(Re)render 3d bounding box to the original single image.

[suvigy]

Hi,

Thanks for the great work. I haven't taken deep dive into the method yet, but my goal is to be able to render a 3d bounding box back to the original (single) image (I have the cam calib params of the original image). Is the original camera position somehow encoded in the decoded output? I can create the 3d Bound box easily, as I've seen the 3d meshes are axis-aligned. But then It is not clear, if I have to do camera pose optimization for silhoutte rgb rendering on meshes/textured meshes. I tried the camera optimization, but in case of even small artifacts this optimization can be suboptimal.

Thanks