论文检索

Zun Wang, Jialu Li, Han Lin, Jaehong Yoon, Mohit Bansal

Storytelling video generation (SVG) aims to produce coherent and visually rich multi-scene videos that follow a structured narrative. Existing methods primarily employ LLM for high-level planning to decompose a story into scene-level descriptions, which are then independently generated and stitched together. However, these approaches struggle with generating high-quality videos aligned with the complex single-scene description, as visualizing such complex description involves coherent composition of multiple objects/events, complex motion synthesis and character customization with sequential motions. To address these challenges, we propose DREAMRUNNER, a novel story-to-video generation method: First, we structure the input script using a large language model (LLM) to facilitate both coarse-grained scene planning as well as fine-grained object-level layout planning. Next, DREAMRUNNER presents retrieval-augmented test-time adaptation to capture target motion priors for objects in each scene, supporting diverse motion customization based on retrieved videos, thus facilitating the generation of new videos with complex, scripted motions. Lastly, we propose a novel spatial-temporal region-based 3D attention and prior injection module SR3AI for fine-grained object-motion binding and frame-by-frame spatial-temporal semantic control. We compare DREAMRUNNER with various SVG baselines, demonstrating state-of-the-art performance in character consistency, text alignment, and smooth transitions. Additionally, DREAMRUNNER exhibits strong fine-grained condition-following ability in compositional text-to-video generation, significantly outperforming baselines on T2V-ComBench. Finally, we demonstrate DREAMRUNNER’s ability to generate multi-character interactions with qualitative examples.

Ziyi Wang, Shengcheng Ye, Faming Fang, Haichuan Song

Modern gaze estimation models can accurately predict human gaze from facial images. However, due to privacy concerns and intricate data collection procedures, gaze estimation datasets are typically smaller and less diverse compared to those for other vision tasks, which directly leads to poor generalization in gaze estimation models. Common solutions, such as domain adaptation models, require additional domain-specific data, yet such data is often difficult to obtain due to privacy restrictions. Meanwhile, domain generalization models suffer from limited performance due to insufficient training data. To address these fundamental challenges---privacy and data diversity---we explore privacy-preserving gaze data generation schemes and propose a novel data-driven generalization solution. Specifically, we develop two diffusion-based generative models, DDPM-Gaze and LDM-Gaze, for synthesizing gaze data. We demonstrate that synthetic data can significantly improve generalization performance when simply used with fine-tuning-based methods. Furthermore, we introduce the Domain Stability Adaptation (DSA) framework, a simple yet effective domain generalization approach that enhances model robustness by increasing the domain uncertainty of input samples while reducing prediction uncertainty. Extensive experiments validate the effectiveness of our synthetic data and demonstrate the superiority of our data-driven generalization solution.

Ziyang Wang, Mengwei Li, Hao Yin, Wenhao Liu, Zilei Wang

Large Vision-Language Models (LVLMs) enhance performance on vision-language tasks by integrating visual features from pre-trained vision encoders into large language models (LLMs). However, the large number of visual tokens introduces significant computational overhead. Existing token pruning methods either perform global selection via [CLS]-based attention in the vision encode or prune within LLM decoding layers. These approaches face two key challenges: (1) [CLS]-based attention primarily focuses on visually salient regions across the entire image, often overlooking semantically important tokens essential for reasoning; and (2) strong positional bias in the shallow decoder layers causes the model to favor later-positioned tokens, while neglecting earlier ones that may carry critical reasoning cues. To address these issues, we propose PosPrune, a training-free, two-stage visual token pruning framework. At the vision encoder, we introduce an Asymmetric Region-aware Pruning (ARP) strategy that retains more tokens in semantically rich regions while discarding more tokens from semantically less informative regions, thus preserving spatial diversity and task-relevant details. In the LLM decoding stage, we find that the positional bias in shallow layers is primarily driven by model architecture rather than task semantics. Based on this insight, we propose a novel Positional Bias Correction (PBC) mechanism to mitigate this bias. To further reduce redundancy, we apply Maximal Marginal Relevance (MMR) to select tokens that best balance textual relevance and diversity. Extensive experiments on various LVLMs and benchmarks demonstrate the general effectiveness of our approach. Notably, when applied to LLaVA-1.5-7B, PosPrune achieves a reduction of 85% in FLOPs while preserving 98.5% of the original performance.

