HD For Indies

13-Inch MacBook Benchmarks 1

Benchmarks from Final Cut Studio. They compared a Dual 2.0 GHz G5, a 15 inch MacBook Pro (2.16 GHz), and a MacBook (2.0 GHz). First up was Final Cut Pro itself - the MacBook is effectively as fast as the G5 for render tests - tie, faster, or slower by 1 second (1:09 vs 1:10 minutes).

Mike's Comments: that said, these were some pretty stupid tests. They use DVCPRO HD, and that's great. They used 3 way color corrector, and that's great. Then they used plugins - that's a good idea.

But then they used the following filters, and let me just classify them as I see them applying to most indie productions:

Realistic constant daily use (smart):
-cross dissolve
-Color Corrector 3 way
-Color Balance
-Unsharp mask

Use occasionally, once or twice a day (maybe OK, don't overuse)
-Gaussian Blur
-Fisheye (I could see using this to correct lens distortion)
-center split side
-channel map
-color key
-Luma Key

I Would Slap You If Used In Serious Production (just plain dumb):
-curl transition
-Swing transition
-pond ripple effect
-no, no they didn't - yes they did! The dreaded lens flare
-page peel
-ripple dissolve
-Wind Blur


...you get the idea. To me, the tests I'm interested in doing as soon as I can would involve mapping the limits of realtime effects:
-can I get a 3way color correct on both clips around a cross dissolve in realtime?
-can I throw a scale on top of that? One or both clips?
-can I put text and/or a lower third on that in realtime as well?
-can I do these in DV? In DVCPRO50? In DVCPRO HD? In HDV? In uncompressed SD? In uncompressed 720p24? 720p60? 1080p24? 1080i60? Etc.
-do yes/no RT testing for full res/full speed, then map out the Dynamic RT performance.
-then test render times for the above, plus some other common color/repo stuff.

I find that covers 80% of day to day editing for storytelling content.

Also, their numbers reflect editing to an external QuickTime file, NOT rendering on the timeline. Again, this is a task that is done, but is done FAR less often than timeline editing. It's good to get preliminary results, but more realistic tests are necessary. They talk about getting faster drives to see if that helps - I seriously doubt that it will - in my own tests on dual and quad G5s, whether rendering to FireWire drive or 8 drive SATA RAID didn't make a difference. If rendering to any compresed format, the data rates involved are well below drive capabilities.

I think it is worth testing the oddball filters, but there are MUCH more important things to test first...you KNOW the programmers are going to spend their time on the common stuff - the color correctors, the basic dissolves - before they optimize the *&*^&%$^^ wind filter. I mean, come ON.

Wind filter? Verrrrrrrry rare special effect shot, and then only maybe. Ripple dissolve? Only as a gag, again 2 or 3 per production tops. Lens flare? My god, that's why I got out of doing mographics and VFX...

(and as long as I'm in a mood to bitch, just watched the trailer for Ghost Rider, and BOY, does that look dumb! How much did they spend on it?)

-mike

(thanks to Jarred Land of the excellent DVXUser.comforum based site for sending in a link to me about this one.

Oh, and they also did some further tests with Motion 2.1 - and the G5 (with an ATI 9600, a lower end card) didn't beat out the laptops by much, even in particle heavy tests (sounds like they did a more realistic Motion test for this series than they did for FCP). The MacBook was actually faster in 3 out of 4 tests, the one where it was slower involved uncompressed video footage, so drive speed may well have played a factor (bus speeds too perhaps?)

Interestingly, the MacBook actually BEAT the Dual G5 in 4 out of 5 Compressor 2.1 encoding tasks, such as SD & HD MPEG-2 encoding, as well as H.264 encoding. The one test where it lost was so short (3 vs 4 seconds) that it is effectively a tie and they should have used a longer test sequence.

This certainly implies that the desktop Macs are going to SMOKE if they are appreciably faster than the laptops.

So MacBooks? Yeah - a real production machine, if you can live with the lack of high speed disk attachment.