Emoreward adapts video emotion evaluation to improve generation quality
EmoReward: Self-Evolving Reward Calibration for Emotional Video Generation RL
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
Evaluating how well a video shows emotions is tricky because it depends on feelings and doesn't have a clear right answer. The authors created EmoReward, a method that learns which parts of its feedback to trust during training, so it better understands the emotional content in videos it generates. This method adjusts how it weighs different clues like facial expressions and colors while making videos, progressively improving its sense of what emotional signals matter. Their tests show that this adaptive approach helps video generation systems create more emotionally accurate content.
What this means in practice
- •For visual effects teams: Improve automated emotional realism in generated video scenes by dynamically adjusting reward signals during training.
- •For game developers: Enhance character animation by adaptively weighing emotional cues to make generated video responses more expressive and context-appropriate.
Authors
Mingzhe Zheng, Yue Wu, Jing Tan, Chang Zou, Dengyang Jiang, Zeyue Tian, Xuanhua He, Ryan Xu, Yunyang Ge, Sihui Ji, Yunlong Lin, Kaihang Pan, Hongru Yan, Sixiang Chen, Dianyi Wang, Julian Zhang, Weijie Kong, Changlin Li, Lanqing Guo, Zhao Zhong, Qifeng Chen, Harry Yang
Abstract
Group Relative Policy Optimization (GRPO) has advanced generative model post-training in verifiable domains such as code and mathematics, but evaluating video quality, especially emotional expression, often depends on subjective aesthetics and has no uniquely correct target. Existing approaches typically rely on fixed rewards for visual quality, motion quality, and text alignment that poorly capture emotional intent and can invite reward hacking, while RLHF is difficult to scale and subject to individual aesthetic differences that complicate the calibration of facial, body motion, and chromatic cues across narrative contexts. We introduce EmoReward, a new self-evolving training mechanism that distills evidence reliability from reward traces collected during an actual training run to adaptively guide multi-reward weight allocation in subsequent training by determining which emotional evidence to trust and how much to weight it. EmoReward first plans specific grounding cues from the emotional prompt to guide evaluation, then uses heterogeneous signal primitives to score the generated video during the act stage. Reflective Reward Gating calibrates their contributions through instance-level gates for phase, face confidence, conflict, and subtlety, then renormalizes weights over the active set to produce the sample's fused reward. RewardMemory then estimates each primitive's reliability from rollout scores and outcomes within each emotional regime in the finished training run, then fixes bounded multipliers that calibrate fusion weights in the next training run without updating reward models. Experiments on HunyuanVideo 1.5 show improvements across multiple RL algorithms and compare EmoReward with other agentic reward methods under SAGE-GRPO. Ablations and transfer experiments on SD 3.5 and FLUX further support adaptive reward calibration for visual generation.