Zixu Wang, Hongye Chen, Xiaochun Zou, Congxuan Zhang, Zhen Chen, Xinbo Zhao

Motion estimation in degraded scenes has long been a significant challenge, primarily attributed to substantial scene variations and insufficient training data. Existing approaches typically address this limitation by incorporating additional training strategies or modifying network architectures within conventional frameworks. However, these solutions not only require cumbersome training procedures or additional modal inputs, but also lack generalization capabilities. To address this problem, we propose a unified optical flow estimation framework specifically designed for degraded scenes. In this work, we employ large-scale pre-trained optical flow foundation models as both teacher and student networks. Our objective is to compensate for feature incompleteness during image degradation through pre-trained large models. Subsequently, we leverage supervised signals for fine-tuning and introduce an intra-inter frame distillation method to enable the student network to adapt to diverse cross-domain scenarios. Our proposed methodology provides deeper insights into learning style-invariant features from these learnable fine-tuning layers. Extensive experiments demonstrate that our approach achieves superior generalization performance and state-of-the-art results in degraded scenes (including low-light, rain, fog and other conditions) while requiring minimal training resources.

Zikang Wang, Boyu Chen, Zhengrong Yue, Yi Wang, Yu Qiao, Limin Wang, Yali Wang

Recent advances in video understanding have been driven by MLLMs. But these MLLMs are good at analyzing short videos, while suffering from difficulties in understanding videos with a longer context. To address this difficulty, several agent paradigms have recently been proposed, using MLLMs as agents for retrieving extra contextual knowledge in a long video. However, most existing agents ignore the key fact that a long video is composed with multiple shots, i.e., to answer the user question from a long video, it is critical to deeply understand its relevant shots like human. Without such insight, these agents often mistakenly find redundant even noisy temporal context, restricting their capacity for long video understanding. To fill this gap, we propose VideoChat-A1, a novel long video agent paradigm. Different from the previous works, our VideoChat-A1 can deeply think with long videos, via a distinct chain-of-shot reasoning paradigm. More specifically, it can progressively select the relevant shots of user question, and look into these shots in a coarse-to-fine partition. By multi-modal reasoning along the shot chain, VideoChat-A1 can effectively mimic step-by-step human thinking process, allowing the interactive discovery of preferable temporal context for thoughtful understanding in long videos. Extensive experiments show that, VideoChat-A1 achieves the state-of-the-art performance on the mainstream long video QA benchmarks, e.g., it achieves 77.0 on VideoMME(w/ subs) and 70.1 on EgoSchema, outperforming its strong baselines (e.g., InternVL2.5-8B and InternVideo2.5-8B), by up to 10.1% and 6.2%. Compared to leading closed-source GPT-4o and Gemini 1.5 Pro, VideoChat-A1 offers competitive accuracy, but only with 7% input frames and 12% inference time on average.

Zihua Wang, Jiarui Wang, Haiyang Xu, Ming Yan, Fei Huang, Xu Yang, Xiu-Shen Wei, Siya Mi, Yu Zhang

In-context learning (ICL) has emerged as a powerful paradigm for Large Visual Language Models (LVLMs), enabling them to leverage a few examples directly from input contexts. However, the effectiveness of this approach is heavily reliant on the selection of demonstrations, a process that is NP-hard. Traditional strategies, including random, similarity-based sampling and infoscore-based sampling, often lead to inefficiencies or suboptimal performance, struggling to balance both efficiency and effectiveness in demonstration selection. In this paper, we propose a novel demonstration selection framework named Coreset-based Dual Retrieval (CoDR). We show that samples within a diverse subset achieve a higher expected mutual information. To implement this, we introduce a cluster-pruning method to construct a diverse coreset that aligns more effectively with the query while maintaining diversity. Additionally, we develop a dual retrieval mechanism that enhances the selection process by achieving global demonstration selection while preserving efficiency. Experimental results demonstrate that our method significantly improves the ICL performance compared to the existing strategies, providing a robust solution for effective and efficient demonstration selection.

