作者Siyu Yan, Yizhen Gao, Yilin Wang, Dongxing Mao, Alex Jinpeng Wang
Text-rich image generation is one of the most challenging settings in image generation, since models must simultaneously produce visually realistic images and render legible, semantically aligned, and layout-consistent text. Existing data pipelines usually follow a static crawl-filter-freeze paradigm. They collect candidate samples, filter them once, and freeze the accepted data for training. However, rejected samples are usually discarded, although they often contain useful failure signals such as OCR errors and semantic mismatches. As a result, later construction rounds may repeat the same failure modes. To address these limitations, we propose DataEvolver, a self-evolving multi-agent framework for text-rich image data construction. DataEvolver treats data construction as feedback-driven construction policy evolution. A Retriever collects candidate samples, a Verifier assigns quality scores and rejection causes, a Critic summarizes round-level feedback into semantic feedback, and a Generator completes under-covered regions through targeted synthesis. The updated feedback memory then guides the next construction round. Experiments on text-rich image generation benchmarks show that DataEvolver produces more useful training data than fixed-dataset baselines under matched data budgets. At the 0.75M scale on PixArt-alpha, DataEvolver improves OCR-F1 over the strongest baseline by 85.3 percent on TextScenesHQ and 35.3 percent on LongTextBench. The improvements are consistent across both evaluated benchmarks and also transfer to Show-o2, indicating that the benefit of DataEvolver is not tied to a single downstream generator. These results suggest that rejected samples can provide actionable feedback for improving text-rich image data construction.
展开完整摘要收起摘要↓
Text-rich image generation is one of the most challenging settings in image generation, since models must simultaneously produce visually realistic images and render legible, semantically aligned, and layout-consistent text. Existing data pipelines usually follow a static crawl-filter-freeze paradigm. They collect candidate samples, filter them once, and freeze the accepted data for training. However, rejected samples are usually discarded, although they often contain useful failure signals such as OCR errors and semantic mismatches. As a result, later construction rounds may repeat the same failure modes. To address these limitations, we propose DataEvolver, a self-evolving multi-agent framework for text-rich image data construction. DataEvolver treats data construction as feedback-driven construction policy evolution. A Retriever collects candidate samples, a Verifier assigns quality scores and rejection causes, a Critic summarizes round-level feedback into semantic feedback, and a Generator completes under-covered regions through targeted synthesis. The updated feedback memory then guides the next construction round. Experiments on text-rich image generation benchmarks show that DataEvolver produces more useful training data than fixed-dataset baselines under matched data budgets. At the 0.75M scale on PixArt-alpha, DataEvolver improves OCR-F1 over the strongest baseline by 85.3 percent on TextScenesHQ and 35.3 percent on LongTextBench. The improvements are consistent across both evaluated benchmarks and also transfer to Show-o2, indicating that the benefit of DataEvolver is not tied to a single downstream generator. These results suggest that rejected samples can provide actionable feedback for improving text-rich image data construction.
VLA models have emerged as a powerful paradigm for transferring semantic knowledge from web-scale data to physical robotic control. However, current single-frame architectures suffer from intrinsic limitations: temporal myopia that discards historical dynamics, reasoning gaps between high-level instructions and low-level motor commands, and inference inefficiency due to autoregressive scalar decoding. In this work, we propose MIRTH, a unified framework designed to address these challenges. MIRTH augments a pretrained VLA backbone with three key innovations: (1) dual-scale temporal memory hubs that compress long-term scene evolution and short-term motion trends into compact embeddings; (2) latent reasoning tokens optimized via a mutual-information objective carving out a semantic plan space to align multimodal context with action trajectories; and (3) a parallel action decoding scheme that replaces autoregressive generation with vector-wise prediction to maximize control throughput. Extensive evaluations on the LIBERO simulation benchmark and a real-world LeRobot platform demonstrate that MIRTH achieves state-of-the-art performance and exhibiting emergent error recovery capabilities. The codes and collected datasets are released at http://github.com/kiva12138/mirth.
展开完整摘要收起摘要↓
VLA models have emerged as a powerful paradigm for transferring semantic knowledge from web-scale data to physical robotic control. However, current single-frame architectures suffer from intrinsic limitations: temporal myopia that discards historical dynamics, reasoning gaps between high-level instructions and low-level motor commands, and inference inefficiency due to autoregressive scalar decoding. In this work, we propose MIRTH, a unified framework designed to address these challenges. MIRTH augments a pretrained VLA backbone with three key innovations: (1) dual-scale temporal memory hubs that compress long-term scene evolution and short-term motion trends into compact embeddings; (2) latent reasoning tokens optimized via a mutual-information objective carving out a semantic plan space to align multimodal context with action trajectories; and (3) a parallel action decoding scheme that replaces autoregressive generation with vector-wise prediction to maximize control throughput. Extensive evaluations on the LIBERO simulation benchmark and a real-world LeRobot platform demonstrate that MIRTH achieves state-of-the-art performance and exhibiting emergent error recovery capabilities. The codes and collected datasets are released at http://github.com/kiva12138/mirth.
As large language model (LLM) inference evolves from text-only to multimodal paradigms, inference systems face three challenges: (1) flexible orchestration of multimodal workflows, where heterogeneous computing units exhibit complex dependencies and concurrent control; (2) efficient transmission of massive intermediate data across processes and nodes, with tensors flowing at high speed among heterogeneous roles; and (3) efficient sharing of KV caches and model weights across roles to eliminate redundant GPU memory. Existing solutions deploy LLMs and diffusion models independently, lacking a system-level abstraction for multimodal pipelines; this scatters orchestration logic, tightly couples transmission paths to specific models, and incurs high cost to integrate new models. To address these challenges, we present Omni-Flow, a distributed scheduling framework for multimodal inference through a three-layer abstraction. The Control Flow layer defines workflows via a Python DSL, orchestrating heterogeneous units into a unified dataflow graph that supports static DAGs and dynamic routing, with built-in service discovery and diverse load-balancing strategies. The Data Flow layer provides a distributed KV cache abstraction beyond prefill/decode separation, unifying allocation and enabling direct cross-role transmission across a three-tier paged storage hierarchy (GPU/CPU/SSD) over zero-copy, low-latency channels. The Compute Flow layer supports complex multimodal prefix matching for KV reuse across multi-turn dialogues, and takes over KV cache and sampling logic via a unified SGLang interface, letting diffusion models directly reuse the LLM forward path under unified parallel semantics. We demonstrate that Omni-Flow supports diverse heterogeneous scenarios with a consistent programming model, including omni-modal dialogue (LongCat-Next) and complex image generation pipelines (HunyuanImage-3).
展开完整摘要收起摘要↓
As large language model (LLM) inference evolves from text-only to multimodal paradigms, inference systems face three challenges: (1) flexible orchestration of multimodal workflows, where heterogeneous computing units exhibit complex dependencies and concurrent control; (2) efficient transmission of massive intermediate data across processes and nodes, with tensors flowing at high speed among heterogeneous roles; and (3) efficient sharing of KV caches and model weights across roles to eliminate redundant GPU memory. Existing solutions deploy LLMs and diffusion models independently, lacking a system-level abstraction for multimodal pipelines; this scatters orchestration logic, tightly couples transmission paths to specific models, and incurs high cost to integrate new models. To address these challenges, we present Omni-Flow, a distributed scheduling framework for multimodal inference through a three-layer abstraction. The Control Flow layer defines workflows via a Python DSL, orchestrating heterogeneous units into a unified dataflow graph that supports static DAGs and dynamic routing, with built-in service discovery and diverse load-balancing strategies. The Data Flow layer provides a distributed KV cache abstraction beyond prefill/decode separation, unifying allocation and enabling direct cross-role transmission across a three-tier paged storage hierarchy (GPU/CPU/SSD) over zero-copy, low-latency channels. The Compute Flow layer supports complex multimodal prefix matching for KV reuse across multi-turn dialogues, and takes over KV cache and sampling logic via a unified SGLang interface, letting diffusion models directly reuse the LLM forward path under unified parallel semantics. We demonstrate that Omni-Flow supports diverse heterogeneous scenarios with a consistent programming model, including omni-modal dialogue (LongCat-Next) and complex image generation pipelines (HunyuanImage-3).
