

cs.HC↗

arXiv2026

Masked Visual Actions for Unified World Modeling

Video models absorb rich priors over how the visual world moves, interacts, and responds to contact, making them promising substrates for robotic world modeling. The central challenge is how to communicate action to such models in a form aligned with the visual space in which they learned these interaction priors, yet still grounded in physical manipulation. We introduce Masked Visual Actions, a pixel-space control interface that expresses action as a partially revealed trajectory of an arbitrary entity in a video. Revealing robot motion makes the model act as a forward dynamics model that predicts the scene's response to low-level robot actions, while revealing desired object motion makes the same model recover robot behavior consistent with that outcome. Finetuned with only 15 hours of masked examples from real videos and simulation, a single checkpoint achieves strong visual fidelity and controllability across diverse scenes and multiple embodiments. In downstream manipulation settings, the model produces imagined rollouts whose outcomes correlate with real-world execution for policy evaluation, improves decision making by ranking candidate futures in model-based planning, and supports inverse modeling by synthesizing robot motion from desired object motion.

cs.CV↗

arXiv2026

FourTune: Towards Fully 4-Bit Efficient Post-Training for Diffusion Models

Diffusion models have become a dominant paradigm for high-quality generative modeling, while post-training is essential for adapting them to diverse downstream applications. However, post-training of large diffusion models is still challenging due to the prohibitive memory footprints and slow training speed, which existing parameter-efficient fine-tuning methods only partially address. To overcome these limitations, we propose FourTune, an efficient post-training framework for diffusion models based on an end-to-end W4A4G4 paradigm. FourTune introduces a triple-branch hybrid pipeline that augments the standard LoRA architecture with a frozen numerical stabilizer to isolate quantization-sensitive outliers, enabling stable training under native 4-bit computation. In addition, FourTune employs hardware-efficient block-wise quantization and customized fused kernels to support efficient quantized backpropagation and reduce memory bandwidth overhead. Across customization, reinforcement learning, and distillation tasks, FourTune matches the quality of full-precision fine-tuning. On FLUX.1-dev (12B), FourTune reduces memory overhead by 2.25$\times$ and increases end-to-end training throughput by 2.27$\times$ compared to BF16 LoRA.

cs.LG↗

arXiv2026

Aggregating LLM-Based Weak Verifiers for Spatial Layout Generation

We present a pipeline for building and aggregating task-specific, LLM-generated weak (imperfect) verifiers into a strong verifier for spatial layout domains. Given a task description, our pipeline asks an LLM to synthesize a collection of verifier programs using a layout verification DSL. Each individual LLM-generated verifier usually provides an imperfect check for a match between the layout and the corresponding task description. We show that by aggregating the responses of many such verifiers we can produce a stronger verifier. Moreover, by applying techniques from weak learning, our pipeline can learn how to aggregate the weak verifiers from a very sparse set of human labeled example layouts (about 10). We find that the strong verifiers produced by our pipeline outperform the status-quo approach of using a set of LLM judges to directly check whether a layout matches a task description, raising F1-scores by up to 7X across a variety of 3D room layout and 2D poster design tasks. We also demonstrate that verifier-guided layout generation using natural language feedback from our strong verifiers improves layout quality of a base layout generator by up to 66.2% according to a human evaluator.

cs.GR↗

arXiv2026

What makes an action sequence enjoyable to watch?

People often seek out ways to watch others perform complex action sequences (e.g., sports). What makes some sequences more enjoyable to watch than others? We generated 24 video clips of gameplay from a Flappy Bird-style video game. Clips varied in difficulty (how often players succeeded on average) and in moment-to-moment uncertainty (how likely the player was to crash at any given step). Participants (N=864) rated each video on one of three dimensions: how much they enjoyed it, how difficult the level appeared, or how dangerous the player's trajectory appeared. We found that participants preferred videos where the player seemed to be completing more difficult obstacle courses, but dangerousness did not predict enjoyment ratings. These findings show how procedurally generated stimuli can isolate the factors that affect how enjoyable an action sequence is to watch.

