Dengming Zhang

Wenda Shi, Chenyuan Pan, Dengming Zhang et al.

Open surface components prevail in real industrial 3D content and support rendering, physical simulation and geometric editing. Garments serve as a typical open surface type, with numerous existing generation methods leveraging sewing patterns to generate 2D panels and stitch them into 3D shapes. Such domain-specific designs lack scalability and cannot generalize to shoes and accessories. Common field-based 3D generators prioritize watertight meshes and tend to create flawed double-layer structures on open surfaces. Though Trellis2 adopts field-free representation, its open surface results still contain normal and topology errors. We present AnySurf, a unified framework generating open, closed and hybrid 3D surfaces with accurate face orientation. Built on directed-edge enhanced Flexible Dual Grid (FDG-D), our representation retains normal direction information via oriented grid edges. We also propose ROS-FT post-training and a lightweight DE-Adapter with merely 1% extra parameters, facilitating directed edge learning while preserving original generation performance. We further construct Outfit3D dataset containing industrial garments and closed accessories. Our work transforms garment modeling into a universal 3D generation task. Experimental results demonstrate superior mesh quality and better practicality for downstream applications.

Shuhong Xiao, Tingting Zhou, Yunnong Chen et al.

Graphic User Interface (GUI) is facing great demand with the popularization and prosperity of mobile apps. Automatic UI code generation from UI design draft dramatically simplifies the development process. However, the nesting layer structure in the design draft affects the quality and usability of the generated code. Few existing GUI automated techniques detect and group the nested layers to improve the accessibility of generated code. In this paper, we proposed our UI Layers Group Detector as a vision-based method that automatically detects images (i.e., basic shapes and visual elements) and text layers that present the same semantic meanings. We propose two plug-in components, text fusion and box attention, that utilize text information from design drafts as a priori information for group localization. We construct a large-scale UI dataset for training and testing, and present a data augmentation approach to boost the detection performance. The experiment shows that the proposed method achieves a decent accuracy regarding layers grouping.

Dengming Zhang, Weitao You, Jingxiong Li et al.

Emotion understanding is critical for making Large Language Models (LLMs) more general, reliable, and aligned with humans. Art conveys emotion through the joint design of visual and auditory elements, yet most prior work is human-centered or single-modality, overlooking the emotion intentionally expressed by the artwork. Meanwhile, current Audio-Visual Language Models (AVLMs) typically require large-scale audio pretraining to endow Visual Language Models (VLMs) with hearing, which limits scalability. We present Vision Anchored Audio-Visual Emotion LLM (VAEmotionLLM), a two-stage framework that teaches a VLM to hear by seeing with limited audio pretraining and to understand emotion across modalities. In Stage 1, Vision-Guided Audio Alignment (VG-Align) distills the frozen visual pathway into a new audio pathway by aligning next-token distributions of the shared LLM on synchronized audio-video clips, enabling hearing without a large audio dataset. In Stage 2, a lightweight Cross-Modal Emotion Adapter (EmoAdapter), composed of the Emotion Enhancer and the Emotion Supervisor, injects emotion-sensitive residuals and applies emotion supervision to enhance cross-modal emotion understanding. We also construct ArtEmoBenchmark, an art-centric emotion benchmark that evaluates content and emotion understanding under audio-only, visual-only, and audio-visual inputs. VAEmotionLLM achieves state-of-the-art results on ArtEmoBenchmark, outperforming audio-only, visual-only, and audio-visual baselines. Ablations show that the proposed components are complementary.

Zhenkai Wu, Xiaowen Ma, Zhenliang Ni et al.

Vision-language models (VLMs) excel at image understanding tasks, but the large number of visual tokens imposes significant computational costs, hindering deployment on mobile devices. Many pruning methods rely solely on token importance and thus overlook inter-token redundancy, retaining numerous duplicated tokens and wasting capacity. Although some redundancy-aware approaches have been proposed, they often ignore the spatial relationships among visual tokens. This can lead to overly sparse selections of retained tokens that fail to adequately cover the regions of target objects. To address these limitations, we propose VLM-Pruner, a training-free token pruning algorithm that explicitly balances redundancy and spatial sparsity. We introduce a centrifugal token pruning paradigm that enables near-to-far selection while prioritizing the preservation of fine-grained object details. Moreover, we design a Buffering for Spatial Sparsity (BSS) criterion that defers the selection of spatially distant tokens. We further adopt a parallel greedy strategy to conduct token selection efficiently. To mitigate information loss from pruning, we selectively fuse salient information from the discarded tokens into the retained ones. Comprehensive comparisons demonstrate that VLM-Pruner consistently outperforms strong baselines across five VLMs with an 88.9\% pruning rate, while delivering an end-to-end inference speedup.