作者Zhiyuan Yao, Zheren Fu, Zhixiao Zheng, Jiajun Li, Yi Tu, Zhendong Mao
Multimodal Large Language Models (MLLMs) are critically hampered by hallucination, generating content inconsistent with the provided image. In this paper, we identify an internal signature of hallucination: progressive degradation of text-to-image cross-attention during generation, leading to specific failure patterns like unfocused or biased attention. Existing mitigation strategies are largely outcome-driven and do not explicitly target this failure mode. To address this problem, we propose ADAPT (Attention Dynamics Alignment with Preference Tuning), an attention-based framework that intervenes directly on text-to-image cross-attention dynamics. We propose ADAPT with three key contributions: a cross-attention visual anchor refined from early decoding to provide stable spatial grounding, an attention-supervised inference mechanism that detects and corrects attention drift online, and a Visual Attention Guidance DPO that aligns preferences toward visually grounded responses. Experiments show that each component of ADAPT contributes to hallucination reduction, and the full framework achieves new best results across multiple hallucination benchmarks, reducing hallucination rates by 40%-60% across mainstream backbones while preserving general multimodal capabilities. Our work provides an attention-based perspective on mitigating hallucinations by exploring the model's internal text-to-image cross-attention behaviors. Code is available at https://github.com/yao-ustc/ADAPT
展开完整摘要收起摘要↓
Multimodal Large Language Models (MLLMs) are critically hampered by hallucination, generating content inconsistent with the provided image. In this paper, we identify an internal signature of hallucination: progressive degradation of text-to-image cross-attention during generation, leading to specific failure patterns like unfocused or biased attention. Existing mitigation strategies are largely outcome-driven and do not explicitly target this failure mode. To address this problem, we propose ADAPT (Attention Dynamics Alignment with Preference Tuning), an attention-based framework that intervenes directly on text-to-image cross-attention dynamics. We propose ADAPT with three key contributions: a cross-attention visual anchor refined from early decoding to provide stable spatial grounding, an attention-supervised inference mechanism that detects and corrects attention drift online, and a Visual Attention Guidance DPO that aligns preferences toward visually grounded responses. Experiments show that each component of ADAPT contributes to hallucination reduction, and the full framework achieves new best results across multiple hallucination benchmarks, reducing hallucination rates by 40%-60% across mainstream backbones while preserving general multimodal capabilities. Our work provides an attention-based perspective on mitigating hallucinations by exploring the model's internal text-to-image cross-attention behaviors. Code is available at https://github.com/yao-ustc/ADAPT
作者Kien T. Pham, I Chieh Chen, Qifeng Chen, Long Chen
Audio-video generation has recently gained unprecedented research attention, aiming to synthesize high-quality sounding video content with fine-grained synchronization and semantic alignment between the auditory and visual components. The preceding methods predominantly adopt a dual-branch design with separate tokenization and generation modules per modality, neglecting the representation gap while necessitating intensive computational resources for proper training. Inspired by recent advancements in one-dimensional visual tokenization, we present AVTok, a novel unified tokenizer designated for holistic audio-video generation. AVTok features a dual-stream transformer-based architecture with shared encoder-decoder and modal-specific learnable queries to efficiently and effectively encode an audio-video pair into a compact one-dimensional latent representation with a unified codebook. To cope with the heterogeneous information imbalance that hinders AVTok from exploiting aligned audio-visual information, we devise a hierarchical training strategy to progressively realize reconstruction capabilities for each modality. Extensive experiments demonstrate that AVTok excels both in audio-video reconstruction and when integrated into downstream pipelines for audio-to-video, video-to-audio, and class-conditional joint audio-video generation. AVTok paves the way for the challenge of joint audio-video tokenization and provides a potential direction to build unified large multimodal models for audio-video generation.
展开完整摘要收起摘要↓
Audio-video generation has recently gained unprecedented research attention, aiming to synthesize high-quality sounding video content with fine-grained synchronization and semantic alignment between the auditory and visual components. The preceding methods predominantly adopt a dual-branch design with separate tokenization and generation modules per modality, neglecting the representation gap while necessitating intensive computational resources for proper training. Inspired by recent advancements in one-dimensional visual tokenization, we present AVTok, a novel unified tokenizer designated for holistic audio-video generation. AVTok features a dual-stream transformer-based architecture with shared encoder-decoder and modal-specific learnable queries to efficiently and effectively encode an audio-video pair into a compact one-dimensional latent representation with a unified codebook. To cope with the heterogeneous information imbalance that hinders AVTok from exploiting aligned audio-visual information, we devise a hierarchical training strategy to progressively realize reconstruction capabilities for each modality. Extensive experiments demonstrate that AVTok excels both in audio-video reconstruction and when integrated into downstream pipelines for audio-to-video, video-to-audio, and class-conditional joint audio-video generation. AVTok paves the way for the challenge of joint audio-video tokenization and provides a potential direction to build unified large multimodal models for audio-video generation.
Modern text-to-image diffusion models, such as diffusion transformers (DiT), rely on timestep or prompt embeddings to modulate the strength of the denoising process in each timestep. While this modulation communicates the current noise level, it does not provide any quality-aware information, which can lead to generated images that are unaligned, visually inconsistent, and lacking in fidelity. In this paper, we propose the Quality Representation Module (QRM), a lightweight transformer module that learns a quality-aware representation based on existing model inputs, and produces a set of vectors $M_{qrm}$. These vectors adjust the adaptive LayerNorm modulation within the DiT transformer blocks, thereby injecting a quality-sensitive signal into the denoising parameters. The QRM introduces no significant changes to the sampling schedule or diffusion backbone. Experiments include ablations on QRM training losses and architectures, as well as empirical results demonstrating consistent image quality improvements over baseline DiT-based models.
展开完整摘要收起摘要↓
Modern text-to-image diffusion models, such as diffusion transformers (DiT), rely on timestep or prompt embeddings to modulate the strength of the denoising process in each timestep. While this modulation communicates the current noise level, it does not provide any quality-aware information, which can lead to generated images that are unaligned, visually inconsistent, and lacking in fidelity. In this paper, we propose the Quality Representation Module (QRM), a lightweight transformer module that learns a quality-aware representation based on existing model inputs, and produces a set of vectors $M_{qrm}$. These vectors adjust the adaptive LayerNorm modulation within the DiT transformer blocks, thereby injecting a quality-sensitive signal into the denoising parameters. The QRM introduces no significant changes to the sampling schedule or diffusion backbone. Experiments include ablations on QRM training losses and architectures, as well as empirical results demonstrating consistent image quality improvements over baseline DiT-based models.
作者Juncheng Ma, Yuxuan Du, Yanan Sun, Zhening Xing, Changlin Li, Zhenyu Tang, Bo Li, Peng-Tao Jiang, Li Yuan, Daquan Zhou, Yonghong Tian
Diffusion Transformers (DiTs) have significantly advanced audio-driven portrait animation, but their high computational cost leads to substantial inference latency. Although training-free diffusion caching accelerates inference significant, existing methods are primarily developed for text-conditioned generation and overlook the spatial and modality imbalances inherent in audio-driven portrait animation. In this paper, we propose SyncCache, a training-free caching acceleration method tailored for DiT-based portrait animation that explicitly exploits asymmetric dynamics. Specifically, high-frequency dynamics driven by audio conditions and concentrated in human regions are more challenging and critical to cache and reuse than the low-frequency visual background in portrait animation. First, we introduce Spatially-Asymmetric Probing to prioritize error sensitivity in dynamic human region. Second, through Modality-Decoupled Caching, we bypass heavy DiT block by reusing stable inter-block residuals, while continuously recomputing lightweight audio blocks to preserve precise lip synchronization. Furthermore, we introduce a cache ratio to control cache capacity and formulate memory-adaptive cache selection as an offline dynamic programming problem without online overhead. Extensive experiments demonstrate that SyncCache achieves superior speed-quality trade-offs, delivering up to 4.12x acceleration on HunyuanVideo-Avatar and 3.75x on Wan-S2V with near-lossless visual fidelity and precise audio alignment.
展开完整摘要收起摘要↓
Diffusion Transformers (DiTs) have significantly advanced audio-driven portrait animation, but their high computational cost leads to substantial inference latency. Although training-free diffusion caching accelerates inference significant, existing methods are primarily developed for text-conditioned generation and overlook the spatial and modality imbalances inherent in audio-driven portrait animation. In this paper, we propose SyncCache, a training-free caching acceleration method tailored for DiT-based portrait animation that explicitly exploits asymmetric dynamics. Specifically, high-frequency dynamics driven by audio conditions and concentrated in human regions are more challenging and critical to cache and reuse than the low-frequency visual background in portrait animation. First, we introduce Spatially-Asymmetric Probing to prioritize error sensitivity in dynamic human region. Second, through Modality-Decoupled Caching, we bypass heavy DiT block by reusing stable inter-block residuals, while continuously recomputing lightweight audio blocks to preserve precise lip synchronization. Furthermore, we introduce a cache ratio to control cache capacity and formulate memory-adaptive cache selection as an offline dynamic programming problem without online overhead. Extensive experiments demonstrate that SyncCache achieves superior speed-quality trade-offs, delivering up to 4.12x acceleration on HunyuanVideo-Avatar and 3.75x on Wan-S2V with near-lossless visual fidelity and precise audio alignment.
