Johnny is correct. Lets say you have two 60 sec animations. The first being a simple talking heads video with very little changing throughout. The second being a high-impact motion graphics speedfest. When you use a lossy compressor like .h264, you will see a huge difference in file size. This is because compression throws out or reuses repetitive material as much as possible to keep bit rates low. The talking head video is very repetitive visual, so its data size will be much smaller than something that is changing constantly.
Now if you render those out to what is considered more lossless codec, like animation or Prores 422 or 444, you will notice that the file sizes are much more similar. That is because lossless codecs make each frame independently from the ones around it, hardly using any compression at all.
It isn’t a good idea to render out .h264 from AE. That’s not its job. AE is not a very good compression app. Johnny is correct in his workflow. Not only can you see an estimated data size, but you will get better results with smaller data sizes.
Seems like an unneeded step, but always make a master in a lossless codec and then use a good compression application to make your distributables. That way if someone asks you for an .flv or .mpeg or whatever, you will be able to quickly compress that from your master file and not have to bother with AE.
Hope this helps,
Reiner
Mac Pro 2 x 3 GHz 6-Core Intel Xeon
32GB Memory
Dual-channel 4Gb Fibre Channel PCI Express card
Dell Display (23″ flat panel)
ATI Radeon HD 5770
AJA Kona LSi SD/HD capture card
Rourke 16 TB
Flanders 2460
SOFTWARE
Mac OS X 10.6.5
FCP 7
After Effects CS5
Boris Continuum
Sapphire Plug Ins
All Trapcode Plugs
Zaxwerks Invig