Multi-subject personalized image generation aims to synthesize customized images containing multiple specified subjects without requiring test-time optimization. However, achieving fine-grained independent control over multiple subjects remains challenging due to difficulties in preserving subject fidelity and preventing cross-subject attribute leakage. We present FocusDPO, a framework that adaptively identifies focus regions based on dynamic semantic correspondence and supervision image complexity. During training, our method progressively adjusts these focal areas across noise timesteps, implementing a weighted strategy that rewards information-rich patches while penalizing regions with low prediction confidence. The framework dynamically adjusts focus allocation during the DPO process according to the semantic complexity of reference images and establishes robust correspondence mappings between generated and reference subjects. Extensive experiments demonstrate that our method substantially enhances the performance of existing pre-trained personalized generation models, achieving state-of-the-art results on both single-subject and multi-subject personalized image synthesis benchmarks. Our method effectively mitigates attribute leakage while preserving superior subject fidelity across diverse generation scenarios, advancing the frontier of controllable multi-subject image synthesis.

While previous multimodal slow-thinking methods have demonstrated remarkable success in single-image understanding scenarios, their effectiveness becomes fundamentally constrained when extended to more complex multi-image comprehension tasks. This limitation stems from their predominant reliance on text-based intermediate reasoning processes. While for human, when engaging in sophisticated multi-image analysis, they typically perform two complementary cognitive operations: (1) continuous cross-image visual comparison through region-of-interest matching, and (2) dynamic memorization of critical visual concepts throughout the reasoning chain. Motivated by these observations, we propose the Complex Multi-Modal Chain-of-Thought (CMMCoT) framework, a multi-step reasoning framework that mimics human-like "slow thinking" for multi-image understanding. Our approach incorporates two key innovations: (1) The construction of interleaved multimodal multi-step reasoning chains, which utilize critical visual region tokens, extracted from intermediate reasoning steps, as supervisory signals. This mechanism not only facilitates comprehensive cross-modal understanding but also enhances model interpretability. (2) The introduction of a test-time memory augmentation module that expands the model’s reasoning capacity during inference while preserving parameter efficiency. Furthermore, to facilitate research in this direction, we have curated a novel multi-image slow-thinking dataset. Extensive experiments demonstrate the effectiveness of our model.

Yuqin Yang, Haowu Zhou, Haoran Tu, Zhiwen Hui, Shiqi Yan, HaoYang Li, Dong She, Xianrong Yao, Yang Gao, Zhanpeng Jin. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

RageSense introduces a novel system for detecting and regulating player frustration during mobile gaming. Instead of relying on coarse emotion labels, RageSense estimates users’ valence and arousal levels in real time using near-ultrasonic acoustic sensing. By analyzing facial muscle movements via built-in smartphone speakers and microphones, our approach enables emotion sensing without requiring cameras or wearables, constituting a more unobtrusive, environment-resilient, and privacy-friendly approach than traditional emotion recognition. To transform detection into action, we integrate a large language model (LLM) that generates empathetic, context-aware interventions based on gameplay screenshots, behavioral signals, and emotional trajectories. These interventions are delivered in real time, tailored to the user’s emotional state, and designed to mitigate rage while enhancing player well-being. In a 53-participant field study, our system improved emotional state immediately after triggers and was preferred over random or template-based messages. To our knowledge, this is the first demonstration of near-ultrasonic, on-phone valence-arousal regression during mobile gameplay that directly drives real-time, context-aware interventions.

Subject-driven image personalization has seen notable advancements, especially with the ReferenceNet paradigm, which excels in integrating reference image features for creative and commercial applications. However, current ReferenceNet implementations mainly function as latent-level feature extractors, limiting their potential. This restricts the delivery of suitable features to the denoising backbone across timesteps, resulting in suboptimal image consistency. In this paper, we revisit reference feature extraction and propose TFCustom, a framework that focuses on reference image features at different temporal and frequency levels. We introduce synchronized ReferenceNet to extract reference features while optimizing noise injection and denoising. We also propose a time-aware frequency refinement module that uses high- and low-frequency filters with time embeddings to adaptively select reference feature injection. Additionally, we introduce a reward-based loss to improve the similarity between reference objects and generated images. Experimental results show that TFCustom outperforms existing methods in single-object and multi-object reference generation, with significant improvements in textual details.