Zihao Wang, Hao Peng, Wei Dong, Yuecen Wei, Li Sun, Zhengtao Yu

Differentially private (DP) image synthesis enables the generation of realistic images while bounding privacy leakage, facilitating secure data sharing across organizations. However, the Gaussian noise injected during DP training, such as via DP-SGD, often severely degrades synthesis quality by disrupting model convergence. To address this, we introduce RPGen, a novel framework that enhances diffusion models' parameter robustness to mitigate DP noise effects without compromising privacy guarantees. At its core, RPGen employs adversarial model perturbation (AMP) during public pre-training to build resilience against perturbations, but we identify and tackle the critical issue of robustness transferability across domains. RPGen achieves this through a three-step process: (1) A pre-trained classifier infers labels for private images, aggregated into a class distribution noised with Gaussian mechanism for DP, and public samples are selected to match this privatized distribution for domain alignment; (2) The diffusion model is pre-trained on this curated subset with adversarial model perturbation to foster robustness; (3) The model undergoes fine-tuning on private data using DP-SGD. This synergy of robustness augmentation and transferability optimization yields high-fidelity synthesis. Extensive evaluations on ImageNet for pre-training, with CelebA and CIFAR-10 for synthesis, show RPGen outperforming state-of-the-art baselines across epsilon in 1, 5, 10. On average, it achieves 20.18% lower FID and 5.45% higher classification accuracy. Ablations confirm the efficacy of domain curation and modest perturbations, establishing RPGen as a new benchmark for privacy-utility trade-offs in image generation.

Zihan Wang, Guansong Pang, Wenjun Miao, Jin Zheng, Xiao Bai

Recent advances in Large Visual Language Models (LVLMs) have demonstrated impressive performance across various vision-language tasks by leveraging large-scale image-text pretraining and instruction tuning. However, the security vulnerabilities of LVLMs have become increasingly concerning, particularly their susceptibility to backdoor attacks. Existing backdoor attacks focus on single-target attacks, i.e., targeting a single malicious output associated with a specific trigger. In this work, we uncover multi-target backdoor attacks, where multiple independent triggers corresponding to different attack targets are added in a single pass of training, posing a greater threat to LVLMs in real-world applications. Executing such attacks in LVLMs is challenging since there can be many incorrect trigger-target mappings due to severe feature interference among different triggers. To address this challenge, we propose MTAttack, the first multi-target backdoor attack framework for enforcing accurate multiple trigger-target mappings in LVLMs. The core of MTAttack is a novel optimization method with two constraints, namely Proxy Space Partitioning constraint and Trigger Prototype Anchoring constraint. It jointly optimizes multiple triggers in the latent space, with each trigger independently mapping clean images to a unique proxy class while at the same time guaranteeing their separability. Experiments on popular benchmarks demonstrate a high success rate of MTAttack for multi-target attacks, substantially outperforming existing attack methods. Furthermore, our attack exhibits strong generalizability across datasets and robustness against backdoor defense strategies. These findings highlight the vulnerability of LVLMs to multi-target backdoor attacks and underscore the urgent need for mitigating such threats.

Zhitao Wang, Hengyu Man, Wenrui Li, Xingtao Wang, Xiaopeng Fan, Debin Zhao

Recent advances in video generation techniques have given rise to an emerging paradigm of generative video coding for Ultra-Low Bitrate (ULB) scenarios by leveraging powerful generative priors. However, most existing methods are limited by domain specificity (e.g., facial or human videos) or excessive dependence on high-level text guidance, which tend to inadequately capture fine-grained motion details, leading to unrealistic or incoherent reconstructions. To address these challenges, we propose Trajectory-Guided Generative Video Coding (dubbed T-GVC), a novel framework that bridges low-level motion tracking with high-level semantic understanding. T-GVC features a semantic-aware sparse motion sampling pipeline that extracts pixel-wise motion as sparse trajectory points based on their semantic importance, significantly reducing the bitrate while preserving critical temporal semantic information. In addition, by integrating trajectory-aligned loss constraints into diffusion processes, we introduce a training-free guidance mechanism in latent space to ensure physically plausible motion patterns without sacrificing the inherent capabilities of generative models. Experimental results demonstrate that T-GVC outperforms both traditional and neural video codecs under ULB conditions. Furthermore, additional experiments confirm that our framework achieves more precise motion control than existing text-guided methods, paving the way for a novel direction of generative video coding guided by geometric motion modeling.