Dengming Zhang, Xiaowen Ma, Zhenliang Ni et al.

Model merging, which combines multiple domain-specialized experts into a single model, offers a practical path to endow Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) with broad capabilities without the cost of joint training or serving many models. However, training-free methods rely on hand-tuned coefficients, whereas training-based methods primarily align parameters rather than downstream task behavior and typically treat all layers uniformly, ignoring inter-layer heterogeneity. We introduce Expert Merging, a training-light method that learns a small set of layer-wise coefficients using only unlabeled calibration data. The coefficients are optimized to explicitly align the merged model's hidden states and logits with those of the corresponding experts, with a coefficient regularizer for stability and task-weighted losses for controllable trade-offs. To capture inter-layer variation, Expert Merging++ augments this design with importance-guided chunking: a normalized layer-importance metric, derived from learned coefficients, task-vector magnitudes, and parameter counts, allocates more chunk-wise coefficients to high-importance layers while keeping low-importance layers lightweight. The result is a label-free, parameter-efficient, and scalable approach to multi-expert model merging across LLMs and MLLMs. Across MLLM backbones (InternVL and Qwen2-VL) and the LLM backbone (Mistral), our method surpasses strong training-free and training-based merging baselines, with Expert Merging++ delivering further gains and, in some cases, even exceeding supervised Mixture Training. The source code is available at https://github.com/Littleor/ExpertMerging.

Wenda Shi, Yiren Song, Dengming Zhang et al.

Visual text rendering are widespread in various real-world applications, requiring careful font selection and typographic choices. Recent progress in diffusion transformer (DiT)-based text-to-image (T2I) models show promise in automating these processes. However, these methods still encounter challenges like inconsistent fonts, style variation, and limited fine-grained control, particularly at the word-level. This paper proposes a two-stage DiT-based pipeline to address these problems by enhancing controllability over typography and style in text rendering. We introduce typography control fine-tuning (TC-FT), an parameter-efficient fine-tuning method (on $5\%$ key parameters) with enclosing typography control tokens (ETC-tokens), which enables precise word-level application of typographic features. To further address style inconsistency in text rendering, we propose a text-agnostic style control adapter (SCA) that prevents content leakage while enhancing style consistency. To implement TC-FT and SCA effectively, we incorporated HTML-render into the data synthesis pipeline and proposed the first word-level controllable dataset. Through comprehensive experiments, we demonstrate the effectiveness of our approach in achieving superior word-level typographic control, font consistency, and style consistency in text rendering tasks. The datasets and models will be available for academic use.

Wenda Shi, Yiren Song, Zihan Rao et al.

Achieving precise word-level typography control within generated images remains a persistent challenge. To address it, we newly construct a word-level controlled scene text dataset and introduce the Text-Image Alignment (TIA) framework. This framework leverages cross-modal correspondence between text and local image regions provided by grounding models to enhance the Text-to-Image (T2I) model training. Furthermore, we propose WordCon, a hybrid parameter-efficient fine-tuning (PEFT) method. WordCon reparameterizes selective key parameters, improving both efficiency and portability. This allows seamless integration into diverse pipelines, including artistic text rendering, text editing, and image-conditioned text rendering. To further enhance controllability, the masked loss at the latent level is applied to guide the model to concentrate on learning the text region in the image, and the joint-attention loss provides feature-level supervision to promote disentanglement between different words. Both qualitative and quantitative results demonstrate the superiority of our method to the state of the art. The datasets and source code will be available for academic use.

Junxian Wu, Weitao You, Heda Zuo et al.

Music enhances video narratives and emotions, driving demand for automatic video-to-music (V2M) generation. However, existing V2M methods relying solely on visual features or supplementary textual inputs generate music in a black-box manner, often failing to meet user expectations. To address this challenge, we propose a novel multi-condition guided V2M generation framework that incorporates multiple time-varying conditions for enhanced control over music generation. Our method uses a two-stage training strategy that enables learning of V2M fundamentals and audiovisual temporal synchronization while meeting users' needs for multi-condition control. In the first stage, we introduce a fine-grained feature selection module and a progressive temporal alignment attention mechanism to ensure flexible feature alignment. For the second stage, we develop a dynamic conditional fusion module and a control-guided decoder module to integrate multiple conditions and accurately guide the music composition process. Extensive experiments demonstrate that our method outperforms existing V2M pipelines in both subjective and objective evaluations, significantly enhancing control and alignment with user expectations.