Jun 21, 2010

Mac Encoder Shootout: Compressor 3 vs. Episode 5 vs. Squeeze 6

Every editor no matter what system they're cutting on has always had the same level of frustration after creating their final masterpiece:  At the very end of their blood, sweat and often tears-effort to bring their visual masterpiece to it's final iteration they are then forced into the time-consuming and gut-wrenching task of compressing the final movie into it's end format, be it DVD, web or even Blu-Ray.

In Mac-land the most often used - and complained about - tool is the Compressor application bundled into Final Cut Studio.  A quick tour of the various web-forum threads will show a plethora of users knee-deep in head-scratching sessions that the default Compressor settings not only don't give optimal results but the final encoding process - or compression - takes painfully long to complete.

Several low-cost third-party offerings have surfaced over the years in an attempt to best Apple's own compression tool and some had widespread success in years past amongst the indie filmmaking community, such as Handbrake, BitVice, MPEGstreamclip and others.  Most of them do a good enough job for the average editor or those who don't require or aren't familiar with things like high-level output, creating transport streams for broadcast or controlling the final output color space, something most if not all of the freeware applications are capable of handling.  But, back in the days of FCS 1 and the now defunct version of Compressor these low-cost or free options often did in fact outperform Compressors lack-luster speed, but not by a huge margin.

However today things have changed; Compressor is now able to create virtual clusters on machines that have multiple processors (this is best suited to quad-core machines, not the Core 2 Duo laptops or iMacs) and has more advanced options for output in fact, Compressor 3.5 (bundled with FCP 7/FCS 2009) now has a very limited option for Blu-Ray export, making the freeware offerings invalid and outdated.

However, despite the increase in capabilities one thing still reigns true over Compressor's many abilities:  It's dog slow, even when using virtual clusters.  (See Brian Gary's Quick Reference Guide on Compressor 3 on the how & why to setup a virtual cluster - but beware: It's not easily understood and an improperly setup system will actually cause problems in encoding).

Have no fear, there are now viable options to Compressors snail-like pace for making your final encodes.  Enter Episode Encoder (Previously called Episode Desktop) and the newly released Squeeze 6.

The testing machine was a 2.5Ghz MacBook Pro running 10.6.2 with a 2TB external array connected via the Sonnet Pro eSATA card.  Compressor 3.0.5 was tested, not  the current version 3.5 mainly because most FCP users have not yet upgraded their FCS2 version for the very reason I did not: it's really a service pack and not a full update to the suite.  Hopefully Apple will grace us with a "real" update later in 2010.

Going head to head with Compressor is Episode Encoder Pro 5.3 and the newly released Sorenson Squeeze 6.  This update to Squeeze is long overdue as Squeeze 5 has been around for almost 3 years with no updates.

The tested file was a 10-second 720p30 DVCPRO-HD clip exported as a stand-alone file from FCP.  Although both Episode and Compressor can talk to reference files for encoding it was determined that to level the playing field for testing that a self-contained QT file was most appropriate.

The 2 most common final encoding tasks are MPEG-2 for DVD authoring along with the AAC audio file and; H.264 for various web and mobile content delivery methods.

For each encode type, both MPEG-2 and H.264 a built-in preset was used for "best" quality in each engine; I used the advanced settings in each application to ensure that, BITRATE, FRAME RATE, VBR SETTINGS, KEYFRAMES  and DE-INTERLACING modes matched up line-for-line.   Each run was done 3 times with the best time out of the 3 posted as the final result.  And the results?

MPEG-2:  10-second clip to MPEG-2 widescreen w/AAC audio

Compressor 3.0.5:                     - 1:52
Squeeze 6:                                   - 1:12
Episode Encoder Pro 5.3:            - 0:28

I've known for some time that Episode was a speed demon compared to Compressor, but this end result surprised even me.  And what's more, is that the final movie in DVDSP4 actually had *better* overall discernible quality than either Compressor or Squeeze!  I still shake my head in disbelief.  On to the next test:

H.264:  10-second clip to 640p Widescreen w/192kbs audio

Compressor 3.0.5:               - 2:08
Squeeze 6:                          - 2:35
Episode Encoder Pro 5.3:   - 0:57

Again, Episode renders a virtual smack-down to it's rivals and Compressor offers up a slightly better time than Squeeze, however the overall quality difference between the 3 encoders is negligible; none of them had any visual superiority over the others.

Now keep in mind the test clip was only 10-seconds long, yet 2 of the 3 encoders took over 2 minutes long to encode it to H.264!  In fact Squeeze took 2 1/2 minutes!  Really?!

(CAVEAT:  Squeeze has proven to be buggy with H.264 encoding on 2 machines that I've tested and I've been able to cause the app to crash repeatedly when doing the H.264 encoding tests.  Clearly something is amiss with this newly released encoder and it could be that the long encoding times are relevant to this bug.  To be sure I'll be delving into the cause of it's maladies but for now the results are what they are).

The good news:  There is a clear winner in this head-to-head speed contest, Episode.

The bad news:  The "pro" version of Episode costs exactly as much as the entire FCS suite!  OK, actually it's a whopping $4 dollars cheaper... to be exact.

For most indie filmmakers who don't output their finals to broadcast and don't often author DVD's for end-user clients the cost of Episode Pro doesn't add up.  But, if your company does a great deal of MPEG-2 encoding for DVD or, you need to make a ton of H.264 files for clients on a regular basis or, if you need to make transport streams for broadcast then the speed offered by Episode could easily pay for itself in much higher productivity.

Keep in mind that my test file was all of 10-seconds long; multiply that by the typical run-time of a DVD project or final movie, which averages out to be 40-70 minutes and the time spent waiting for an encode to finish becomes exponentially shorter using Episode than Compressor.

As an example:  I had an HD final that needed to be compressed to MPEG2 for widescreen DVD for a client.  Time was on my side so I ran a test using that final movie and went head-to-head again using Compressor and Episode.  Compressor was literally an all-night encoding job.  It started at 9pm and finished when I woke up at 6am the next morning - actually it was still crunching the file until 6:45.

By contrast I put that same HD final into Episode Pro at my lunch break that same day, around 1pm, ran a few errands and made it back to my office in time for dinner around 6pm.  The encode was finished by the time I arrived.

You do the math.

1 comment:

  1. Now this is in actual fact cooperative. It’s very openhanded of you to share this with us.
    thai dating