We introduce Orca, an initial instantiation of a general world foundation model. Orca learns a unified world latent space from multimodal world signals and exposes it through multimodal readout interfaces. Rather than optimizing isolated next-token, next-frame, or next-action prediction, we are centered on Next-State-Prediction modeling, offering a unified state-transition modeling route toward understanding, predicting, and acting upon the world. Orca learns through two complementary paradigms: unconscious learning captures dense natural state transitions from continuous videos, and conscious learning models sparse meaningful state transitions by language-described events and VQA supervision. For pre-training, we construct a large-scale world-learning inventory data, including 125K hours of video data and 160M event annotations. After pre-training, Orca learns a unified world latent space. To examine whether the learned latent supports downstream, we evaluate it by three representative downstream readouts: text generation, image prediction, and embodied action generation. Orca's backbone is frozen, and only the lightweight modality-specific decoders are trainable. Experiments show the scalability of the proposed paradigm and verify that stronger world latent enables stronger downstream readouts. Orca outperforms similar-sized specialized baselines. These results show that Orca, as a general world foundation model, presents a promising approach to understanding, predicting, and acting upon the world. Finally, we discuss the current limitations, aiming to provide useful insights and inspiration for the community.
展开完整摘要收起摘要↓
We introduce Orca, an initial instantiation of a general world foundation model. Orca learns a unified world latent space from multimodal world signals and exposes it through multimodal readout interfaces. Rather than optimizing isolated next-token, next-frame, or next-action prediction, we are centered on Next-State-Prediction modeling, offering a unified state-transition modeling route toward understanding, predicting, and acting upon the world. Orca learns through two complementary paradigms: unconscious learning captures dense natural state transitions from continuous videos, and conscious learning models sparse meaningful state transitions by language-described events and VQA supervision. For pre-training, we construct a large-scale world-learning inventory data, including 125K hours of video data and 160M event annotations. After pre-training, Orca learns a unified world latent space. To examine whether the learned latent supports downstream, we evaluate it by three representative downstream readouts: text generation, image prediction, and embodied action generation. Orca's backbone is frozen, and only the lightweight modality-specific decoders are trainable. Experiments show the scalability of the proposed paradigm and verify that stronger world latent enables stronger downstream readouts. Orca outperforms similar-sized specialized baselines. These results show that Orca, as a general world foundation model, presents a promising approach to understanding, predicting, and acting upon the world. Finally, we discuss the current limitations, aiming to provide useful insights and inspiration for the community.
作者Mohamed el Amine Boudjoghra, Ivan Laptev, Angela Dai
Articulated 3D objects are essential for interactive environments in embodied AI, robotics, and virtual reality, but reconstructing their structure and motion from sparse observations remains challenging. Existing approaches remain largely constrained by lack of supervised data or lack the priors needed to reliably recover articulation, hidden geometry, and internal object structure. We present the first debate-driven agentic approach to articulated 3D object reconstruction from text or image inputs that both grounds articulation reasoning in concrete motion and exposes the occluded geometry revealed under articulation. High-level agents reason about object semantics and motion using knowledge from vision-language and video models, while low-level agents estimate articulation parameters and interaction points; together, they engage in a two-round structured debate that first exploits global--local disagreement and then grounds the agents in freely generated video. The same video prior, conditioned on the agreed articulation, then drives each part through its motion to expose occluded interiors and geometry that cannot be inferred from a single static view. By combining agentic reasoning with a video generative prior, our approach jointly infers articulation and reconstructs complete 3D articulated objects, producing high-fidelity geometry, internal structure, and motion-consistent states beyond directly observed surfaces.
展开完整摘要收起摘要↓
Articulated 3D objects are essential for interactive environments in embodied AI, robotics, and virtual reality, but reconstructing their structure and motion from sparse observations remains challenging. Existing approaches remain largely constrained by lack of supervised data or lack the priors needed to reliably recover articulation, hidden geometry, and internal object structure. We present the first debate-driven agentic approach to articulated 3D object reconstruction from text or image inputs that both grounds articulation reasoning in concrete motion and exposes the occluded geometry revealed under articulation. High-level agents reason about object semantics and motion using knowledge from vision-language and video models, while low-level agents estimate articulation parameters and interaction points; together, they engage in a two-round structured debate that first exploits global--local disagreement and then grounds the agents in freely generated video. The same video prior, conditioned on the agreed articulation, then drives each part through its motion to expose occluded interiors and geometry that cannot be inferred from a single static view. By combining agentic reasoning with a video generative prior, our approach jointly infers articulation and reconstructs complete 3D articulated objects, producing high-fidelity geometry, internal structure, and motion-consistent states beyond directly observed surfaces.
作者Jianjiang Yao, Ke Xian, Renxiang Dai, Robert Caiming Qiu
We present FFAvatar, a Transformer-based 3D Gaussian framework for fast construction of high-quality and animatable 4D head avatars from one or more reference portrait images. Unlike existing feed-forward approaches that require a fixed number of input views, FFAvatar supports incremental reconstruction, progressively refining the avatar representation as additional reference images become available. At the core of our method is an alternating attention mechanism that disentangles identity appearance from expression and viewpoint variations, enabling the reconstruction of a canonical 3D appearance that remains consistent across poses and facial expressions. To balance visual fidelity and computational efficiency, we introduce a sparse-to-dense learning paradigm. Coarse appearance features are first learned using sparse primitives anchored to the FLAME vertex level and are subsequently densified in the UV domain to capture fine-grained geometric and texture details. We further propose a plug-and-play motion refinement module that enables subject-specific dynamic personalization by modeling residual motion beyond parametric deformation. Extensive experiments demonstrate that FFAvatar efficiently produces high-fidelity and controllable 4D head avatars, achieving superior flexibility, driving efficiency, and identity-consistent rendering across diverse expressions and viewpoints.
展开完整摘要收起摘要↓
We present FFAvatar, a Transformer-based 3D Gaussian framework for fast construction of high-quality and animatable 4D head avatars from one or more reference portrait images. Unlike existing feed-forward approaches that require a fixed number of input views, FFAvatar supports incremental reconstruction, progressively refining the avatar representation as additional reference images become available. At the core of our method is an alternating attention mechanism that disentangles identity appearance from expression and viewpoint variations, enabling the reconstruction of a canonical 3D appearance that remains consistent across poses and facial expressions. To balance visual fidelity and computational efficiency, we introduce a sparse-to-dense learning paradigm. Coarse appearance features are first learned using sparse primitives anchored to the FLAME vertex level and are subsequently densified in the UV domain to capture fine-grained geometric and texture details. We further propose a plug-and-play motion refinement module that enables subject-specific dynamic personalization by modeling residual motion beyond parametric deformation. Extensive experiments demonstrate that FFAvatar efficiently produces high-fidelity and controllable 4D head avatars, achieving superior flexibility, driving efficiency, and identity-consistent rendering across diverse expressions and viewpoints.
Existing instruction-based video editing datasets commonly focus on single-task appearance editing, failing to meet the complex creative demands of real-world scenarios. To bridge this gap, we present Goku, a large-scale dataset featuring 2 million high-quality, instruction-aligned video editing pairs, which is the first to extend task boundaries from basic appearance editing to multi-task and structural manipulations(e.g., precise control of subject movement). To tackle the data synthesis challenges inherent in these complex tasks, we design an efficient data synthesis pipeline that decomposes complex edits into controllable sub-problems and introduce a progressive filtering system for data reliability throughout the whole process. Furthermore, we explore the optimal network structures on Goku, and propose Goku-Edit. To deeply comprehend complex editing instructions, Goku-Edit leverages an MLLM as its text encoder and adopts a decoupled dual-branch design: a dedicated mask branch handles structural control, freeing the main branch for appearance rendering. A comprehensive video editing benchmark, Goku-Bench, is also proposed with 1,000 human-verified test cases and 7 novel editing-specific metrics. Evaluated on Goku-Bench, Goku-Edit obtains up to +8% improvement on other open-source models in terms of instruction following.
展开完整摘要收起摘要↓
Existing instruction-based video editing datasets commonly focus on single-task appearance editing, failing to meet the complex creative demands of real-world scenarios. To bridge this gap, we present Goku, a large-scale dataset featuring 2 million high-quality, instruction-aligned video editing pairs, which is the first to extend task boundaries from basic appearance editing to multi-task and structural manipulations(e.g., precise control of subject movement). To tackle the data synthesis challenges inherent in these complex tasks, we design an efficient data synthesis pipeline that decomposes complex edits into controllable sub-problems and introduce a progressive filtering system for data reliability throughout the whole process. Furthermore, we explore the optimal network structures on Goku, and propose Goku-Edit. To deeply comprehend complex editing instructions, Goku-Edit leverages an MLLM as its text encoder and adopts a decoupled dual-branch design: a dedicated mask branch handles structural control, freeing the main branch for appearance rendering. A comprehensive video editing benchmark, Goku-Bench, is also proposed with 1,000 human-verified test cases and 7 novel editing-specific metrics. Evaluated on Goku-Bench, Goku-Edit obtains up to +8% improvement on other open-source models in terms of instruction following.
