CVPR Findings2026
Blend-Aware Latent Diffusion: Mitigating Stitched Seams in Image Inpainting
Yunpeng Liu, Xingzhong Hou, Jie Wu, Boxiao Liu, Yi Zhang, Guanglu Song, Yu Liu, Changyao Tian, Gen Luo, Haihang You
Abstract
Image inpainting aims to fill missing or masked regions of an image in a manner that blends with the surrounding context. While diffusion models have significantly improved the visual fidelity of inpainting, they still suffer from noticeable stitched seams, including boundary discontinuity and content inconsistency between the preserved and generated regions. We argue that these issues originate from a fundamental limitation: the latent blending of the two regions in inference, which is unaccounted for in training, creates a piece-wise latent manifold. Firstly, the masked input encoded by the Variational AutoEncoder (VAE) does not perfectly align with the resized mask, resulting in boundary discontinuity that persists in the reconstruction and denoising processes. Second, the piece-wise latent manifold deviates from the assumption of data coherence in diffusion models since the two regions follow distinct distributions, leading to content inconsistency. In this work, we propose Blend-Aware Latent Diffusion, a unified framework that explicitly resolves these issues by aligning the model's training dynamics with the blend nature of inference. Our framework consists of two complementary components: BlendRecon, a blend-aware VAE that learns to decode blended latents continuously; and BlendGen, a novel denoising loss that explicitly regularizes the generated content to harmonize with the surrounding context. Extensive experiments on Brush-Bench and MISATO demonstrate that Blend-Aware Latent Diffusion effectively mitigates stitched seams and improves perceptual quality across various scenarios, including inpainting and outpainting.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- High-Resolution Image Synthesis with Latent Diffusion Models cites Denoising Diffusion Probabilistic Models
- High-Resolution Image Synthesis with Latent Diffusion Models cites Denoising Diffusion Implicit Models
- Blended Diffusion for Text-driven Editing of Natural Images cites Denoising Diffusion Probabilistic Models
- Blended Diffusion for Text-driven Editing of Natural Images cites Denoising Diffusion Implicit Models
- Denoising Diffusion Implicit Models cites Denoising Diffusion Probabilistic Models
- Blended Latent Diffusion cites High-Resolution Image Synthesis with Latent Diffusion Models
- Blended Latent Diffusion cites Denoising Diffusion Probabilistic Models
- Blended Latent Diffusion cites Denoising Diffusion Implicit Models
Builds on15
- Denoising Diffusion Probabilistic ModelsJonathan Ho, Ajay Jain, Pieter AbbeelNeurIPS 2020 · 35,902 citations
- High-Resolution Image Synthesis with Latent Diffusion ModelsRobin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser et al.CVPR 2022 · 13,123 citations
- Denoising Diffusion Implicit ModelsJiaming Song, Chenlin Meng, Stefano ErmonICLR 2021 · 11,743 citations
- Blended Diffusion for Text-driven Editing of Natural ImagesOmri Avrahami, Dani Lischinski, Ohad FriedCVPR 2022 · 670 citations
- Blended Latent DiffusionOmri Avrahami, Ohad Fried, Dani LischinskiSIGGRAPH 2023 · 339 citations
Related papers
- Hierarchical Masked 3D Diffusion Model for Video OutpaintingFanda Fan, Chaoxu Guo, Litong Gong, Biao Wang et al.ACM MM 2023 · 12 citations
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color ConsistencyYikai Wang, Chenjie Cao, Junqiu Yu, Ke Fan et al.CVPR 2025
- One Stone with Two Birds: A Null-Text-Null Frequency-Aware Diffusion Models for Text-Guided Image InpaintingHaipeng Liu, Yang Wang, Meng WangNeurIPS 2025 · 8 citations
- Towards Coherent Image Inpainting Using Denoising Diffusion Implicit ModelsGuanhua Zhang, Jiabao Ji, Yang Zhang, Mo Yu et al.ICML 2023 · 99 citations
- InverFill: One-Step Inversion for Enhanced Few-Step Diffusion InpaintingDuc Vu, Kien Nguyen, Trong-Tung Nguyen, Ngan Nguyen et al.CVPR 2026 · 4 citations
