In the example, 1.66:1 films would be letterboxed on 4:3 screens. But that aspect ratio needs to be pillarboxed on 16:9 screens to maintain the correct shape. I agree that this is poorly written, because it only seems to take into account one of the two possible scenarios for aspect ratios that are not natively 4:3 or 16:9 (we deal with lots of them daily, because we do a lot of film-originated work, so 1.85:1, 1.66:1, 2.35:1, etc.
I guess what it's saying is that if you're going to make both SD and HD streams available and you have a non-video aspect ratio (say, 1.66:1), you'd make two files:
SD at 720x432 (this will be letterboxed by the player to fit a 4:3 screen)
HD at 1800x1080 (this will be pillarboxed by the player to fit a 16:9 screen)
Or in the case of my question, where the source is a 2k 4:3 film scan:
SD at 720x480 (this will be full frame on a 4:3 screen)
HD at 1440x1080 (this will be pillarboxed by the player to fit a 16:9 screen)
Is that how it works?
Thanks!