Zhijie Wang, Lin Wang, Zhenyu Wen, Cong Wang

Latent Diffusion Models have become a powerful tool for generating high-fidelity unrestricted adversarial examples. However, the existing methods typically perturb only the initial latent or rely on prompt engineering, which is ill-suited to the iterative nature of the diffusion process, plus optimization instability due to external text prompts and cumulative drift that push the adversarial images off the data manifold. In this paper, we propose a hierarchical attack framework that operates in alignment with the model's generative manifold and leverages intermediate denoising states to maximize attack transferability and visual fidelity. Extensive experiments show that the proposed attack improves adversarial transferability by 10-20% against a diverse set of normally-trained models and achieves over 10.5% higher success rate against adversarially-defended models, while simultaneously enhancing visual quality by 1.0-1.2 FID reduction and 16.7% LPIPS improvements.

Zhigang Wang, Zhenguang Liu, Shaojing Fan, Sifan Wu, Yingying Jiao

Video-based human pose estimation has long been a nontrivial task due to its dynamic nature and challenging detection scenarios such as occlusion and defocus. Inspired by the success of diffusion models, researchers have applied them to video pose estimation, outperforming traditional joint detection methods. However, existing diffusion model-based methods still face challenges like slow convergence and unstable pose generation. To tackle these issues, we propose DiffusionPose, a novel framework for video pose estimation that integrates diffusion models with optimization strategies: (1) We combine the emerging Mamba with Transformers to balance global and local spatio-temporal modeling. (2) We integrate Markov Random Fields into the reverse diffusion process to enhance the denoising of pose heatmaps, particularly addressing the issue of confused generation of occluded joints. (3) We mathematically formulate a Markov objective to supervise the heatmap denoising process, enabling the model to generate anatomically plausible skeletons. Our method achieves state-of-the-art performance on three large-scale benchmark datasets. Interestingly, it shows surprising robustness in challenging video scenarios, improving the accuracy of the most difficult ankle joint by 16.9% compared to the previous best diffusion model-based method on the Challenging-PoseTrack dataset.

Zhengxue Wang, Yuan Wu, Xiang Li, Zhiqiang Yan, Jian Yang

Depth super-resolution has achieved impressive performance, and the incorporation of multi-frame information further enhances reconstruction quality. Nevertheless, statistical analyses reveal that video depth super-resolution remains affected by pronounced long-tailed distributions, with the long-tailed effects primarily manifesting in spatial non-smooth regions and temporal variation zones. To address these challenges, we propose a novel SpatioTemporal Difference Network (STDNet) comprising two core branches: a spatial difference branch and a temporal difference branch. In the spatial difference branch, we introduce a spatial difference mechanism to mitigate the long-tailed issues in spatial non-smooth regions. This mechanism dynamically aligns RGB features with learned spatial difference representations, enabling intra-frame RGB-D aggregation for depth calibration. In the temporal difference branch, we further design a temporal difference strategy that preferentially propagates temporal variation information from adjacent RGB and depth frames to the current depth frame, leveraging temporal difference representations to achieve precise motion compensation in temporal long-tailed areas. Extensive experimental results across multiple datasets demonstrate the effectiveness of our STDNet, outperforming existing approaches.

Zhen Wang, Jiaojiao Zhao, Qilong Wang, Yongfeng Dong, Wenlong Yu

Fine-Grained Domain Generalization (FGDG) presents greater challenges than conventional domain generalization due to the subtle inter-class differences and relatively pronounced intra-class variations inherent in fine-grained recognition tasks. Under domain shifts, the model becomes overly sensitive to fine-grained cues, leading to the suppression of critical features and a significant drop in performance. Cognitive studies suggest that humans classify objects by leveraging both common and specific attributes, enabling accurate differentiation between fine-grained categories. However, current deep learning models have yet to incorporate this mechanism effectively. Inspired by this mechanism, we propose Concept-Feature Structuralized Generalization (CFSG). This model explicitly disentangles both the concept and feature spaces into three structured components: common, specific, and confounding segments. To mitigate the adverse effects of varying degrees of distribution shift, we introduce an adaptive mechanism that dynamically adjusts the proportions of common, specific, and confounding components. In the final prediction, explicit weights are assigned to each pair of components. Extensive experiments on three single-source benchmark datasets demonstrate that CFSG achieves an average performance improvement of 9.87% over baseline models and outperforms existing state-of-the-art methods by an average of 3.08%. Additionally, explainability analysis validates that CFSG effectively integrates multi-granularity structured knowledge and confirms that feature structuralization facilitates the emergence of concept structuralization.

