Encoding Tests

A couple of days ago, I was able to make some encoding time comparisons – betwween 4.07 and 4.1. For curiosity's sake I also ran some tests comparing SL and Mavericks (on the same 2010 iMac)


There were some surprises – perhaps the biggest one being that Send to Compressor, a workflow that I've typically avoided, is now much faster using the latest versions of FCP and Compressor. Time will tell whether it also proves to be reliable. And I only tested it for a single setting, but the results were impressive enough to encourage its use going forward. Another surprise was that the "good old" Quick Cluster days for this particular target weren't so great after all. YMMV – especially with other targets and longer sequences.


My source clips were 1080i – the Canon MPEG encoding in an MXF wrapper. I assembled those into a simple 6 minute and 25 second sequence with no effects added. My objective was to produce a 720p 24 fps mp4 file. So there were multiple processes in this job: de-interlacing; down-scaling; re-timing and transcoding. I used a 2 pass quality setting for all tests.


Here are the Mavericks results:


In 10.1 Share>Master File Pro Res 422, 2:44 (quick, as expected). In 4.1, encode to mp4, 15:.38 Total combined time from native FCP sequence to mp4: 18:22. (not very impressive – but again, a lot going on – and it was multi-pass).


In 10.1, Share > Apple Devices>Settings>720p, 12:14.


In 10.1, Send to Compressor>mp4 setting, 10:07 (surpise #1).


In 4.07, Quick Cluster (3 instances), the Pro Res Mastr file encoded to mp4 in 16:22; so total time from native FCP sequence was 19:04. (surprise #2; requires more investigation).



Here are the 10.0.9/Snow Leopard results; (expecting everything to be slower…and I wasn't disappointed):


In 10.0.9, Share>Master File, 3:04. (pretty quick, but 11% slower than 10.1 on 10.9). In 4.0.7, Quick Cluster (3 instances) to mp4, 25:56. Total combined time, 29:00 (painful).


In 10.0.9, Send to Compressor>mp4, 40:17 (yikes!).


Russ






Posted on Dec 26, 2013 9:47 AM

Reply
13 replies

Dec 27, 2013 4:47 PM in response to Russ H

Here are my numbers with 4.0.7 and 4.1... Video 1280x 720p... Encoded to h.264 and x264 (2 passes). Computer, 4 cores and 4 hyperthreads for a total of 8 cores. Running OX 9.1 (Computer specs below)


4.0.7 -

h.264 24:16, used about 600% processing power. (4 instances, 8 segments)

x264 23:00, used about 750% processing power. (4 instances, 8 segments)



4.1

h.264 30:00, used about 500% processing power. (2 instances, 4 segments)

x264 21:51, used about 750% processing power. (2 instances, 4 segments)



It's not how many instances the computer is using to make a faster encoding... it's the amount of possessing power the computer is using. In fact segmenting the video can take longer to encode (using one computer). Instances/segmenting really only comes into play when your using a rendering farm (more than on computer to encode the video). But then again 4.1 is slower than 4.0.7 when encoding to h.264.


I do like the new layout!!!



MY Computer:


27" iMac mid-2011


3.4GHz Quad-Core Intel Core i7

16GB 1333MHz DDR3 SDRAM - 4x4GB

2TB Serial ATA Drive + 256GB Solid State Drive

AMD Radeon HD 6970M 2GB GDDR5

Dec 27, 2013 5:34 PM in response to Russ H

BTW... The above video was a 30 minute video. To use the qmaster (multiple instances) in both versions, I could only have one version installed at a time.


When I'm up to it I'll try the same video at 1080p and maybe 4k. I know a 1080 video will push and use more of the computers possessing power.


(Apple's pushing 4k and it would be nice when they come out with an h.265 plug-in. I was able to encode the same video above (using the 1080p version) to h.265 at 6,000 kbits/sec. At 6,000 kbits/sec I can't tell the difference from an h.264 encoded at 19,000 kbits/sec. I saw blocking at around 3,000-4,000 kbits/sec using h.265.)

Jan 4, 2014 9:30 AM in response to David M Brewer

If you haven't seen it, Larry Jordan dot Biz has published an article on compression, comparing the new MP with the iMac. Interestingly, there is an embedded link to another LJ article on when and when not to use multiple instances.


Russ


PS, The ACS software will not let me post links today. But you shouldn't have any trouble tracking it down.

This thread has been closed by the system or the community team. You may vote for any posts you find helpful, or search the Community for additional answers.

Encoding Tests

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple Account.