cs.HC↗

arXiv2026

CubePart: An Open-Vocabulary Part-Controllable 3D Generator

Interactive 3D assets used in games and simulation are typically decomposed into specific semantic parts to support animation, physics, and scripted behaviors, yet most generative 3D models produce either monolithic meshes or arbitrary part decompositions that cannot be aligned with application-specific requirements. We present CubePart, a generative framework for open-vocabulary, part-controllable 3D mesh generation that exposes part structure as an explicit inference-time control signal. Given a global text prompt and a user-defined parts schema expressed as an open-ended list of part names, our method generates a set of meshes - one per schema element - that assemble into a coherent object while respecting the specified semantic structure. To enable this capability, we introduce a scalable data pipeline to construct a large open-vocabulary, part-labeled 3D dataset, along with a two-stage generative architecture that separates global shape synthesis from part-level decoding. We demonstrate that the resulting assets can be directly integrated into game engines and driven by animation and behavior scripts without manual post-processing. Project Page: https://cubepart.github.io/

cs.AI↗

arXiv2026

EvalVerse: Pipeline-Aware and Expert-Calibrated Benchmarking for Professional Cinematic Video Generation

The rapid evolution of generative video foundation models has propelled the field toward professional-grade cinematic synthesis. To achieve such demanding quality, the community transitions towards Reinforcement Learning (RL) and agentic workflows. However, reliable evaluation has emerged as a critical bottleneck. Existing benchmarks predominantly evaluate ''whether it is right'' (basic prompt-following) while fundamentally neglecting ''whether it is good'' (cinematic quality, acting, and aesthetics). Furthermore, current automated metrics lack the domain-specific rigor required to provide trustworthy signals, creating a severe credibility gap between human aesthetic perception and machine scoring. To bridge this gap, we introduce EvalVerse, a comprehensive, pipeline-aware, and expert-calibrated evaluation framework. We treat video generation assessment not merely as an engineering task, but as a core scientific problem: the systematic digitization of subjective cinematic expertise. First, we organize domain knowledge into an evaluation taxonomy aligned with the professional filmmaking workflow (pre-production, production, and post-production). Second, we distill human expert judgments into a curated dataset with large-scale human annotations. Third, we inject this knowledge into Vision-Language Models (VLMs) through an expert-calibrated fine-tuning strategy, enabling the VLM to perform explicit Chain-of-Thought reasoning. Compared to previous works, EvalVerse not only retains compatibility with foundational ''rightness'' metrics, but also significantly expands the criteria to ''goodness'' and broaden the task coverage to complex multi-shot sequencing and audio-visual integration. Consequently, by providing granular diagnostic signals, EvalVerse transcends a static leaderboard and establishes a fundamental infrastructure for future work, such as reward models and evaluator agent.

cs.CV↗

arXiv2026

Self-Consistency for LLM-Based Motion Trajectory Generation and Verification

Self-consistency has proven to be an effective technique for improving LLM performance on natural language reasoning tasks in a lightweight, unsupervised manner. In this work, we study how to adapt self-consistency to visual domains. Specifically, we consider the generation and verification of LLM-produced motion graphics trajectories. Given a prompt (e.g., "Move the circle in a spiral path"), we first sample diverse motion trajectories from an LLM, and then identify groups of consistent trajectories via clustering. Our key insight is to model the family of shapes associated with a prompt as a prototype trajectory paired with a group of geometric transformations (e.g., rigid, similarity, and affine). Two trajectories can then be considered consistent if one can be transformed into the other under the warps allowable by the transformation group. We propose an algorithm that automatically recovers a shape family, using hierarchical relationships between a set of candidate transformation groups. Our approach improves the accuracy of LLM-based trajectory generation by 4-6%. We further extend our method to support verification, observing 11% precision gains over VLM baselines. Our code and dataset are available at https://majiaju.io/trajectory-self-consistency .

