Chameleon: Style-Content Disentangled Framework for Cross-Domain Object Compositing

• URL: http://arxiv.org/abs/2606.01079v1
• Date: Sun, 31 May 2026 07:54:26 GMT
• Title: Chameleon: Style-Content Disentangled Framework for Cross-Domain Object Compositing
• Abstract summary: Cross-domain compositing is relatively underexplored and remains challenging.<n>Existing approaches largely rely on training-free blending and refinement strategies.<n>We propose Chameleon, a novel two-stage training-based cross-domain compositing framework.
• Score: 14.170738532380712
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Image compositing aims to seamlessly insert a foreground object into a background image, and recent advances in diffusion models have significantly enhanced the quality, especially when the foreground and background images come from the same domain (e.g., natural images). However, cross-domain compositing, where the foreground and background come from different domains, is relatively underexplored and remains challenging because the model must preserve the foreground object's identity while stylizing it to match the background domain. Existing cross-domain compositing approaches largely rely on training-free blending and refinement strategies. This is partly due to the lack of large-scale paired datasets for cross-domain compositing, limiting the development of training-based solutions. As a result, they are limited to tone-level alignment and often produce style-inconsistent or overstylized results. To overcome such limitations, we construct ChameleonDataset, the first large-scale training dataset for cross-domain compositing, with a comprehensive evaluation benchmark, built through a scalable data construction pipeline. Building on this, we propose Chameleon, a novel two-stage training-based cross-domain compositing framework. In the first stage, we propose Joint Hard Contrastive Learning (JHCL) to train ChameleonEncoder, which effectively disentangles style and content representations. In the second stage, we introduce Spatio-Temporal Attention Gating (STAG) into a diffusion transformer for effective stylization, adaptively regulating how style tokens from the first-stage encoder are injected across spatial and temporal dimensions. Our method outperforms state-of-the-art in-domain and cross-domain compositing models, sequential pipelines and commercial models, achieving improvements in both compositional plausibility and stylistic fidelity.

Related papers

• Dual-Foundation Models for Unsupervised Domain Adaptation [2.279449016085348]
  Training segmentation models requires costly, labor-intensive annotations on real-world datasets.<n>Unsupervised Domain Adaptation (UDA) addresses this by training models on labeled synthetic data and adapting them to unlabeled real images.<n>We propose a dual-foundation UDA framework that leverages two complementary foundation models.
  arXiv  Detail & Related papers  (2026-05-05T04:52:08Z)
• CoCoDiff: Correspondence-Consistent Diffusion Model for Fine-grained Style Transfer [85.217605146499]
  CoCoDiff is a training-free and low-cost style transfer framework for computer vision.<n>It exploits pretrained latent diffusion models to achieve fine-grained, semantically consistent stylization.<n>CoCoDiff delivers state-of-the-art visual quality and strong quantitative results, outperforming methods that rely on extra training or annotations.
  arXiv  Detail & Related papers  (2026-02-16T04:52:29Z)
• Bézier Meets Diffusion: Robust Generation Across Domains for Medical Image Segmentation [18.618250617122392]
  Training robust learning algorithms across different medical imaging modalities is challenging due to the large domain gap.<n>Unsupervised domain adaptation (UDA) mitigates this problem by using annotated images from the source domain and unlabeled images from the target domain to train the deep models.<n>Existing approaches often rely on GAN-based style transfer, but these methods struggle to capture cross-domain mappings in regions with high variability.
  arXiv  Detail & Related papers  (2025-09-26T15:23:17Z)
• AIComposer: Any Style and Content Image Composition via Feature Integration [3.227277661633987]
  Cross-domain image composition remains under-explored.<n>Our method does not require text prompts, allowing natural stylization and seamless compositions.<n>Our method outperforms state-of-the-art techniques in both qualitative and quantitative evaluations.
  arXiv  Detail & Related papers  (2025-07-28T11:19:14Z)
• Cross-domain and Cross-dimension Learning for Image-to-Graph Transformers [48.74331852418905]
  Direct image-to-graph transformation is a challenging task that involves solving object detection and relationship prediction in a single model.<n>Due to this task's complexity, large training datasets are rare in many domains, making the training of deep-learning methods challenging.<n>We introduce a set of methods enabling cross-domain and cross-dimension learning for image-to-graph transformers.
  arXiv  Detail & Related papers  (2024-03-11T10:48:56Z)
• Contrast, Stylize and Adapt: Unsupervised Contrastive Learning Framework for Domain Adaptive Semantic Segmentation [18.843639142342642]
  We present CONtrastive FEaTure and pIxel alignment for bridging the domain gap at both the pixel and feature levels. Our experiments demonstrate that our method outperforms existing state-of-the-art methods using DeepLabV2.
  arXiv  Detail & Related papers  (2023-06-15T12:50:46Z)
• Generalized One-shot Domain Adaption of Generative Adversarial Networks [72.84435077616135]
  The adaption of Generative Adversarial Network (GAN) aims to transfer a pre-trained GAN to a given domain with limited training data. We consider that the adaptation from source domain to target domain can be decoupled into two parts: the transfer of global style like texture and color, and the emergence of new entities that do not belong to the source domain. Our core objective is to constrain the gap between the internal distributions of the reference and syntheses by sliced Wasserstein distance.
  arXiv  Detail & Related papers  (2022-09-08T09:24:44Z)
• Modeling Image Composition for Complex Scene Generation [77.10533862854706]
  We present a method that achieves state-of-the-art results on layout-to-image generation tasks. After compressing RGB images into patch tokens, we propose the Transformer with Focal Attention (TwFA) for exploring dependencies of object-to-object, object-to-patch and patch-to-patch.
  arXiv  Detail & Related papers  (2022-06-02T08:34:25Z)
• Few Shot Generative Model Adaption via Relaxed Spatial Structural Alignment [130.84010267004803]
  Training a generative adversarial network (GAN) with limited data has been a challenging task. A feasible solution is to start with a GAN well-trained on a large scale source domain and adapt it to the target domain with a few samples, termed as few shot generative model adaption. We propose a relaxed spatial structural alignment method to calibrate the target generative models during the adaption.
  arXiv  Detail & Related papers  (2022-03-06T14:26:25Z)
• Controllable Person Image Synthesis with Spatially-Adaptive Warped Normalization [72.65828901909708]
  Controllable person image generation aims to produce realistic human images with desirable attributes. We introduce a novel Spatially-Adaptive Warped Normalization (SAWN), which integrates a learned flow-field to warp modulation parameters. We propose a novel self-training part replacement strategy to refine the pretrained model for the texture-transfer task.
  arXiv  Detail & Related papers  (2021-05-31T07:07:44Z)
• Image-to-image Mapping with Many Domains by Sparse Attribute Transfer [71.28847881318013]
  Unsupervised image-to-image translation consists of learning a pair of mappings between two domains without known pairwise correspondences between points. Current convention is to approach this task with cycle-consistent GANs. We propose an alternate approach that directly restricts the generator to performing a simple sparse transformation in a latent layer.
  arXiv  Detail & Related papers  (2020-06-23T19:52:23Z)
• Introducing Pose Consistency and Warp-Alignment for Self-Supervised 6D Object Pose Estimation in Color Images [38.9238085806793]
  Most successful approaches to estimate the 6D pose of an object typically train a neural network by supervising the learning with annotated poses in real world images. A two-stage 6D object pose estimator framework that can be applied on top of existing neural-network-based approaches is proposed.
  arXiv  Detail & Related papers  (2020-03-27T11:53:38Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.