Human image animation, which aims to generate a video of a reference subject following a provided action sequence, has received increasing research interest. With the development of diffusion-based/flow-based video foundation models, existing animation works have began to upgrade the guidance information from 2D skeleton/pose to 3D modeling conditions. Despite achieving reasonable results, these approaches face challenges in synthesizing trajectory-controllable human motion within natural scene under changed camera views. In this work, we present a scene-adaptive human image animation framework that controls both human motion and camera trajectories within a reconstructed 3D environment for video generation. To achieve this, we first develop a ground-adaptive 3D motion retargeting approach to enable user-friendly motion trajectory control adapting to the changes of elevations of ground and orientations automatically. Then we design a viewpoint-adaptive latent fusion mechanism to inject point-cloud geometric priors through scene-visibility masking into the generative process, providing precise guidance of viewpoint changes under camera control. Experiments on two standard human image animation benchmark datasets demonstrate remarkable improvements of our method over the state of the arts in related video generation metics. Project page: https://robinhood256100.github.io/web-disp
展开完整摘要收起摘要↓
Human image animation, which aims to generate a video of a reference subject following a provided action sequence, has received increasing research interest. With the development of diffusion-based/flow-based video foundation models, existing animation works have began to upgrade the guidance information from 2D skeleton/pose to 3D modeling conditions. Despite achieving reasonable results, these approaches face challenges in synthesizing trajectory-controllable human motion within natural scene under changed camera views. In this work, we present a scene-adaptive human image animation framework that controls both human motion and camera trajectories within a reconstructed 3D environment for video generation. To achieve this, we first develop a ground-adaptive 3D motion retargeting approach to enable user-friendly motion trajectory control adapting to the changes of elevations of ground and orientations automatically. Then we design a viewpoint-adaptive latent fusion mechanism to inject point-cloud geometric priors through scene-visibility masking into the generative process, providing precise guidance of viewpoint changes under camera control. Experiments on two standard human image animation benchmark datasets demonstrate remarkable improvements of our method over the state of the arts in related video generation metics. Project page: https://robinhood256100.github.io/web-disp
作者Qin Guo, Hao Luo, Dongxu Yue, Weixuan Jin, Xiao Fu, Fan Wang, Dan Xu
Recent advances in diffusion models have shown impressive performance in controllable image generation and dense prediction tasks. However, existing approaches typically treat diffusion-based controllable generation and dense prediction as separate tasks, overlooking the potential benefits of jointly modeling the heterogeneous distributions. In this work, we introduce UniGP, a framework built upon MMDiT, which unifies controllable generation and dense prediction through simple joint training, without the need for complex task-specific designs or losses, while preserving the backbone's versatile priors. By learning controllable generation and prediction under different conditions, our model effectively captures the joint distribution of image-geometry pairs. UniGP is capable of versatile controllable generation, dense prediction, and joint generation. Specifically, the proposed UniGP consists of DUGP and a unified dataset training strategy. The former, following the principle of Occam's razor, uses only a copied image branch of MMDiT to model dense distributions beyond RGB, while the latter integrates heterogeneous datasets into a unified training framework to jointly model generation and perception tasks. Extensive experiments demonstrate that our unified model surpasses prior unified approaches and performs on par with specialized methods. Furthermore, we demonstrate that multi-task joint training provides complementary benefits: generative priors enrich perceptual details, while perceptual learning improves structural alignment in generation.
展开完整摘要收起摘要↓
Recent advances in diffusion models have shown impressive performance in controllable image generation and dense prediction tasks. However, existing approaches typically treat diffusion-based controllable generation and dense prediction as separate tasks, overlooking the potential benefits of jointly modeling the heterogeneous distributions. In this work, we introduce UniGP, a framework built upon MMDiT, which unifies controllable generation and dense prediction through simple joint training, without the need for complex task-specific designs or losses, while preserving the backbone's versatile priors. By learning controllable generation and prediction under different conditions, our model effectively captures the joint distribution of image-geometry pairs. UniGP is capable of versatile controllable generation, dense prediction, and joint generation. Specifically, the proposed UniGP consists of DUGP and a unified dataset training strategy. The former, following the principle of Occam's razor, uses only a copied image branch of MMDiT to model dense distributions beyond RGB, while the latter integrates heterogeneous datasets into a unified training framework to jointly model generation and perception tasks. Extensive experiments demonstrate that our unified model surpasses prior unified approaches and performs on par with specialized methods. Furthermore, we demonstrate that multi-task joint training provides complementary benefits: generative priors enrich perceptual details, while perceptual learning improves structural alignment in generation.
We present Sequential Planning via Anchored Robotic Keypoints, SPARK, a training-free neurosymbolic manipulation system that reaches 43.7% on six LIBERO-PRO position & task cells, more than doubling CaP-Agent0 and Vision-Language-Action (VLA) baselines. CaP-Agent0, a multi-turn code-generation agent, achieves 18.2% by re-querying an LLM at every turn, but its restart-from-scratch solution proves costly against minor policy failures. Perception is the layer that fails most under position and task changes so SPARK spends its computation there. A single Gemini call composes the plan as a typed behavior tree (BT) of composable primitives, each already containing the low-level control (motion, grasping, depth geometry) a code-generation agent would otherwise regenerate on every trial. The rest of the budget goes to perception: a second Gemini call proposes three alternative text prompts per object, SAM3 evaluates each, and we keep the prompt$\to$label pair with the most confident detection and a recovery loop then retries a failed primitive against freshly detected objects, with no new LLM call. The alternative prompts add +27.7 points on the spatial suite and +10.0 on the object suite, with the recovery loop adding +5.0 overall. SPARK runs the same primitives on three robot families (UR10e, Franka FR3, bimanual Franka) across nine unique tasks at twenty trials each, averaging 68%. Since the detector, planner, and controller modules sit behind the typed plan, they swap independently without training, and each primitive's checkable post-condition traces a failure to the corresponding module or a kinematic limit. Every trial logs a verified, labeled trajectory, so a training-free planner that already beats VLAs can supply the data those policies need without teleoperation. Project page: https://cwru-aism.github.io/spark-page/
展开完整摘要收起摘要↓
We present Sequential Planning via Anchored Robotic Keypoints, SPARK, a training-free neurosymbolic manipulation system that reaches 43.7% on six LIBERO-PRO position & task cells, more than doubling CaP-Agent0 and Vision-Language-Action (VLA) baselines. CaP-Agent0, a multi-turn code-generation agent, achieves 18.2% by re-querying an LLM at every turn, but its restart-from-scratch solution proves costly against minor policy failures. Perception is the layer that fails most under position and task changes so SPARK spends its computation there. A single Gemini call composes the plan as a typed behavior tree (BT) of composable primitives, each already containing the low-level control (motion, grasping, depth geometry) a code-generation agent would otherwise regenerate on every trial. The rest of the budget goes to perception: a second Gemini call proposes three alternative text prompts per object, SAM3 evaluates each, and we keep the prompt$\to$label pair with the most confident detection and a recovery loop then retries a failed primitive against freshly detected objects, with no new LLM call. The alternative prompts add +27.7 points on the spatial suite and +10.0 on the object suite, with the recovery loop adding +5.0 overall. SPARK runs the same primitives on three robot families (UR10e, Franka FR3, bimanual Franka) across nine unique tasks at twenty trials each, averaging 68%. Since the detector, planner, and controller modules sit behind the typed plan, they swap independently without training, and each primitive's checkable post-condition traces a failure to the corresponding module or a kinematic limit. Every trial logs a verified, labeled trajectory, so a training-free planner that already beats VLAs can supply the data those policies need without teleoperation. Project page: https://cwru-aism.github.io/spark-page/
DiT video generation is latency-intensive due to iterative full-frame denoising, while prior cloud-edge methods largely rely on static inter-step decoupling and cannot leverage inter-frame similarity or adapt to system dynamics. We propose EcoVideo, an entropy-orchestrated framework for dynamic inter-frame decoupling: early-stage self-attention entropy provides a training-free estimate of frame-wise information density for frame selection; a cloud large model denoises sparse high-entropy keyframes; and an edge lightweight model reconstructs the remaining frames via motion-aware interpolation with refinement for temporal stability. EcoVideo further adapts the keyframe budget and edge refinement depth to real-time bandwidth and compute availability, optimizing end-to-end latency under constraints. Experiments on representative DiT video generators show improved quality--efficiency trade-offs and up to 2.9x end-to-end speedup in low-bandwidth, compute-limited edge settings. Code is available at https://github.com/IF-LAB-PKU/EcoVideo.
