Brian's Brain

Mar 27 2008 10:51AM | Permalink | Email this | Comments (7) |
Blog This! using:  Blogger.com | LiveJournal |
Digg This | Slashdot This | add to Del.icio.us

Following up on my 3+ weeks-back experiment, last night I finally found a high-def flick from the iTunes Store worth of a $5 rental test. It was The Darjeeling Limited (given my long-standing love of India-influenced Nepal, are you surprised?), and I started the download at 7:39 PM. Here are some stats to refresh your memory of my prior standard-def test over my recently bandwidth-upgraded 2.5 Mbps downstream DSL connection:

• The total runtime of Saved! was 1 hour, 32 minutes, and 22 seconds.
• Each percent-downloaded 'tick' took ~35 seconds
• Apple TV reported the movie as 'ready to play', thanks to the service's progressive download capabilities, at the 2.5%-downloaded mark, or said another way, less than 2 minutes after I concluded the rental transaction and the download began.

Last night's high-def experience was quite different, much more so than I would have predicted in advance. To wit, I didn't realize just how much bigger the high-def files served by the iTunes Store were, versus their standard-def alternatives.

• The total runtime of The Darjeeling Limited, as reported by the online store description, is ~91 minutes.
• Each percentage-downloaded 'tick' took ~1 minute, 35 seconds (i.e. roughly 3x longer than the standard-def predecessor), and therefore
• Apple TV reported the movie as 'ready to play' at the 43%-downloaded mark i.e. 68 minutes after I concluded the rental transaction and the download began.

Not exactly instant gratification, was it? In fact, I haven't yet started viewing the movie; by the time I got the 'ready to play' notification, it was almost 9PM, and I didn't want to stay up for another 1.5 hours. I'll watch it tonight and let you know if I encounter anything odd. However, given the data above, I can extrapolate some other statistics that you might find interesting:

• Assuming my DSL line's 2.5 Mbps downstream potential held constant through the entire download cycle, the movie was completely stored on my Apple TV's HDD after 2 hours, 38 minutes and 20 seconds (or, said another way, ~0.6x real-time). Compare this to the 58-minute (~1.6x real-time) total-download time for standard-def Saved!, and the high-def delay penalty (along with the reason for the extended progressive-download latency) will be quickly apparent.
• The Darjeeling Limited's total file size is, if I've done the math right, approximately 3 GBytes (i.e. roughly 3x the size of the standard-def predecessor). Keep in mind, too, that each anamorphic-encoded high-def video frame is 1280x534 (683,520 total) pixels in resolution since the film employs a 2.35:1 aspect ratio, versus 854x480 (409,920 total) per-frame pixels for the standard-definition 16:9 aspect ratio Saved!, whose total file size I'd earlier estimated as ~1.1 GBytes. Considering the aspect ratio, therefore per-frame resolution, differences between the two films, I'm unable to draw any definitive conclusions regarding standard-vs-high def compression efficiency of the H.264 (aka MPEG-4 AVC, aka MPEG-4 Part 10) video codec that Apple employs.
• I also estimate the average playback bitrate for The Darjeeling Limited to be ~4.4 Mbps. Compare this to the earlier ~1.6 Mbps average playback bitrate estimate for Saved! Keep in mind, however, that I strongly suspect (though I don't know for sure) Apple's using VBR (variable bitrate) encoding for its iTunes Store-served rental and purchased video content, in order to maximize video quality across a range of content compression complexity. The playback bitrate could therefore wildly fluctuate on a frame-by-frame basis, which greatly complicates progressive-download calculations and makes streaming near impossible (leading me to also suspect that the movie previews are constant bitrate, i.e. CBR, encoded).

I'm half-tempted, after watching the movie tonight, to spend an incremental $3.99 on the standard-definition version of the film, in order to provide a more apples-to-apples comparison. Stay tuned for a follow-up report if I decide to drop the incremental cash...