arXiv 2606.269302026-06-26cs.CV

PortraitGen: 基于示例驱动的双奖励引导GRPO的逼真肖像生成

PortraitGen: Exemplar-Driven GRPO with Dual-Reward Guidance for Photorealistic Portrait Generation

Dalian University of Technology（大连理工大学）；University of Electronic Science and Technology of China（电子科技大学）；Zhejiang University（浙江大学）；WeChat, Tencent Inc.（腾讯微信）

Xiaomin Li, Qian Liang, Yinan Li, Ying Zhang, Chen Li, Jing Lyu, Huchuan Lu, Xu Jia

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出PortraitGen框架，通过将真实图像引入GRPO采样组并设计双奖励机制（OmniReward和AI-Portrait），有效抑制AI伪影，实现逼真肖像生成。

摘要：

像组相对策略优化（GRPO）这样的强化学习显著推进了文本到图像的后训练。然而，当前方法往往偏好表面美学，例如过饱和的颜色，而忽略了关键缺陷，如AI伪影和生物学不合理性。我们将这些限制归因于两个主要因素：（1）后训练期间缺乏真实图像，使得GRPO采样局限于原始分布，无法打破固有的生成边界；（2）优化过程缺乏针对细粒度伪影（如过度油腻的皮肤和其他AI伪影）的特定奖励。为了解决这个问题，我们提出了PortraitGen，一个专为逼真肖像生成设计的新颖框架。首先，我们通过直接将真实图像引入GRPO采样组来打破固有的生成边界，其中使用图像反转来获得它们的转移概率和潜变量。其次，为了明确引导模型走向逼真度，我们引入了一个互补的双奖励机制：用于一般质量的OmniReward和用于以人为中心的保真度的AI-Portrait。此外，我们策划了PortraitBench，一个全面的以肖像为中心的基准测试。大量实验表明，PortraitGen显著优于现有基线，有效抑制了AI伪影并实现了前所未有的逼真度。

英文摘要：

Reinforcement Learning like Group Relative Policy Optimization (GRPO) has significantly advanced text-to-image post-training. However, current methods often favor superficial aesthetics, such as over-saturated colors, leaving critical flaws like AI artifacts and biological implausibilities unresolved. We attribute these limitations to two primary factors: (1) The absence of real images during post-training confines GRPO sampling to the original distribution, failing to break inherent generative boundaries; (2) the optimization process lacks specific rewards targeting fine-grained artifacts like overly oily skin and other AI artifacts. To address this, we propose PortraitGen, a novel framework tailored for photorealistic portrait generation. First, we break inherent generative boundaries by directly introducing real images into the GRPO sampling groups, where image inversion is employed to obtain their transition probabilities and latents. Second, to explicitly steer the model toward photorealism, we introduce a complementary dual-reward mechanism: OmniReward for general quality and AI-Portrait for human-centric fidelity. Furthermore, we curate PortraitBench, a comprehensive portrait-centric benchmark. Extensive experiments demonstrate that PortraitGen significantly outperforms existing baselines, effectively suppressing AI artifacts and achieving unprecedented photorealism.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。