Multiple Camera Compositing

[side-effect]

Hi, I'm working on a VR template in after effects, and need some help. The basic idea is that there is a 3D scene, and I collapse-transform it into six comps, each with a different camera rotated 30 degrees more than the last. These are then composited and stitched in a 'master comp'. The problem is that I have to copy all the lighting from the scene into the six different comps (it used to be eight in my prototype), as well as any 3D 'illusions' (e.g. camera-reactive particle effects). Is there any way to siphon off the views of six different cameras, all in the same comp, into my master comp? Sorry if the explanation came out a bit convoluted, I'm really tired.
Anywho, I'd be really glad if someone could help me out.

[Rick Gerard]

My first suggestion is to create the first comp, then add 6 cameras to the comp, then turn on one camera at a time and render a lossless digital intermediate, then drop all 6 rendered cameras in a new comp and set up your VR.

If that doesn't work, duplicate the master comp with all six cameras in it, then turn on only one camera in all six comps, then blend those comps in your master comp.

If you are using a camera in the master comp to fly around the scene then you are going to run into a serious problem if you turn on collapse transformation, even if you have cameras in your six original comps. The bad news is that when you turn on collapse transformations only the camera in the master comp is used. The good news is that all of the lights in the main comp will effect the 3D layers in the slave comps.

Without a much more detailed description of what you are trying to do it's hard to give you more than this...

If I were working on a 360º video I would start here: Mettle | 360/VR + 3D Plug-ins for Adobe Software... That 3rd party plug-in has quickly become the standard for working with VR video. Trying to cobble something up without the proper tools is going to be a mess.

[side-effect]

I've seen the buzz around Skybox, and I think it's not very warranted. It's incredibly overpriced and I'm attempting to build something with just the inbuilt tools of AE; I believe it's possible without spending a hundred dollars. Your second suggestion was exactly what I was already doing, but then I had to copy all the lighting effects and anything a collapse transform doesn't carry through all the comps. The first suggestion would be impractical, because I'm working on a very high resolution, and on a 3D project, and exporting footage like that is a hard and time-consuming process (and I also want the effect completely procedural, so any changes in the 3D scene carry over).
By the way, to avoid confusion, the master comp only houses 2D compositing. The 3D stuff all goes on in the scene comp, which is then collapse-transformed into six comps (with a width a 6th of the master comp) with 1 camera in each, all parented to a main motion control camera in the scene. These six comps are then composited and stitched in the master comp. My question is: Can you, instead of stitching six different comps, stitch the output of six cameras in the same comp? Hope I managed to put it slightly better that time, I work very intuitively and I'm not very good at explaining.

[side-effect]

Also, thanks a ton for the quick reply.

[Rick Gerard]

First point  - if your time is worth more than $20/hr then Sky Box is a great value. Take it from somebody that uses it. You can set up a project in a half hour and start a render something that would take a most of a day to set up with AE's built in tools. Two short projects and it's paid for....

Second point, if your main comp is strictly 2D, even if you are working with 4K or even higher resolution footage then everything in a main comp using six cameras and then rendering a lossless DI or an image sequence from each of the six cameras is going to take less time to render than putting six 3D comps in one main comp and trying to render everything at once, especially if you are using 3D lights. Take it from someone that routinely builds composites that are complex enough to take 5 to 7 minutes a frame to render. Render times go up logarithmically, they are not linear. If it takes 1 minute to render a 2K frame it will take 4 minutes to render an 4K comp and 16 minutes to render 8K comp. Digital intermediates are an industry standard for complex projects because they save time in the long run. In your project if it took 1 minute to render one of your camera comps, it would take 6 minutes to render all 6 of them cutting the render time at least 75%. If your just stacking up the 6 2D renders in a main comp and not using any complex effects your comp would probably render at about 3 or 4 seconds a frame. Even it it took one minute per frame you're still saving a bunch of time in the render. More steps but less time is usually the rule, especially if you are rendering to a common delivery format.

And  last point - no, you can't tell a main comp to use camera a different camera from a nested comp. That's why I suggested putting all 6 cameras in a main comp, duplicating the comps, which takes about 6 seconds, then creating your master comp, or better yet rendering all six, and sending things off for the final render. Even if you decide to put all 6 comps in a main comp you'll save time rendering a high resolution DI (digital intermediate) and then rendering your compressed deliverable file because the frame size also has the same effect when rendering compressed files.

[side-effect]

Thanks a bunch for the great advice!

[Kevin-Monahan]

Hi side-effect,

There's also this news: https://blogs.adobe.com/conversations/2017/06/adobe-acquires-mettles-skybox-plug-ins.html

Did Rick's advice help you solve your issue? Please let us know.

Thanks,
Kevin