In a November 2007 article, Posting HDV: How Much Power Do You Need for Speed?, I wrote about some of the practical options that were available to event videographers making the switch from the DV to HDV workflow on the Windows XP/PC platform. This time around I’ll be looking at some of the same issues on the Mac OS side of the editing world.
While there are many editing options on both platforms, the PC article focused on Adobe Premiere Pro CS3 (which is also available now on the Mac) and tested for postproduction performance in three scenarios: native HDV editing without additional hardware, native HDV editing with the assistance of the Matrox RT.X2, and HDV editing using the CineForm intermediate codec. Since most Mac editors use Final Cut Pro (FCP) and there are more options available for it, I used FCP for my experiments. While I’m a longtime Premiere Pro user, I had little difficulty learning FCP for this article because 90% of the commands are the same.
For the purpose of these tests Apple supplied a very nice, top-of-the-line Quad Core Mac Pro. The specs are two Dual Core Intel Xeon processors, 8GB RAM, four 500GB hard drives, and an ATI x1900xt 512MB graphics card, with the system running OS 10.5 Leopard.
The Tests I Ran
I tried three different options for editing: editing 1080i native in FCP, using the Apple ProRes 422 intermediary codec, and CineForm’s visually lossless intermediary codec. On the PC side, there are accelerator cards to speed or eliminate rendering, such as the Matrox RT.X2 and Axio cards. While there are cards from AJA and Blackmagic Design as well as the external MXO2 from Matrox to get video into and out of the Mac, none of these products provide real acceleration on preview rendering for editing.
As a result I ran all of my tests using the software solutions only. I could have used the AJA Kona LHe card to capture HDV decompressed over its component analog inputs, but the company sent the Kona 3 that only had HD-SDI and SD-SDI inputs, which were not much use in our study other than the ability to output to an analog monitor. If I had the correct card, it would have been less taxing on the CPUs, but it would not have been very space-efficient.
A Word About the Mac
Before I get into the numbers and spreadsheets, I’d like to talk a little bit about my first real hands-on experience with an Apple computer since 1992, when I chose an Amiga 2000 and a Video Toaster as my first (linear) editing system over an early Mac-based NLE that only did postage stamp-sized offline video. I did find a lot to like about the Mac Pro and its OS X Leopard operating system.
Having no experience on the Mac, I was able to easily learn to navigate the file system and install programs without calling for tech support. I even used Boot Camp to install Windows XP and make it a dual-boot system with ease. I found the internal configuration very different from the HP workstations I am used to. I’m someone who likes to dive in and upgrade his own machines, and there isn’t a lot of room to do so in this one. I brought the Mac Pro test system to a local association meeting and asked if anyone would like to see the insides. Final Cut aficionado David Robin of david robin | films came for a peek with others who had never looked inside their own Macs. “Looking inside is for techies, not artists,” I was informed.
Sliding the full-length AJA Kona/Xena card in took a lot of wiggling. The way the hard drives go in is nice, but if you need to remove one of the first two drives, the Mac Pro will draw blood unless you wear gloves. The fact that you can’t add off-the-shelf internal DVD burners is annoying. I found the “mighty mouse” that shipped with the test very un-ergonomic. I donated it to a lab that does animal testing on rodents and replaced it with a standard USB mouse. The way the RAM was installed was kind of neat in the way it was on its own back planes; it’s also the easiest of the hardware to upgrade. Overall, I found the Mac Pro a very nice machine that will give you excellent performance no matter what your OS.
The tests I ran to compare HDV postproduction performance on the Mac using the two intermediate codecs and going native are a sampling of the many different types of edits and effects many may use in editing event videos. By looking at the results of these different tests you can gauge how long these various operations would take using the different editing codecs in real situations.
From the test results you can extrapolate how long renders and previews will take in one of the methods reviewed here. For instance, if you like to use FCP’s Posterize filter with the CineForm codec on a 6-minute clip on a comparable system, it will take about 14 minutes and 45 seconds to render that section of the video.
Here are the tests I did:
Test Two: One-Minute Photo Montage. In this test I animated still frames over a couple of Digital Juice backgrounds (uncompressed HD AVI). Handling still frames is a common requirement of NLEs in event work. In this test all of the elements were exactly the same, with no proprietary codecs. Here, CineForm was nearly 4 minutes quicker to preview and was tied with native HDV for export at 0:16. ProRes 422 trailed at 7:35 to export.
- Test Three: One-Minute Color Balance. This test is just what it sounds like: I color-balanced an HDV video clip, previewed, and exported the clip. All codecs previewed in real time. Export was similar on all three with ProRes coming in the quickest at 2:02.
- Test Four: 10 Seconds of RGB Color Correction. In this test, the video clip was color-corrected using the RGB Color Correction filter. All codecs previewed in real time. Native HDV was the winner on export at 0:29.
- Test Five: One Minute of Posteriori. All codecs previewed in real time in this test. Native HDV, at 2:25, was the winner by 2 seconds over CineForm.
- Test Six: Five-Minute Project Rendered to Standard-Definition DVD Stream (.m2v). Here, ProRes 422 won by nearly 6 minutes over native HDV and more than 8 minutes over CineForm.
- Test Seven: Five-Minute Project Rendered to HDV Stream (.m2t) for Blu-ray. ProRes 422 won this round by nearly 4 minutes over native HDV and more than 15 minutes over CineForm.
Interpreting the Results
All in all, there was no clear winner in these seven rounds of testing since all the codecs won the same amount of tests. Nonetheless, here are some things to consider. When editing native HDV, you may run into issues doing effects and issues with compression artifacts, and you will likely find that working with the HDV codec will bog down lesser systems.
ProRes is part of FCP and decompresses the HDV so it will retain its quality and require less processor power. CineForm is similar to ProRes 422, but it is a $249 add-on for the encoder. The Neo Player is a free download. Here’s how it works: You can have the CineForm encoder on one computer, either a PC or Mac, and once your footage is encoded, it can be played back on any system that has the free player installed. CineForm would be more beneficial if you wanted to edit in a non-FCP application, like Adobe Premiere Pro, that won’t work with ProRes. In the end, you may still want to look at the AJA or Blackmagic Design cards or Matrox’s MXO2. Even if they don’t accelerate the processing, they will allow you to send an HD or SD signal to an external monitor and allow for different types of analog and digital inputs.
Marc Franklin (marcfvp at yahoo.com) has been shooting video since 1982 and has run Franklin Video Productions since 1992. He has been featured in the Hollywood Reporter, Forbes, and TV Technology and has written for WEVA.