For every CGI shot in a scene, let us say a typical 8 second shot with a CG character walking (like Buzz from Toy Story walking), when it comes to render time, are all the renders elements - e.g. occlusion, reflection, specular etc. selected in the render settings of the 3D software render settings, and then one single EXR file containing those elements for the 8 second shot rendered out before it is sent to the compositing software (e.g. After Effects or Nuke)?
Is the above where OpenEXR files are concerned, the industry standard workflow?
Do the top industry VFX houses use a workflow where the occlusion pass is for the 8 second shot is rendered in a single OpenEXR file, then the reflection pass for the 8 second shot is rendered in a single OpenEXR file etc. and then all those EXR files are sent to the compositing software?
With regards to AE this is easy to answer: It simply sucks with multichannel EXRs and the performance is terrible, no matter what they claim. Always use conventional EXR files. otehr programs like Nuke handle this much better and will gladly accept multichannel files and other stuff like tiled EXRs, though depending on the number of buffers/ channels the wiring can get complex, so often a mix of both variations is used.