展开完整摘要收起摘要↓
DiT video generation is latency-intensive due to iterative full-frame denoising, while prior cloud-edge methods largely rely on static inter-step decoupling and cannot leverage inter-frame similarity or adapt to system dynamics. We propose EcoVideo, an entropy-orchestrated framework for dynamic inter-frame decoupling: early-stage self-attention entropy provides a training-free estimate of frame-wise information density for frame selection; a cloud large model denoises sparse high-entropy keyframes; and an edge lightweight model reconstructs the remaining frames via motion-aware interpolation with refinement for temporal stability. EcoVideo further adapts the keyframe budget and edge refinement depth to real-time bandwidth and compute availability, optimizing end-to-end latency under constraints. Experiments on representative DiT video generators show improved quality--efficiency trade-offs and up to 2.9x end-to-end speedup in low-bandwidth, compute-limited edge settings. Code is available at https://github.com/IF-LAB-PKU/EcoVideo.
Recent progress in humanoid behavior models has been driven in large part by abundant human motion data, but comparable motion data is scarce for non-humanoid legged robots such as quadrupeds, hexapods, and quadruped manipulators. A promising alternative is to repurpose human motion across embodiments; however, direct retargeting often produces motions that are visually plausible yet physically inconsistent or difficult to track under robot dynamics. We present X-Morph, a human-motion-to-robot-behavior pipeline that converts human motion into deployable locomotion and loco-manipulation policies for diverse non-humanoid legged morphologies. A cross-morphology retargeting stage converts human motions into kinematically plausible, intent-preserving robot references, which are then tracked by a privileged RL policy and distilled into a causal student policy. We evaluate X-Morph on three morphologically distinct platforms: a quadruped, a hexapod, and a quadruped equipped with a manipulator. The resulting policies track diverse retargeted motions, generalize to unseen human motions, and support downstream use cases including video-based teleoperation, behavior-prior control, and text-conditioned motion generation. These results suggest that large-scale human motion can serve as a substrate for learning broad, reusable behavior priors beyond humanoid robots. Project page: https://maker-rat.github.io/morph/
展开完整摘要收起摘要↓
Recent progress in humanoid behavior models has been driven in large part by abundant human motion data, but comparable motion data is scarce for non-humanoid legged robots such as quadrupeds, hexapods, and quadruped manipulators. A promising alternative is to repurpose human motion across embodiments; however, direct retargeting often produces motions that are visually plausible yet physically inconsistent or difficult to track under robot dynamics. We present X-Morph, a human-motion-to-robot-behavior pipeline that converts human motion into deployable locomotion and loco-manipulation policies for diverse non-humanoid legged morphologies. A cross-morphology retargeting stage converts human motions into kinematically plausible, intent-preserving robot references, which are then tracked by a privileged RL policy and distilled into a causal student policy. We evaluate X-Morph on three morphologically distinct platforms: a quadruped, a hexapod, and a quadruped equipped with a manipulator. The resulting policies track diverse retargeted motions, generalize to unseen human motions, and support downstream use cases including video-based teleoperation, behavior-prior control, and text-conditioned motion generation. These results suggest that large-scale human motion can serve as a substrate for learning broad, reusable behavior priors beyond humanoid robots. Project page: https://maker-rat.github.io/morph/
General-purpose robot policies should be modeled as dynamical systems, yet many VLA and generative imitation policies still rely on present observations or short windows. This Markovian shortcut fails in memory-dependent manipulation: identical observations can demand different actions after different histories. We present Chronos, a physics-informed full-history framework for non-Markovian long-horizon manipulation. The key idea is to elevate observation history from auxiliary context to the latent state of the policy dynamics. At each physical control step, Chronos forms one state-representative token by fusing observation and proprioception, so the token sequence is aligned one-to-one with physical time. A selective state space model propagates this causal historical state, which conditions a multimodal coarse action prior through implicit maximum likelihood estimation (IMLE). This prior is then refined by a second-order Schrodinger-inspired bridge that predicts acceleration fields, yielding smoother and more physically grounded robot motion. Across 16 simulated tasks and 4 real-world experiments, Chronos is evaluated on precision insertion, general manipulation, and memory-dependent long-horizon control. On RMBench, where success requires remembering task phase, Chronos achieves 73.6% average success, outperforming Markovian VLA baseline pi0.5 by +62.4 percentage points, a 6.6x relative gain, while using 10x fewer parameters. It also surpasses the memory VLA Mem-0 by 22.8 points while using over 30x fewer parameters. In real-world dual-arm experiments using a single RGB camera, Chronos achieves 78% average success over four tasks, including 72% on the three memory-dependent tasks, whereas pi0.5 achieves 7% overall and 0% on the memory-dependent subset. These results suggest that history should not be treated as auxiliary context, but as the latent state of the manipulation policy.
展开完整摘要收起摘要↓
General-purpose robot policies should be modeled as dynamical systems, yet many VLA and generative imitation policies still rely on present observations or short windows. This Markovian shortcut fails in memory-dependent manipulation: identical observations can demand different actions after different histories. We present Chronos, a physics-informed full-history framework for non-Markovian long-horizon manipulation. The key idea is to elevate observation history from auxiliary context to the latent state of the policy dynamics. At each physical control step, Chronos forms one state-representative token by fusing observation and proprioception, so the token sequence is aligned one-to-one with physical time. A selective state space model propagates this causal historical state, which conditions a multimodal coarse action prior through implicit maximum likelihood estimation (IMLE). This prior is then refined by a second-order Schrodinger-inspired bridge that predicts acceleration fields, yielding smoother and more physically grounded robot motion. Across 16 simulated tasks and 4 real-world experiments, Chronos is evaluated on precision insertion, general manipulation, and memory-dependent long-horizon control. On RMBench, where success requires remembering task phase, Chronos achieves 73.6% average success, outperforming Markovian VLA baseline pi0.5 by +62.4 percentage points, a 6.6x relative gain, while using 10x fewer parameters. It also surpasses the memory VLA Mem-0 by 22.8 points while using over 30x fewer parameters. In real-world dual-arm experiments using a single RGB camera, Chronos achieves 78% average success over four tasks, including 72% on the three memory-dependent tasks, whereas pi0.5 achieves 7% overall and 0% on the memory-dependent subset. These results suggest that history should not be treated as auxiliary context, but as the latent state of the manipulation policy.
Customized Portrait Generation (CPG) technologies have been widely used to generate high-fidelity person images given an input image indicating the identity and a text prompt indicating the required edits. Yet these methods pose significant privacy risks by spreading fake visual information. Against such risks, each public generator should be able to suppress its generation ability for a particular person when requested. Therefore, in this work we investigate the identity unlearning problem for CPG. Since there are no previous methods in this field, we propose a simple baseline that updates the image encoder by minimizing identity similarity between generated and input images for target identities to be unlearned, while maximizing it for identities to be retained. However, we find such a global perturbation in the feature space harms the fidelity of generated images for other identities to be retained. To solve this problem, we propose a novel method IREU, which first locates identity-related features in an offline manner and then only performs feature perturbations on them. The experimental results show that our proposed method IREU achieves better identity unlearning performance for target identities to be unlearned, and also keeps high fidelity for other identities to be retained. In addition, our unlearned image encoder is generalizable across different generators with the same encoder without fine-tuning, which is friendly for deployment in practice.
展开完整摘要收起摘要↓
Customized Portrait Generation (CPG) technologies have been widely used to generate high-fidelity person images given an input image indicating the identity and a text prompt indicating the required edits. Yet these methods pose significant privacy risks by spreading fake visual information. Against such risks, each public generator should be able to suppress its generation ability for a particular person when requested. Therefore, in this work we investigate the identity unlearning problem for CPG. Since there are no previous methods in this field, we propose a simple baseline that updates the image encoder by minimizing identity similarity between generated and input images for target identities to be unlearned, while maximizing it for identities to be retained. However, we find such a global perturbation in the feature space harms the fidelity of generated images for other identities to be retained. To solve this problem, we propose a novel method IREU, which first locates identity-related features in an offline manner and then only performs feature perturbations on them. The experimental results show that our proposed method IREU achieves better identity unlearning performance for target identities to be unlearned, and also keeps high fidelity for other identities to be retained. In addition, our unlearned image encoder is generalizable across different generators with the same encoder without fine-tuning, which is friendly for deployment in practice.
Automatic evaluation of image and video captioning is essential for benchmarking multimodal systems, although standard evaluation metrics show limited alignment with human judgments. Recent approaches using large language models (LLMs), commonly referred to as LLM-as-a-Judge, have improved alignment with human judgments but still suffer from a mismatch between large-vocabulary language modeling and evaluation over a small label set. To address this, we propose Rigel, an automatic evaluation metric for image and video captioning, based on self-distilled score adaptation. The metric employs an evaluation-specific scoring head distilled from a frozen LLM, which captures judgment signals in a task-aligned space without relying on large-vocabulary token sets. We then refine the LLM backbone with human judgment data. To train Rigel, we constructed the Vid-Lepus dataset, which contains 3,338 video clips, 33,380 reference captions, and 5,637 candidate captions. Experiments on multiple benchmarks show that Rigel outperforms state-of-the-art metrics, achieving over 10-point improvements on ActivityNet-Fact in the reference-free setting.
