Feature Request: Rendering the viewer camera without the gui


I want to create a video on a headless machine which displays the behavior of all the environments. While I could do this by recording individual camera sensors and stitching the images together, recording the viewer camera is a natural way to accomplish this.

However, the viewer camera seems to be rendered only when calling gym.draw_viewer(). Is it possible to separate the rendering of the viewer camera from the gui in the api?