[This is an excerpt from a recent on-line video training: Video Compression Comparison, which you can download from our store. You can also access all our training when you become a Video Training Library subscriber.]
Video compression can take a long time. And, with Apple Compressor, two attributes which can improve compression speed are turned off by default.
If you are compressing files which are stored on a network server, you can significantly improve compression speed by building a “compression network” by combining multiple computers into a single compression engine. In this video, Larry Jordan explains how to create it, when this network is a good idea and when it is not.
TRT: 4:40 — MPEG-4 HD movie
NOTE: This video may not play inside FireFox, if so, please use a different browser.
Our website offers lots more video training in our store.
One Response to Compressor: Build a Compression Network [Video]
Hi Larry,
I am enthralled by compression and find it as mystical and confusing as colour correction; something else I struggle with. I have one of your earlier Webinars on compression and so only watched the above excerpt. I am amazed at how compression times change depending on what recipe you use in which program but also whether you run it directly from the timeline or from an exported file.
Anyhow, after watching the above excerpt I ran a short experiment with a movie I’m still working on but have an early draft of as an exported file. It’s 1080 25p (UK resident), was captured to a field recorder in ProRes 422 and as an early draft it’s about 9 minutes long, weighing in at around 8GB. I discovered that my Late 2013 iMac (3.5GHz, i7) with 16GB RAM allowed me to set one additional instance of Compressor so I gave it a shot with the ‘Apple Devices HD (most compatible)’ 960×540 setting and got a time of 4’04”, which I thought was pretty good. Then I thought I’d try a similar setting using my Matrox MX02 Mini with Max, through Compressor, and it failed. I tried it twice more before identifying that the error code was showing the Matrox reported to be already running with Compressor. Of course, two instances of Compressor and only one Matrox is never going to work and perhaps it’s not really that surprising. So I reverted back to just one instance and got a satisfying Matrox encoding time of 3’10”.
The strange thing, however, was when I went to perform the ‘control’ experiment and repeat my first encode but with just the single instance of Compressor running. This time, it all went smoothly but the encode was 3’55”, almost 10 seconds faster.
No wonder this stuff is such a mystery!
Keep up the good work!