ue 微笑 audio2emotion 学习配置

目录自己制作微笑表情：直接拉表情曲线 —— 适合你现在这种 AnimBP 流程😀 标准微笑需要这几块：🎯 在 AnimBP 里做：mh_arkit_mapping_pose_a2f接的modify curvecurve_map 中添加了值：MouthSmile_L = 0.7 MouthSmile_R = 0.7 CheekRaise_L = 0.4 CheekRaise_R = 0.4 EyeSquint_L = 0.2 EyeSquint_R = 0.2自己制作微笑，测试ok：抬眉 与挑眉：方法三（高级）：用 Emotion 数据自动微笑docker启动：docker run -d --name audio2face-3d --gpus all \ -p 52000:52000 -p 50052:50052 -p 8002:8000 \ -e NGC_API_KEY="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" \ -e NIM_OFFLINE_MODE=1 \ -e NIM_LOCAL_MODEL_PATHS_JSON='{ "audio2face_Claire": "/opt/nim/.cache/ngc/hub/models--nim--nvidia--audio2face_3d_model/snapshots/claire_v2.3_gb20x_fp16_bs8_v1/claire_v2.3.trt", "audio2face_James": "/opt/nim/.cache/ngc/hub/models--nim--nvidia--audio2face_3d_model/snapshots/james_v2.3_gb20x_fp16_bs8_v1/james_v2.3.trt", "audio2face_Mark": "/opt/nim/.cache/ngc/hub/models--nim--nvidia--audio2face_3d_model/snapshots/mark_v2.3_gb20x_fp16_bs8_v1/mark_v2.3.trt", "