Zeyu Wang, Jiawei Feng, Jiayu Wang, Pengjie Wang, Haiyu Song

Image Fusion (IF) aims to integrate complementary features from multiple source images into a single image. However, a key challenge in this field is the lack of large-scale real-world training datasets. Existing models typically rely on either small datasets or synthetic, less realistic datasets. To address this, we propose SigFusion, a unified signal-level self-supervised learning paradigm for various IF tasks.The core idea is to use signal-level Pseudo-Label Generation Networks (PLGN) to automatically synthesize training sets and pseudo labels with real multi-source signal characteristics from vast unlabeled natural images.PLGN includes two critical components: learnable 1D Signal Modulators (SM) and SigFormer. SM learns implicit 1D signal patterns across various source images and embeds them into natural images, reducing the domain gap between synthetic and real datasets. SigFormer integrates Transformer with signal processing methods, establishing an appropriate signal representation space for SM. Its cascaded, multi-level design allows hierarchical feature learning from coarse to fine detail. Moreover, SigFormer can serve as a flexible backbone for IF, as its design adheres to the classic decomposition-reconstruction paradigm. Experimental results demonstrate that SigFusion achieves state-of-the-art performance across multiple IF tasks, including medical image fusion, infrared-visible image fusion, multi-focus image fusion, and multi-exposure image fusion.

Zeyu Wang, Jiayu Wang, Haiyu Song

3D medical image fusion (MIF) and segmentation (MIS) are critical and inherently synergistic tasks in medical image analysis. However, fundamentally integrating them remains highly challenging, since effective collaborative paradigms are still scarce and their optimization objectives fundamentally diverge. Moreover, existing continual learning techniques are unable to achieve truly advanced performance for both tasks using a shared weight. To address these challenges, we propose M²-CoFS, a unified model capable of jointly handling both tasks. Our core contribution is a “network-guided network learning” paradigm designed to break the task boundaries. We model the weight spaces of MIF and MIS as high-dimensional manifolds and innovatively use a lightweight neural network to implicitly construct a shared manifold. Interestingly, this network yields a unified weight for both tasks. To ensure the shared manifold retains the intrinsic geometry of both original manifolds, we embed manifold distances into the loss function of this network as a constraint. Additionally, we design a tailored three-stage training paradigm for our core contribution mentioned above. Stage I focuses on independent task optimization for high-quality weights; Stage II aims to reduce parameter-space distance between tasks via our cross-task weight adaptation strategy; Our core innovation serves as stage III. Experimental results show that M²-CoFS consistently outperforms state-of-the-art comparison models on both MlF and MIS.

Zerun Wang, Jiafeng Mao, Xueting Wang, Toshihiko Yamasaki

Generative models have become a powerful tool for synthesizing training data in computer vision tasks. Current approaches solely focus on aligning generated images with the target dataset distribution. As a result, they capture only the common features in the real dataset and mostly generate "easy samples", which are already well learned by models trained on real data. In contrast, those rare "hard samples", with atypical features but crucial for enhancing performance, cannot be effectively generated. Consequently, these approaches must synthesize large volumes of data to yield appreciable performance gains, yet the improvement remains limited. To overcome this limitation, we present a novel method that can learn to control the learning difficulty of samples during generation while also achieving domain alignment. Thus, it can efficiently generate valuable "hard samples" that yield significant performance improvements for target tasks. This is achieved by incorporating learning difficulty as an additional conditioning signal in generative models, together with a designed encoder structure and training–generation strategy. Experimental results across multiple datasets show that our method can achieve higher performance with lower generation cost. Specifically, we obtain the best performance with only 10% additional synthetic data, saving 63.4 GPU hours of generation time compared to the previous SOTA on ImageNet. Moreover, our method provides insightful visualizations of category-specific hard factors, serving as a tool for analyzing datasets.

