the official website of the dallas final cut pro user group

As I’ve written about previously our department is undergoing a massive re-encode, about 1600 files. We were previously using Sorenson Squeeze 4.5.7. It’s main problem revolved around the time it took to transcode a Quicktime movie into a .mp4. To create one mp4 took about three hours. And for each Quicktime file we need to output two mp4’s and one mp3. Now 6+ hours to get through one file is just rediculous.

It became important to take advantage of our multi-core machines, which Squeeze 4.5.7 wasn’t doing. We decided to take the plunge and upgrade to Squeeze 5 because of it’s claims to utilizing multiple cores. Come to find out that it does and it doesn’t. It does utilize multi-core machines, but only in certain instances. For example, if you important files manually it can use all the cores you have. If, however, you use a watch folder to encode files you only get one core of fun. Now that seems pretty short sighted. Watch folders, in a multi editor studio is of the utmost importance. Since everything we do is in a networked envirnoment watch folders make the encoding process much simplier.

For those who may be unaware of what or how a watch folder works, let me explain. A watch folder is a designated folder that is “watched” by the encoding software. Whenever a file of an appropriate type, say a Quicktime .mov, is placed into the folder the encoding software will transcode the file into whatever settings are specified. So from one watch folder you may have five different output formats. In other words, one .mov goes in and out comes a mp3, mp4, wmv, flv, and mpeg-2. This makes repetative tasks easily managable. And it makes repetitative taks that come from multiple editors easily mangable through one piece of software.

The upside to Squeeze 5 is when it does take advantage of your multiple cores it allows you to rock ‘n roll. Instead of doing 5 files in a 24 hour period, my 8 core machine was cranking through 26 files in the same time. Now that’s a serious time saver! My recommendation to Sorenson is to tweak their programming so that multi-core encodes happen whether you import manually or setup watch folders. The software has an extremely intuitive interface with simple, easy-to-use plugins. For the price point and its feature set, not much can touch it. Overall, I give squeeze a 7 out of 10.

So now that I’m utilizing all my cores I’ll keep ya posted on how long it takes to transcode 1600 files! Maybe one day we’ll have a faster solution to these kinds of tasks….but until then….

Last week I posted how we were in the midst of searching for a way to adequately encode almost 2000 files using the h.264 codec into some sort of wrapper. The problem was deciding what type of wrapper to use. Should we use .mp4, .mov, or .m4a?

The problem centered around how long it took Sorenson Squeeze (our transcoding software) to create a mp4 from our full quality mov. What I failed to consider was how much unnecessary work Squeeze was doing. For instance, we exported from Final Cut Pro a DV25 720×480 interlaced timeline. Squeeze had to resize the file down to our output resolution of 640×480, deinterlace the footage because we were delivering via web (and you don’t put interlaced material on a progressive display like a LCD or Plasma), and then create a new mp4 file. So Squeeze took anywhere from 4-6 hours just to do 1 file. That just isn’t going to work. I’ve seen mud move faster. Hence our dilemma.

We looked at Compressor as a solution, but it didn’t offer a mp4 file using the h.264 codec. If we want h.264 it either meant wrapping it as a m4v or as a mov. The solution: stick with Squeeze but take some work off the programs shoulders, force FCP to do much of the grunt work upfront. So when we batch export from FCP our sequences were manipulated from FCP into our target resolution (640×480) and deinterlaced. Encoding a file that Squeeze no longer has to resize and deinterlace took the transcoding time down to 1-2 hours. Much better. Still not light speed, but not as slow as molasses either.

What we learned: sometimes its easier to let FCP do the grunt work, like resizing or deinterlacing, and then let a secondary software, like Squeeze, do what it’s made to do, namely, transcode.