Metahuman model gets distorted in lower LODs

Hi everyone,

I have been working with mocap and the metahumans models and found some issues with it when trying to visualize the recorded animations. After saving a mocap animation using the Take Recorder option, I found that the lower the LOD goes the more distorted the model gets. For example, this is the model in LOD 3:

And this is the same model in LOD 0:

The same kind of distortion is visible when exporting a video with animation a MP4.

Have any of you found a similar problem or how to fix it? For reference the Mocap system used is the Perception Neuron.

Thanks in advance!