展开完整摘要收起摘要↓
Automatic evaluation of image and video captioning is essential for benchmarking multimodal systems, although standard evaluation metrics show limited alignment with human judgments. Recent approaches using large language models (LLMs), commonly referred to as LLM-as-a-Judge, have improved alignment with human judgments but still suffer from a mismatch between large-vocabulary language modeling and evaluation over a small label set. To address this, we propose Rigel, an automatic evaluation metric for image and video captioning, based on self-distilled score adaptation. The metric employs an evaluation-specific scoring head distilled from a frozen LLM, which captures judgment signals in a task-aligned space without relying on large-vocabulary token sets. We then refine the LLM backbone with human judgment data. To train Rigel, we constructed the Vid-Lepus dataset, which contains 3,338 video clips, 33,380 reference captions, and 5,637 candidate captions. Experiments on multiple benchmarks show that Rigel outperforms state-of-the-art metrics, achieving over 10-point improvements on ActivityNet-Fact in the reference-free setting.
作者Habin Lim, Jae-Ho Lee, Hah Min Lew, Ji-Su Kang, Gyeong-Moon Park
Natural face-to-face conversation requires real-time speech generation together with synchronized facial motion. Existing systems only partially address this problem: speech-only full-duplex models can generate speech in real time but do not produce facial motion, while audio-driven facial motion models animate a face from already available audio rather than jointly generating speech and motion online. To bridge this gap, we first formalize full-duplex joint speech-facial motion generation, where speech tokens and facial motion tokens are produced together every step. Building on this formulation, we propose FacePlex, a unified streaming framework with two key components. First, Rolling Flow Matching adapts flow matching to online motion generation by committing new motion frames at each streaming step. Second, Rolling Cross-Attention couples the streaming audio queue with the motion queue, allowing speech and facial motion to condition each other as generation progresses. Through extensive experiments, ablation studies, and a user study, we show that FacePlex enables full-duplex joint speech-facial motion generation under online streaming constraints, while achieving stronger lip-sync quality and motion fidelity than audio-driven facial motion baselines.
展开完整摘要收起摘要↓
Natural face-to-face conversation requires real-time speech generation together with synchronized facial motion. Existing systems only partially address this problem: speech-only full-duplex models can generate speech in real time but do not produce facial motion, while audio-driven facial motion models animate a face from already available audio rather than jointly generating speech and motion online. To bridge this gap, we first formalize full-duplex joint speech-facial motion generation, where speech tokens and facial motion tokens are produced together every step. Building on this formulation, we propose FacePlex, a unified streaming framework with two key components. First, Rolling Flow Matching adapts flow matching to online motion generation by committing new motion frames at each streaming step. Second, Rolling Cross-Attention couples the streaming audio queue with the motion queue, allowing speech and facial motion to condition each other as generation progresses. Through extensive experiments, ablation studies, and a user study, we show that FacePlex enables full-duplex joint speech-facial motion generation under online streaming constraints, while achieving stronger lip-sync quality and motion fidelity than audio-driven facial motion baselines.
While Text-to-Image (T2I) models have shown remarkable success in generating photorealistic visual content, they still struggle with the rigorous semantic alignment and logical reasoning required for scientific imagery. Inspired by Peirce's Semiotic Triad, we introduce Scientific Image Reasoning (SciIR), a comprehensive resource for training and evaluation of scientific image generation. We formalize scientific reasoning into three core dimensions: Entity Structure (Icon), Scientific Process (Index), and Scientific Law (Symbol). Specifically, to overcome the scarcity of training data in scientific image generation, we elaborately create SciIR-82k, a large-scale dataset containing over 80,000 high-quality scientific image-text pairs from cutting-edge publications. The dataset is hierarchically organized according to the semiotic dimensions and incorporates a Scientific Reasoning Chain-of-Thought (Sci-RCoT) to explicitly model underlying visual logic. For evaluation, we propose SciIR-Bench, which aligns with these three semiotic levels and employs an Atomic Checklist to convert the outcome-oriented scientific accuracy into process-oriented, verifiable, fine-grained questions. Our extensive experiments reveal significant deficiencies in current models' scientific reasoning capabilities. Furthermore, by fine-tuning on the SciIR-82k dataset, we developed the Qwen-Image-SciIR model, which achieves a substantial improvement on the SciIR-Bench, increasing the final score from 35% to 43%, laying a solid foundation for future advances in scientific image generation.
展开完整摘要收起摘要↓
While Text-to-Image (T2I) models have shown remarkable success in generating photorealistic visual content, they still struggle with the rigorous semantic alignment and logical reasoning required for scientific imagery. Inspired by Peirce's Semiotic Triad, we introduce Scientific Image Reasoning (SciIR), a comprehensive resource for training and evaluation of scientific image generation. We formalize scientific reasoning into three core dimensions: Entity Structure (Icon), Scientific Process (Index), and Scientific Law (Symbol). Specifically, to overcome the scarcity of training data in scientific image generation, we elaborately create SciIR-82k, a large-scale dataset containing over 80,000 high-quality scientific image-text pairs from cutting-edge publications. The dataset is hierarchically organized according to the semiotic dimensions and incorporates a Scientific Reasoning Chain-of-Thought (Sci-RCoT) to explicitly model underlying visual logic. For evaluation, we propose SciIR-Bench, which aligns with these three semiotic levels and employs an Atomic Checklist to convert the outcome-oriented scientific accuracy into process-oriented, verifiable, fine-grained questions. Our extensive experiments reveal significant deficiencies in current models' scientific reasoning capabilities. Furthermore, by fine-tuning on the SciIR-82k dataset, we developed the Qwen-Image-SciIR model, which achieves a substantial improvement on the SciIR-Bench, increasing the final score from 35% to 43%, laying a solid foundation for future advances in scientific image generation.
作者Hiroyasu Akada, Jesús Pérez, Emre Aksan, Vasileios Choutas, Cristian Romero, Alberto Garcia-Garcia, Vladislav Golyanik, Christian Theobalt, Thabo Beeler
Generating 3D hand-object interactions is essential for applications in robotics, XR, and synthetic data generation, where flexible controllability and strong generalization to diverse object geometries are required. However, existing methods rarely satisfy these requirements, limiting their practical applicability. We present DCGrasp, a distance-aware controllable grasp generation system built on a novel grasp energy term. This term computes Distance Profile, a signed distance from each hand vertex to the nearest object point, coupled with distance-aware weighting, effectively capturing the semantically similar hand-object interaction in near-contact regions while remaining invariant to object and hand identity. Given various controllable signals, DCGrasp first generates a Distance Profile based on a Diffusion Transformer, together with a corresponding candidate hand pose. We then refine the candidate pose through optimization, enforcing consistency between the optimized hand pose and the generated Distance Profile in near-contact regions. Our experiments show that DCGrasp produces high-quality, physically plausible grasps with flexible user control, generalizing to diverse object and hand shapes and scales. Our work establishes a robust and versatile pipeline for the synthesis of controllable 3D hand-object interactions.
展开完整摘要收起摘要↓
Generating 3D hand-object interactions is essential for applications in robotics, XR, and synthetic data generation, where flexible controllability and strong generalization to diverse object geometries are required. However, existing methods rarely satisfy these requirements, limiting their practical applicability. We present DCGrasp, a distance-aware controllable grasp generation system built on a novel grasp energy term. This term computes Distance Profile, a signed distance from each hand vertex to the nearest object point, coupled with distance-aware weighting, effectively capturing the semantically similar hand-object interaction in near-contact regions while remaining invariant to object and hand identity. Given various controllable signals, DCGrasp first generates a Distance Profile based on a Diffusion Transformer, together with a corresponding candidate hand pose. We then refine the candidate pose through optimization, enforcing consistency between the optimized hand pose and the generated Distance Profile in near-contact regions. Our experiments show that DCGrasp produces high-quality, physically plausible grasps with flexible user control, generalizing to diverse object and hand shapes and scales. Our work establishes a robust and versatile pipeline for the synthesis of controllable 3D hand-object interactions.
