In 2014, if you're model has an animation which does not result in any body deformations, the animation can now be translated "all in one shot" to PhotoView.  This means that rather than the model being saved, and re-loaded to PhotoView for each rendered frame, it only has to translate one time.  This is a HUGE performance benefit.

See the attached video.  Sorry it's a bit long/wordy.  One extra point I'll mention here...  We are looking in to re-working how we detect, and choose to process the "new" way (one translation for the whole animation) vs the "old" way (translate the model for every frame).  This should hopefully remove the step I have shown in this video, where I lock the in-context references.

-Ron