Xiaodong Cun
Xiaodong Cun
6G的话不要从SDwebui启动，直接跑inference.py 或者 app.py. sd的checkpoint会占用3G的显存
the results before enhancer are here: /content/stable-diffusion-webui/outputs/SadTalker/fedb9ea0-629f-48d6-aaac-e64fcc654ba9/tmp64wrnel0##2-0-100_full.mp4. SIGKILL:9 seems you do not have larger RAMs, we will optimize the algorithm on the longer video. currently, you can try to reduce...
请更新代码到最新版本，你的好像是之前的。 https://github.com/Winfredy/SadTalker/blob/7a8b48dc6383de1332dd4cd94ddc773dd7918ce4/src/facerender/animate.py#L199
可能是音频的问题。确保是mp3或者wav文件
这里有一个实现：https://replicate.com/cjwbw/sadtalker
Hi, can you give me more information in command line log of sdwebui?
Cool! I will find out how to make it more stable.
抱歉，没有太理解你这边的意思
full model 中这ref_eyeblink 和 ref_pose我还得check一下代码
目前感觉ref_eyeblink应该不会影响head pose