tr3e
Inquiry About Potential Anomalies in SMPL Annotations
Hello! Thank you so much for sharing your excellent work and providing this dataset. I’m currently visualizing the data and noticed some unusual behavior in the SMPL annotations for a few sequences (shown below). Could you please let me know if this might be due to lower-quality sequences or if there may be an issue with my visualization code? Thank you in advance for your help!
Sequence ID: 10
https://github.com/user-attachments/assets/0bc67a49-1089-4342-a979-63b2e55e5fb3
Sequence ID: 1002
https://github.com/user-attachments/assets/7cf8e314-79e4-41cb-9e56-d6d33b7f8a78
I came across the same problem, the motion was weird. Have you solved it?
您好！非常感谢您分享您的出色工作并提供此数据集。我目前正在进行数据可视化，发现一些序列的 SMPL 注释出现了一些异常（如下所示）。请问这可能是由于序列质量较低造成的，还是我的可视化代码存在问题？在此先感谢您的帮助！
序列 ID：10
10.mp4 序列 ID：1002
12.mp4
你好，@jyuntins Hello, I'd like to know how to achieved the realistic human visualizations shown in the issue? I have been reproducing the work from the paper and obtained the prompt-generated stick figure motion sequences in mp4 format. I have also looked into the SMPL model. How can I combine the motion data stored in a .npy file with the SMPL model through code? Thank you for your help!
您好，想问问您在issue里面呈现这个真人形动画的可视化效果是怎么实现的？我按照论文的工作进行复现，得到了Prompt下生成的关于mp4格式的火柴人运动动作序列，我查了一些关于SMPL模型的信息，怎样可以把.npy这样的动作文件通过代码结合到模型上呢？感谢您的帮助！