Recent progress in Text-to-Image generation benefits from large-scale Text-Image pairs. However, the scarcity of Text-LiDAR pairs often causes over-smoothed scenes and limited controllability. In this paper, we rethink the limitations of Text-LiDAR generation task, focusing on alleviating insufficient training priors and constructing controllable Text-LiDAR data. We propose a Text-to-LiDAR Diffusion Model for LiDAR scene generation, T2LDM++, with a Self-Conditioned Representation Guidance (SCRG). Specifically, to alleviate object over-smoothing, SCRG employs a Guidance Network (GN) to provide reconstruction-based soft supervision to the Denoising Network (DN). This enables DN to learn geometry-aware representations through reconstruction guidance, leading to more accurate denoising in DDPMs. Meanwhile, through analysis and design, SCRG exhibits more effective and lightweight, while decoupled in inference, avoiding computational overhead. Furthermore, we construct two high-quality Text-LiDAR benchmarks ($>$100K samples) using a generalized strategy of geometric annotations, along with a controllability metric. Moreover, a directional position prior is designed to mitigate street distortion, further improving scene fidelity. Additionally, T2LDM++ supports multiple conditions, including (Semantic, Box, BEV, Camera)-to-LiDAR, Sparse-to-Dense, and Dense-to-Sparse generation, by learning a control encoder via frozen DN. With effective prior modeling and high-quality Text-LiDAR benchmarks, T2LDM++ can generate realistic LiDAR scenes with rich geometric details in unconditional and conditional settings.
展开完整摘要收起摘要↓
Recent progress in Text-to-Image generation benefits from large-scale Text-Image pairs. However, the scarcity of Text-LiDAR pairs often causes over-smoothed scenes and limited controllability. In this paper, we rethink the limitations of Text-LiDAR generation task, focusing on alleviating insufficient training priors and constructing controllable Text-LiDAR data. We propose a Text-to-LiDAR Diffusion Model for LiDAR scene generation, T2LDM++, with a Self-Conditioned Representation Guidance (SCRG). Specifically, to alleviate object over-smoothing, SCRG employs a Guidance Network (GN) to provide reconstruction-based soft supervision to the Denoising Network (DN). This enables DN to learn geometry-aware representations through reconstruction guidance, leading to more accurate denoising in DDPMs. Meanwhile, through analysis and design, SCRG exhibits more effective and lightweight, while decoupled in inference, avoiding computational overhead. Furthermore, we construct two high-quality Text-LiDAR benchmarks ($>$100K samples) using a generalized strategy of geometric annotations, along with a controllability metric. Moreover, a directional position prior is designed to mitigate street distortion, further improving scene fidelity. Additionally, T2LDM++ supports multiple conditions, including (Semantic, Box, BEV, Camera)-to-LiDAR, Sparse-to-Dense, and Dense-to-Sparse generation, by learning a control encoder via frozen DN. With effective prior modeling and high-quality Text-LiDAR benchmarks, T2LDM++ can generate realistic LiDAR scenes with rich geometric details in unconditional and conditional settings.
Recent co-speech gesture generation methods often overlook cultural differences, limiting their effectiveness in human-agent interaction. Moreover, culture-conditioned models are rarely evaluated under speaker-disjoint splits, so apparent "cultural" behavior may be confounded with speaker-specific gesturing style. We introduce SICAGE, a modular framework for culture-aware co-speech gesture generation that conditions motion synthesis models on speaker-independent cultural representations. SICAGE learns these representations from audio and text by treating each speaker as a separate domain while imposing invariance across speakers. This encourages representations to remain culture-discriminative while reducing dependence on speaker identity. The resulting cultural embeddings condition a multimodal generator to produce culturally appropriate gestures. We instantiate this idea with two domain generalization approaches: adversarial learning and Fishr regularization. We further introduce ALaDiT, a real-time diffusion-based gesture generator designed to efficiently incorporate the learned cultural embeddings. To validate our method, we built TED4C-L, a 106-hour multimodal dataset of 764 TED speakers from four cultural groups. Experiments show that SICAGE improves motion realism, diversity, beat synchronization, semantic relevance, and cultural consistency.
展开完整摘要收起摘要↓
Recent co-speech gesture generation methods often overlook cultural differences, limiting their effectiveness in human-agent interaction. Moreover, culture-conditioned models are rarely evaluated under speaker-disjoint splits, so apparent "cultural" behavior may be confounded with speaker-specific gesturing style. We introduce SICAGE, a modular framework for culture-aware co-speech gesture generation that conditions motion synthesis models on speaker-independent cultural representations. SICAGE learns these representations from audio and text by treating each speaker as a separate domain while imposing invariance across speakers. This encourages representations to remain culture-discriminative while reducing dependence on speaker identity. The resulting cultural embeddings condition a multimodal generator to produce culturally appropriate gestures. We instantiate this idea with two domain generalization approaches: adversarial learning and Fishr regularization. We further introduce ALaDiT, a real-time diffusion-based gesture generator designed to efficiently incorporate the learned cultural embeddings. To validate our method, we built TED4C-L, a 106-hour multimodal dataset of 764 TED speakers from four cultural groups. Experiments show that SICAGE improves motion realism, diversity, beat synchronization, semantic relevance, and cultural consistency.
Music-driven dance video generation aims to synthesize expressive human motion that is temporally aligned with music while maintaining high visual fidelity. Despite recent progress, existing methods still face two key limitations: the lack of large-scale, high-quality dance video datasets, and the absence of principled frameworks for integrating music as a complementary conditioning signal into Video Generation Foundation Models. To address these limitations, we introduce CIPE-Dance, a large-scale Internet-sourced dance video dataset with choreography-informed text annotations, constructed via a progressive expert pipeline. To the best of our knowledge, CIPE-Dance is the largest dataset for dance video generation to date, comprising 300k high-quality clips over 400 hours and covering diverse dancers, environments, and dance genres. We further propose OmniDance, a framework-level recipe for integrating music into a TI2V foundation model without sacrificing its original controllability or visual fidelity. Motivated by the complementary roles of text as low-frequency semantics and music as high-frequency temporal dynamics, OmniDance co-designs a depth-aware specialization architecture, an anchored easy-to-hard curriculum learning strategy, and a modality-specialized time-dependent CFG strategy, enabling unified TI2V, MI2V, and MTI2V generation. Extensive experiments on CIPE-Dance demonstrate that OmniDance achieves state-of-the-art performance across all three tasks and exhibits robust multimodal integration capability. Project is available at https://github.com/AMAP-ML/OmniDance.
展开完整摘要收起摘要↓
Music-driven dance video generation aims to synthesize expressive human motion that is temporally aligned with music while maintaining high visual fidelity. Despite recent progress, existing methods still face two key limitations: the lack of large-scale, high-quality dance video datasets, and the absence of principled frameworks for integrating music as a complementary conditioning signal into Video Generation Foundation Models. To address these limitations, we introduce CIPE-Dance, a large-scale Internet-sourced dance video dataset with choreography-informed text annotations, constructed via a progressive expert pipeline. To the best of our knowledge, CIPE-Dance is the largest dataset for dance video generation to date, comprising 300k high-quality clips over 400 hours and covering diverse dancers, environments, and dance genres. We further propose OmniDance, a framework-level recipe for integrating music into a TI2V foundation model without sacrificing its original controllability or visual fidelity. Motivated by the complementary roles of text as low-frequency semantics and music as high-frequency temporal dynamics, OmniDance co-designs a depth-aware specialization architecture, an anchored easy-to-hard curriculum learning strategy, and a modality-specialized time-dependent CFG strategy, enabling unified TI2V, MI2V, and MTI2V generation. Extensive experiments on CIPE-Dance demonstrate that OmniDance achieves state-of-the-art performance across all three tasks and exhibits robust multimodal integration capability. Project is available at https://github.com/AMAP-ML/OmniDance.
作者Zhiqi Li, Chengrui Dong, Zhenhua Du, Hangning Zhou, Cong Qiu, Hailong Qin, Mu Yang, Dongxu Wei, Peidong Liu
Interactive video generation systems for camera-controlled world exploration roll out growing sequences of latent video frames, entangling state transition with high-frequency observation synthesis. We propose Walking in the Implicit, a scene-centric paradigm that changes the rollout variable from frame latents to a fixed-length, renderable implicit state, termed Neural Implicit Scene (NIS). This factorizes interactive generation into stochastic transition of a compact scene state and deterministic pose-conditioned rendering given the sampled state. We instantiate this paradigm as NeuWorld: a transformer VAE learns locally anchored NIS from sparse posed frames, and a diffusion transformer evolves NIS conditioned on future camera trajectories and geometry-aware retrieved history. By reusing the VAE encoder as a unified conditioner, NeuWorld maps camera, reference-image, and history cues into the same NIS modality, avoiding external heterogeneous encoders. Trained from scratch on public posed-view data without pretrained video backbones or auxiliary 3D reconstructors, NeuWorld achieves strong long-horizon consistency with favorable inference efficiency.
