研究人员推出了一种新颖的双分支架构CustomShift，用于文本到图像生成中的主体驱动图像定制。该方法基于Stable Diffusion 3，通过将定制表述为条件注意力分布迁移来解决现有方法的局限性。CustomShift旨在通过对齐参考图像特征与潜在表示，并整合文本和参考线索，来同时提高语义保真度和主体一致性。 AI
影响 通过增强身份保持和语义保真度，改进了主体驱动的图像生成。
排序理由 该集群包含一篇详细介绍图像定制新方法和架构的研究论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2606.16866v1 Announce Type: new Abstract: Subject-driven image customization aims to generate images that not only follow textual instructions but also preserve the identity of a given reference subject. Existing approaches, including test-time fine-tuning, encoder-based me…
Subject-driven image customization aims to generate images that not only follow textual instructions but also preserve the identity of a given reference subject. Existing approaches, including test-time fine-tuning, encoder-based methods, and token competition in shared attention…