cs.CV↗

arXiv2026

View-oriented Conversation Compiler for Agent Trace Analysis

We observe that an agent trace is a structured document. A coding agent session contains user turns, assistant text, chain of thought blocks, tool calls, tool results, subagent invocations, compaction boundaries, and harness injected directives, and may exceed ten thousand JSONL lines. Effective trace analysis requires a lossless record, a session level overview, and content retrieval with conversational roles. To this end, we propose VCC, namely View oriented Conversation Compiler, which lexes, parses, and lowers a raw JSONL log into three views based on one intermediate representation. The full view provides the lossless transcript and defines the line number coordinate system. The UI view reconstructs the interaction perceived by the user. The adaptive view selects relevant trace content and preserves role annotations and line range pointers. Line numbers are assigned before view lowering, ensuring that all pointers can be resolved to the full view. We evaluate VCC in a context engineering experiment on AppWorld by varying the input format of the reflector. Experiments with three model configurations and two test splits show that VCC improves aggregate task goal pass rates by 1.1 to 4.2 points, reduces reflector token consumption by one half to two thirds, and generates smaller memory files. These results demonstrate the effectiveness of trace format as an important component of context engineering infrastructure.

cs.AI↗

arXiv2026

Mode Seeking meets Mean Seeking for Fast Long Video Generation

Scaling video generation from seconds to minutes faces a critical bottleneck: while short-video data is abundant and high-fidelity, coherent long-form data is scarce and limited to narrow domains. To address this, we propose a training paradigm where Mode Seeking meets Mean Seeking, decoupling local fidelity from long-term coherence based on a unified representation via a Decoupled Diffusion Transformer. Our approach utilizes a global Flow Matching head trained via supervised learning on long videos to capture narrative structure, while simultaneously employing a local Distribution Matching head that aligns sliding windows to a frozen short-video teacher via a mode-seeking reverse-KL divergence. This strategy enables the synthesis of minute-scale videos that learns long-range coherence and motions from limited long videos via supervised flow matching, while inheriting local realism by aligning every sliding-window segment of the student to a frozen short-video teacher, resulting in a few-step fast long video generator. Evaluations show that our method effectively closes the fidelity-horizon gap by jointly improving local sharpness, motion and long-range consistency. Project website: https://primecai.github.io/mmm/.

cs.CV↗

arXiv2025

TinyHistory: Lightweight Video History Embeddings via Two-Stage Context Learning

History context is central to autoregressive video generation, driving consistency and storytelling for both commercial models and personal use cases. For example, personal users, offline workflows, and individual-scale finetuning need to encode longer video histories under tight compute and memory budgets. We observe that content and identity consistency is an essential requirement, and that complete, uninterrupted history coverage together with content query and interpretation capabilities is broadly desired. We present TinyHistory, a lightweight history embedding learned through two-stage context learning. In the first stage, we pretrain the encoder on large-scale video data with a randomized frame query objective; in the second stage, we repurpose the pretrained encoder within an autoregressive video diffusion model to learn content-level consistency. As a result, we show that the learned lightweight embeddings achieve consistency comparable (by VLM, VBench, ELO, etc) to heavier alternatives, while reducing training overhead and extending the encodable history length within a given memory budget. We conduct ablation studies to analyze the influence and trade-offs of each component.

cs.CV↗

arXiv2025

LouvreSAE: Sparse Autoencoders for Interpretable and Controllable Style Transfer

Artistic style transfer in generative models remains a significant challenge, as existing methods often introduce style only via model fine-tuning, additional adapters, or prompt engineering, all of which can be computationally expensive and may still entangle style with subject matter. In this paper, we introduce a training- and inference-light, interpretable method for representing and transferring artistic style. Our approach leverages an art-specific Sparse Autoencoder (SAE) on top of latent embeddings of generative image models. Trained on artistic data, our SAE learns an emergent, largely disentangled set of stylistic and compositional concepts, corresponding to style-related elements pertaining brushwork, texture, and color palette, as well as semantic and structural concepts. We call it LouvreSAE and use it to construct style profiles: compact, decomposable steering vectors that enable style transfer without any model updates or optimization. Unlike prior concept-based style transfer methods, our method requires no fine-tuning, no LoRA training, and no additional inference passes, enabling direct steering of artistic styles from only a few reference images. We validate our method on ArtBench10, achieving or surpassing existing methods on style evaluations (VGG Style Loss and CLIP Score Style) while being 1.7-20x faster and, critically, interpretable.

