[Matthew Belinkie]

Honestly, I can’t tell the difference between ProRes and ProRes LT, even when there’s a decent amount of motion. But even if the footage LOOKS fine, I’ve heard that ProRes LT will start to degrade quickly when you add effects to it and compress it for the web. I just wanted to get some opinions from people who have done this more.

I’ll probably use ProRes, because better safe than sorry.

[Matthew Belinkie]

I’m using a comically large source layer for CC Sphere, so that’s not a problem. So the idea is to do a camera move in on the globe while it finishes rotating, so the right portion of the world is already filling the screen when the first comp ends? It certainly gives me a lot more options. But then when the second comp begins, the next layer of the earth zoom is already going to be filling the screen. Making the Earth bigger compounds the problem of how I gracefully transition from a spherical layer to a bunch of flat maps. Then again, maybe zooming in on the Earth in the FIRST comp lets me skip a layer of 2D map, so it balances out.

I suspect I’ll need to fudge this transition in a couple ways:
1. Use a layer of clouds to mask the layer fading in below, then fade the clouds away as we zoom “through” them.
2. Speed up the first part of the zoom, thus increasing the motion blur enough to mask the imperfections. So we start in space, and do a really fast zoom through the top layer of clouds, then slow down as we get to the country level.

Basically, the part where I’m layering a 2D map on top of a CC Sphere effects is never going to be perfect, so I want to spend as few frames on that as possible. Right?

[Matthew Belinkie]

I’m using CS4, if anyone wants to do further research on how 0% opacity effects rendering time. It may be that the rules are different for 2D layers (like the ones I’m using) whereas 3D layers would still need to be processed?

Anyway, even if I were going to manually trim layers, I still love your script because it shows WHERE the layers should be trimmed, and it adjusts dynamically as the comp changes. The project I’m working on will be 3 minutes of zooming in and out on various map locations, so I’m looking to keep things tidy and efficient.

[Matthew Belinkie]

Thanks a lot! Your expression had the desired effect – layers that were either very small or very large were automatically faded to 0% opacity. It didn’t look any different, and it actually DID render much faster. Without the script, the comp took 3 hours. With it, it took just over 1 hour!

To improve it even more, I found a script that automatically does what you suggest: it trims each layer that begins or ends with 0% opacity.
https://www.aenhancers.com/viewtopic.php?f=11&t=1096

However, I found that it did NOT reduce render time further. The version with the trimmed layers finished within a minute of the version that just had the opacity adjustment. So in this case, for whatever reason, adjusting the opacity to 0% IS just as good as trimming the layer.

Anyway, for anyone who is doing an earth zoom in the future, sticking this expression on all your layers (BESIDES the one parent layer) is a great way to reduce rendering time without impacting quality.

[Matthew Belinkie]

Thanks for the tip! But this is trickier than I thought; let me explain a little. I’m using the basic map zoom technique demoed by Andrew Kramer:
https://www.videocopilot.net/tutorials/earth_zoom/
You get a bunch of maps at various scales, parent them all to the first layer, and then scale that one layer up logarithmically. Works great, but it means that you’ve got 10 or so layers, and most of them are either too small to see, or giant (and therefore completely covered up by the next layer). It can get very processor intensive, and I thought that the expression you recommended would dramatically reduce rendering time by turning off the layers that couldn’t be seen.

The problem is that because every layer is parented to the first, their scales don’t change at all! So what I really want to do is make After Effects apply the expression to each layer’s “true” scale (the scale it would be if it weren’t parented) but still keep it parented. I know I’ve seen expressions that can use a layer’s “true” POSITION in the comp even if it’s parented, but can the same thing be done with scale?

Thanks!

[Matthew Belinkie]

Okay, I didn’t find a particularly elegant solution, but I got it done:
https://dl.dropbox.com/u/606016/Reel/Spin-Zoom.mov
What helped was using a very large image of the world to make my initial CC Sphere. It was about 6,000 across, and my CC Sphere had a radius of 1,300 pixels. That way, by the time I scaled it up to 100%, I was only looking at a tiny section of the map, and curvature wasn’t as big an issue. My next-largest layer in the world zoom was just Italy. If you stop on those frames, you can see that North Africa is pretty blurry. But nobody is looking there anyway.