Magic Music Visuals

Getting ready for a performance in two weeks, here's my question:

Will it be smarter to have one long (10-minute) video file with all my assets placed at 20-second intervals,
and then call the position of those elements by a time position?
Or, have some 30-odd separate files/video assets? Or will it even make a difference?
I already have the video as one giant file, since I was creating a consistent look to the whole batch.

If you want an idea of what I am talking about, listen to this ditty, jump in to about the 3-minute mark;
every audio file sample has an associated video.
https://soundcloud.com/robertdorschel/d ... -broadcast

Just wondering about performance general practices, as well as ease of loading/reloading assets.

Great question. It actually depends quite a bit on what video codec you use, and the specific encoding settings.

The reason is the video keyframes. If the keyframes are spaced exactly at every 20 seconds, you won't have any issues. It's fast to jump to keyframes. But if you have to jump to any other frame, all the intermediate frames (starting at the previous keyframe) have to be decoded, which is a major performance hit. It works this way in pretty much any video playback application.

So, I think the simplest way to go is to have separate files. That way you don't need to worry about encoding settings, because the first frame of a video file is always a keyframe. It's always fast to jump to frame 0.

If you want to read a bit further: https://en.wikipedia.org/wiki/Key_frame ... ompression

*Edit: one thing I should add is that some codecs, like M-JPEG, treat every frame like a keyframe. So you could jump to any frame without a performance hit. But M-JPEG video files are much larger as a result (because they store more information). Thus, a faster hard drive is necessary. See https://en.wikipedia.org/wiki/Motion_JPEG#Encoding

Great, thanks!
I'll let you know how it turns out once I get the scene trigger layout completed.