Zeqing Wang, Bowen Zheng, Xingyi Yang, Zhenxiong Tan, Yuecong Xu, Xinchao Wang

Diffusion Transformer (DiT)-based video diffusion models generate high-quality videos at scale but incur prohibitive processing latency and memory costs for long videos. To address this, we propose a novel distributed inference strategy, termed DualParal. The core idea is that, instead of generating an entire video on a single GPU, we parallelize computation by partitioning both video frames and model layers across multiple GPUs. However, a naive parallel implementation is not feasible. Because all frames need to share the same noise level, they can't be processed independently. Instead, every step must wait for all others to finish, which cancels out the speed benefits of parallel processing. We overcome this obstacle with a block-wise denoising scheme. Namely, we segment the video into sequential blocks, each with a different noise level. As a result, we process them in a pipeline across the GPUs. Each GPU, holding a subset of the model layers, processes a specific block of frames and passes the results to the next GPU, enabling asynchronous computation and communication. To further optimize performance, we incorporate two key enhancements. Firstly, each GPU uses a feature cache technique to reduce the overhead of smooth transitions by reusing only features involved in cross-frame computation from the prior block, minimizing inter-GPU communication and redundant computation. Secondly, we employ a coordinated noise initialization strategy, ensuring globally consistent temporal dynamics by sharing initial noise patterns across GPUs. Together, these enable fast, artifact-free, and infinitely long video generation. Applied to the latest diffusion transformer video generator, our method efficiently produces 1,025-frame videos with up to 6.54x lower latency and 1.48x lower memory cost on 8xRTX 4090 GPUs.

Zehao Wang, Xinpeng Liu, Yudonglin Zhang, Xiaoqian Wu, Zhou Fang, Yifan Fang, Junfu Pu, Cewu Lu, Yong-Lu Li

Multimodal Large Language Models (MLLMs) have garnered significant attention recently and demonstrate outstanding capabilities in various tasks such as OCR, VQA, captioning, etc. However, hallucination remains a persistent issue. While numerous methods have been proposed to mitigate hallucinations, achieving notable improvements, these methods primarily focus on mitigating hallucinations related to object/noun concepts. Verb concepts, which are crucial for understanding human actions, have been largely overlooked. In this paper, to the best of our knowledge, we are the first to investigate the verb hallucination phenomenon of MLLMs from various perspectives. Our findings reveal that most state-of-the-art MLLMs suffer from severe verb hallucination. To assess the effectiveness of existing mitigation methods for object concept hallucination in relation to verb hallucination, we evaluated these methods and found that they do not effectively address verb hallucination. To address this issue, we propose a baseline method based on fine-tuning with rich verb knowledge, achieving decent superiority. The experiment results demonstrate that our method significantly reduces hallucinations related to verbs.

Yuzhong Wang, Wenmin Wang, Shixiong Zhang, Xinxing Yu, Zhongheng Chen

We present MCGS (Markov Chain Gaussian Splatting), a novel approach for high-fidelity dynamic scene reconstruction via combining Markov chain and 3D Gaussian splatting. Our method addresses the critical challenge of artifact-free temporal consistency in dynamic neural rendering. By integrating a Markov chain-based deformation network with multi-head temporal attention, MCGS effectively captures motion patterns and temporal dependencies, producing more accurate and stable 3D representations over time. The key innovations include: (1) a Markov Deform Network that models state transitions while preserving temporal coherence, (2) a temporal attention mechanism that adaptively weights historical states within a sliding window, and (3) strategic noise injection during training to enhance model robustness and generalization. Experiments on representative dynamic scene datasets demonstrate that MCGS outperforms previous methods in both visual quality and temporal coherence, while maintaining competitive rendering speed and efficiency. These results suggest the practical applicability of our approach to real-world dynamic scene understanding and synthesis.

Yuxuan Wang, Xuanyu Yi, Qingshan Xu, Yuan Zhou, Long Chen, Hanwang Zhang

Personalizing 3D scenes from a single reference image enables intuitive user-guided editing, which requires achieving both multi-view consistency across perspectives and referential consistency with the input image. However, these goals are particularly challenging due to the viewpoint bias caused by the limited perspective provided in a single image. Lacking the mechanisms to effectively expand reference information beyond the original view, existing methods of image-conditioned 3DGS personalization often suffer from this viewpoint bias and struggle to produce consistent results. Therefore, in this paper, we present Consistent Personalization for 3D Gaussian Splatting (CP-GS), a framework that progressively propagates the single-view reference appearance to novel perspectives. In particular, CP-GS integrates pre-trained image-to-3D generation and iterative LoRA fine-tuning to extract and extend the reference appearance, and finally produces faithful multi-view guidance images and the personalized 3DGS outputs through a view-consistent generation process guided by geometric cues. Extensive experiments on real-world scenes show that our CP-GS effectively mitigates the viewpoint bias, achieving high-quality image-conditioned 3DGS personalization that significantly outperforms existing methods.