H.264 Blues: Why Isn't Starbuck Smooth?

Continued from ‘NAB 2008: Is Full-Screen Hulu Jerky For You, Too?‘…

I first viewed the first few minutes’ worth of Season 4, Episode 6, and I terminated playback at the same scene in all cases. This is standard-definition material that, in fullscreen mode, requires substantial upscaling to fill the LCD. For contrast, I then accessed the trailer for the movie Street Kings, which being natively high-definition requires less system-side upscaling prior to fullscreen playback. For the Dell, I played back the content over both the laptop’s integrated 12" widescreen LCD and the VGA-tethered Hanns.G 28" external LCD. With the MacBook, I captured screenshots only in conjunction with the integrated 13.1" widescreen LCD. 

Dell Inspiron 700m, Battlestar Galactica, normal playback window, integrated LCD

Dell Inspiron 700m, Battlestar Galactica, fullscreen playback, integrated LCD

Dell Inspiron 700m, Battlestar Galactica, normal playback window, external LCD

Dell Inspiron 700m, Battlestar Galactica, fullscreen playback, external LCD

Dell Inspiron 700m, Street Kings, normal playback window, integrated LCD

Dell Inspiron 700m, Street Kings, fullscreen playback, integrated LCD

Dell Inspiron 700m, Street Kings, normal playback window, external LCD

Dell Inspiron 700m, Street Kings, fullscreen playback, external LCD

Apple MacBook, Battlestar Galactica, normal playback window, integrated LCD

Apple MacBook, Battlestar Galactica, fullscreen playback, integrated LCD

Apple MacBook, Street Kings, normal playback window, integrated LCD

Apple MacBook, Street Kings, fullscreen playback, integrated LCD

H.264 thoughts? Well, to paraphrase Frank Sinatra, I have a few:

• Generally speaking, full-screen playback of the high-def trailer was smoother than that of the standard-def Battlestar Galactica material. I realize that I’m in a sense comparing apples and oranges here; it would have been better for me to have access to standard- and high-definitions of the exact same content. However, I still feel confident in saying that per-frame up-scaling seemed to seemed to consume more of the CPU’s resources than did the fundamental video decode task, even though the H.264 decode algorithm ran entirely on the CPU.
• Note the undulating CPU utilization (unless, of course, the processor was completely swamped the entire time) throughout the playback duration. This inconsistency reflects varying within-frame and frame-to-frame complexity throughout the clip.
• Focus on the MacBook profiles and it’ll be immediately apparent to you that the Flash 9 Player plugin is multi-threaded; the benefits both of the higher per-core clock speed and dual-core attributes of the MacBook’s processor versus its single-core predecessor in the Inspiron 700m were very evident to my eyeballs.

And versus my weekly Lost-on-On2 experiences? Again, definitive conclusions are difficult to ascertain; not only am I comparing completely different content (albeit with some common big-picture characteristics….lots of frame-to-frame movement and luminance changes, etc), but now I’m also comparing Adobe’s Flash playback engine against ABC’s VP7-based Full Episode Player. With that said, what I can tell you is that whereas full-screen 1920×1200 playback of H.264-encoded high-def source material on the Inspiron 700m is definitely a sub-optimal ‘jerky’ experience, the VP7 counterpart is ’smooth as butter’, as the saying goes, and with quite acceptable quality.

I realize that in a few years, when H.264 hardware decode support is pervasive and the bulk of today’s PCs and other video playback widgets are retired, this study will largely be a moot point. But for now, my conclusion is that Bill Joll’s NAB claims have a lot of merit behind them. What say you, readers?