cs.CV↗

arXiv2025

Teaching Spell Checkers to Teach: Pedagogical Program Synthesis for Interactive Learning

Spelling taught through memorization often fails many learners, particularly children with language-based learning disorders who struggle with the phonological skills necessary to spell words accurately. Educators such as speech-language pathologists (SLPs) address this instructional gap by using an inquiry-based approach to teach spelling that targets the phonology, morphology, meaning, and etymology of words. Yet, these strategies rarely appear in everyday writing tools, which simply detect and autocorrect errors. We introduce SPIRE (Spelling Inquiry Engine), a spell check system that brings this inquiry-based pedagogy into the act of composition. SPIRE implements Pedagogical Program Synthesis, a novel approach for operationalizing the inherently dynamic pedagogy of spelling instruction. SPIRE represents SLP instructional moves in a domain-specific language, synthesizes tailored programs in real-time from learner errors, and renders them as interactive interfaces for inquiry-based interventions. With SPIRE, spelling errors become opportunities to explore word meanings, word structures, morphological families, word origins, and grapheme-phoneme correspondences, supporting metalinguistic reasoning alongside correction. Evaluation with SLPs and learners shows alignment with professional practice and potential for integration into writing workflows.

cs.HC↗

arXiv2025

StreamDiffusionV2: A Streaming System for Dynamic and Interactive Video Generation

Generative models are reshaping the live-streaming industry by redefining how content is created, styled, and delivered. Previous image-based streaming diffusion models have powered efficient and creative live streaming products but have hit limits on temporal consistency due to the foundation of image-based designs. Recent advances in video diffusion have markedly improved temporal consistency and sampling efficiency for offline generation. However, offline generation systems primarily optimize throughput by batching large workloads. In contrast, live online streaming operates under strict service-level objectives (SLOs): time-to-first-frame must be minimal, and every frame must meet a per-frame deadline with low jitter. Besides, scalable multi-GPU serving for real-time streams remains largely unresolved so far. To address this, we present StreamDiffusionV2, a training-free pipeline for interactive live streaming with video diffusion models. StreamDiffusionV2 integrates an SLO-aware batching scheduler and a block scheduler, together with a sink-token--guided rolling KV cache, a motion-aware noise controller, and other system-level optimizations. Moreover, we introduce a scalable pipeline orchestration that parallelizes the diffusion process across denoising steps and network layers, achieving near-linear FPS scaling without violating latency guarantees. The system scales seamlessly across heterogeneous GPU environments and supports flexible denoising steps (e.g., 1--4), enabling both ultra-low-latency and higher-quality modes. Without TensorRT or quantization, StreamDiffusionV2 renders the first frame within 0.5s and attains 58.28 FPS with a 14B-parameter model and 64.52 FPS with a 1.3B-parameter model on four H100 GPUs, making state-of-the-art generative live streaming practical and accessible--from individual creators to enterprise-scale platforms.

cs.CV↗

arXiv2025

Hollywood Town: Long-Video Generation via Cross-Modal Multi-Agent Orchestration

Recent advancements in multi-agent systems have demonstrated significant potential for enhancing creative task performance, such as long video generation. This study introduces three innovations to improve multi-agent collaboration. First, we propose OmniAgent, a hierarchical, graph-based multi-agent framework for long video generation that leverages a film-production-inspired architecture to enable modular specialization and scalable inter-agent collaboration. Second, inspired by context engineering, we propose hypergraph nodes that enable temporary group discussions among agents lacking sufficient context, reducing individual memory requirements while ensuring adequate contextual information. Third, we transition from directed acyclic graphs (DAGs) to directed cyclic graphs with limited retries, allowing agents to reflect and refine outputs iteratively, thereby improving earlier stages through feedback from subsequent nodes. These contributions lay the groundwork for developing more robust multi-agent systems in creative tasks.

cs.MA↗