[James Strecker] “Is there some magic way I can do 16-bit, lossless, and not gigantic file sizes?”
Probably in the same way someone can have their cake and eat it too. By definition lossless is lossless. That said people usually use these lossy but “visually lossless” intermediate codecs like Prores, DNxHD or Cineform.
[James Strecker] “what is the point of working in 16-bit depth in AE?”
The math is working differently with the different bit depths. There will be a difference (especially in gradients) if you’re comping in 8, 16 or 32 bits. Also there’s probably no service/basic media player that will accept 16 bit files anyway.
We do our CG shots in EXR half float and that’ll keep the quality high through the whole pipeline and only at the delivery stage we render it down to whatever we need to deliver.