SeedEdit 3.0: Fast and High-Quality Generative Image Editing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Peng, Shi, Yichun, Lian, Xiaochen, Zhai, Zhonghua, Xia, Xin, Xiao, Xuefeng, Huang, Weilin, Yang, Jianchao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910991462170624
author Wang, Peng
Shi, Yichun
Lian, Xiaochen
Zhai, Zhonghua
Xia, Xin
Xiao, Xuefeng
Huang, Weilin
Yang, Jianchao
author_facet Wang, Peng
Shi, Yichun
Lian, Xiaochen
Zhai, Zhonghua
Xia, Xin
Xiao, Xuefeng
Huang, Weilin
Yang, Jianchao
contents We introduce SeedEdit 3.0, in companion with our T2I model Seedream 3.0, which significantly improves over our previous SeedEdit versions in both aspects of edit instruction following and image content (e.g., ID/IP) preservation on real image inputs. Additional to model upgrading with T2I, in this report, we present several key improvements. First, we develop an enhanced data curation pipeline with a meta-info paradigm and meta-info embedding strategy that help mix images from multiple data sources. This allows us to scale editing data effectively, and meta information is helpfult to connect VLM with diffusion model more closely. Second, we introduce a joint learning pipeline for computing a diffusion loss and reward losses. Finally, we evaluate SeedEdit 3.0 on our testing benchmarks, for real/synthetic image editing, where it achieves a best trade-off between multiple aspects, yielding a high usability rate of 56.1%, compared to SeedEdit 1.6 (38.4%), GPT4o (37.1%) and Gemini 2.0 (30.3%).
format Preprint
id arxiv_https___arxiv_org_abs_2506_05083
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SeedEdit 3.0: Fast and High-Quality Generative Image Editing
Wang, Peng
Shi, Yichun
Lian, Xiaochen
Zhai, Zhonghua
Xia, Xin
Xiao, Xuefeng
Huang, Weilin
Yang, Jianchao
Computer Vision and Pattern Recognition
We introduce SeedEdit 3.0, in companion with our T2I model Seedream 3.0, which significantly improves over our previous SeedEdit versions in both aspects of edit instruction following and image content (e.g., ID/IP) preservation on real image inputs. Additional to model upgrading with T2I, in this report, we present several key improvements. First, we develop an enhanced data curation pipeline with a meta-info paradigm and meta-info embedding strategy that help mix images from multiple data sources. This allows us to scale editing data effectively, and meta information is helpfult to connect VLM with diffusion model more closely. Second, we introduce a joint learning pipeline for computing a diffusion loss and reward losses. Finally, we evaluate SeedEdit 3.0 on our testing benchmarks, for real/synthetic image editing, where it achieves a best trade-off between multiple aspects, yielding a high usability rate of 56.1%, compared to SeedEdit 1.6 (38.4%), GPT4o (37.1%) and Gemini 2.0 (30.3%).
title SeedEdit 3.0: Fast and High-Quality Generative Image Editing
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.05083