展开完整摘要收起摘要↓
Interactive video generation systems for camera-controlled world exploration roll out growing sequences of latent video frames, entangling state transition with high-frequency observation synthesis. We propose Walking in the Implicit, a scene-centric paradigm that changes the rollout variable from frame latents to a fixed-length, renderable implicit state, termed Neural Implicit Scene (NIS). This factorizes interactive generation into stochastic transition of a compact scene state and deterministic pose-conditioned rendering given the sampled state. We instantiate this paradigm as NeuWorld: a transformer VAE learns locally anchored NIS from sparse posed frames, and a diffusion transformer evolves NIS conditioned on future camera trajectories and geometry-aware retrieved history. By reusing the VAE encoder as a unified conditioner, NeuWorld maps camera, reference-image, and history cues into the same NIS modality, avoiding external heterogeneous encoders. Trained from scratch on public posed-view data without pretrained video backbones or auxiliary 3D reconstructors, NeuWorld achieves strong long-horizon consistency with favorable inference efficiency.
The advancement of generative AI models capable of producing text and image marks a critical step forward in the realm of multimodal intelligence, particularly for tasks involving the interleaving of both modalities. To advance this intelligence to the next stage, it is crucial for models to autonomously generate free-form interleaved text-image sequences. In this paper, we introduce ILLUME-X, an advanced unified multimodal paradigm that enables high-quality, free-form interleaved text-image generation by improving multimodal data efficiency and stabilizing the multimodal training process. ILLUME-X comprises three key components: (i) an expanded training data pipeline optimized for interleaved text-image generation, (ii) a progressive training strategy with self-adaptive objectives for free-length multimodal token sequences, and (iii) an objective and comprehensive evaluation method ILScore for interleaved text-image sequences. Notably, our ILLUME-X outperforms previous unified models across multiple interleaved text-image generation tasks like style transfer, image decomposition and storytelling.
展开完整摘要收起摘要↓
The advancement of generative AI models capable of producing text and image marks a critical step forward in the realm of multimodal intelligence, particularly for tasks involving the interleaving of both modalities. To advance this intelligence to the next stage, it is crucial for models to autonomously generate free-form interleaved text-image sequences. In this paper, we introduce ILLUME-X, an advanced unified multimodal paradigm that enables high-quality, free-form interleaved text-image generation by improving multimodal data efficiency and stabilizing the multimodal training process. ILLUME-X comprises three key components: (i) an expanded training data pipeline optimized for interleaved text-image generation, (ii) a progressive training strategy with self-adaptive objectives for free-length multimodal token sequences, and (iii) an objective and comprehensive evaluation method ILScore for interleaved text-image sequences. Notably, our ILLUME-X outperforms previous unified models across multiple interleaved text-image generation tasks like style transfer, image decomposition and storytelling.
We present DreamForge-World 0.1 Preview, a preview foundational world model for real-time interactive world simulation. The system adapts the LongLive 1 autoregressive video stack, itself derived from Wan2.1-T2V-1.3B, with a residual action pathway inspired by the Matrix-Game family. DreamForge-World 0.1 Preview focuses on a complementary axis to frontier-scale world simulators: low-compute adaptation, consumer-GPU runtime, and broad interactive capability coverage. It supports live keyboard and mouse control, multimodal initialization, mid-stream reprompting, dual-view operation, and minute-scale interactive rollouts at native 480p resolution, reaching up to 14 to 15 FPS FPS on a single RTX 4090 with a low memory footprint. By leveraging open video backbones and applying targeted adaptation runs, we build the preview system with high cost-efficiency. DF-World 0.1 Preview is not yet a memory-complete or frontier-quality world simulator, but demonstrates a practical low-compute route toward real-time controllable world-model previews on consumer GPUs.
展开完整摘要收起摘要↓
We present DreamForge-World 0.1 Preview, a preview foundational world model for real-time interactive world simulation. The system adapts the LongLive 1 autoregressive video stack, itself derived from Wan2.1-T2V-1.3B, with a residual action pathway inspired by the Matrix-Game family. DreamForge-World 0.1 Preview focuses on a complementary axis to frontier-scale world simulators: low-compute adaptation, consumer-GPU runtime, and broad interactive capability coverage. It supports live keyboard and mouse control, multimodal initialization, mid-stream reprompting, dual-view operation, and minute-scale interactive rollouts at native 480p resolution, reaching up to 14 to 15 FPS FPS on a single RTX 4090 with a low memory footprint. By leveraging open video backbones and applying targeted adaptation runs, we build the preview system with high cost-efficiency. DF-World 0.1 Preview is not yet a memory-complete or frontier-quality world simulator, but demonstrates a practical low-compute route toward real-time controllable world-model previews on consumer GPUs.
Recent text-to-video (T2V) diffusion models rely heavily on auxiliary reward signals (e.g., via reward models or DPO) to align generated content with human aesthetics and improve realism. These signals, however, incur substantial computational overhead, require costly human annotations, and often yield limited improvement in fine-grained local details. In this paper, we argue that your data manifold is secretly a reward model. By explicitly modeling the manifold structure of high-quality Supervised Fine-Tuning (SFT) data and encouraging video latents to lie on this manifold, we derive dense, differentiable, and nearly cost-free reward signals that significantly improve video quality, particularly in mitigating low-level distortions. Our modeling builds upon Local Coordinate Coding (LCC), which captures the `skeleton' of the manifold. However, directly applying LCC suffers from mean regression, pulling latents toward the geometric mean and losing high-frequency details. We therefore extend it to Shell Local Coordinate Coding (Shell-LCC), which models the manifold `surface' as an isotropic shell to align with the true high-density region. Experiments demonstrate that our approach improves realism, enhances high-frequency details, reduces over-smoothing artifacts, and alleviates motion blur.
展开完整摘要收起摘要↓
Recent text-to-video (T2V) diffusion models rely heavily on auxiliary reward signals (e.g., via reward models or DPO) to align generated content with human aesthetics and improve realism. These signals, however, incur substantial computational overhead, require costly human annotations, and often yield limited improvement in fine-grained local details. In this paper, we argue that your data manifold is secretly a reward model. By explicitly modeling the manifold structure of high-quality Supervised Fine-Tuning (SFT) data and encouraging video latents to lie on this manifold, we derive dense, differentiable, and nearly cost-free reward signals that significantly improve video quality, particularly in mitigating low-level distortions. Our modeling builds upon Local Coordinate Coding (LCC), which captures the `skeleton' of the manifold. However, directly applying LCC suffers from mean regression, pulling latents toward the geometric mean and losing high-frequency details. We therefore extend it to Shell Local Coordinate Coding (Shell-LCC), which models the manifold `surface' as an isotropic shell to align with the true high-density region. Experiments demonstrate that our approach improves realism, enhances high-frequency details, reduces over-smoothing artifacts, and alleviates motion blur.
Text-to-image (T2I) diffusion models often fail to faithfully render explicit textual descriptions, instead defaulting to strongly learned visual priors due to a phenomenon referred to as concept association bias. We show that such bias is particularly strong for one-and-only (OAO) objects, entities that exist in a single canonical form, such as celestial bodies, landmarks, and artworks. The deeply ingrained visual identity for these concepts often resists modification through prompting alone. Addressing this challenge, we first identify through an information-theoretic analysis that the final text embedding discards concept-level information present in the intermediate-layer text representations, reducing the mutual information available to the subsequent denoising process. We then propose Intermediate Text Representation (IR)-guided diffusion, which injects intermediate hidden states of the text encoder into the conditioning signal during early denoising steps, recovering suppressed concepts without any additional training, optimization, or external models. To systematically evaluate the challenging task of aligning generative outputs with unusual prompts for OAO objects, we introduce OAO-AttackBench, a benchmark comprising counterfactual prompts that directly conflict with the core visual identity of OAO objects. Experiments on four benchmarks, including OAO-AttackBench, show that our method achieves up to a 19.1 percentage-point improvement in VQAScore while preserving generation fidelity and human preference. Project page: https://soyoun-won.github.io/one-and-only-ir-guidance/.
展开完整摘要收起摘要↓
Text-to-image (T2I) diffusion models often fail to faithfully render explicit textual descriptions, instead defaulting to strongly learned visual priors due to a phenomenon referred to as concept association bias. We show that such bias is particularly strong for one-and-only (OAO) objects, entities that exist in a single canonical form, such as celestial bodies, landmarks, and artworks. The deeply ingrained visual identity for these concepts often resists modification through prompting alone. Addressing this challenge, we first identify through an information-theoretic analysis that the final text embedding discards concept-level information present in the intermediate-layer text representations, reducing the mutual information available to the subsequent denoising process. We then propose Intermediate Text Representation (IR)-guided diffusion, which injects intermediate hidden states of the text encoder into the conditioning signal during early denoising steps, recovering suppressed concepts without any additional training, optimization, or external models. To systematically evaluate the challenging task of aligning generative outputs with unusual prompts for OAO objects, we introduce OAO-AttackBench, a benchmark comprising counterfactual prompts that directly conflict with the core visual identity of OAO objects. Experiments on four benchmarks, including OAO-AttackBench, show that our method achieves up to a 19.1 percentage-point improvement in VQAScore while preserving generation fidelity and human preference. Project page: https://soyoun-won.github.io/one-and-only-ir-guidance/.