This is just noddling with the encoder, mind you - the bitrate is a relatively-low 25 Mb/s, and the settings weren't as fine-tuned as they would be if I were actually creating a transfer for official replication. With these settings I can whiz by at 5-6 frames per second... at "professional" levels, I'm lucky to get 1.25fps, and that's with all four cores running at 100%!
BD25 re-encode (TOP) and BD50 original (BOTTOM), lossless PNG:
All of these are exact frame matches using the same decoder, and saved to a lossless compression format: for all intents and purposes, this is what's on the original disc and what the single-layer transfer I've concocted looks like. I'm not dancing around looking for the best samples, either; a few of these were chosen specifically because they were some of the worst frames I could find, and by "worst" I mean "the least like the source frame". When you have the luxury to A/B the two, side-by-side in different tabs or whatever, you can see a slight level of softening, a minor shift towards 'smoothing' in grain structure, and some very minor blocking artifacts on the re-encode... but even with less than optimal settings, the difference are much more subtle than I had expected.
Still, having shaved a third of the original bitrate off I don't think these are anything to sneeze at: the source was 37 Mb/s, and the single layer transfer here is 25 Mb/s, which - after two lossless stereo tracks - means there's still ample room for SD bonus features without requiring a dual layered disc. Make no mistake, the 37 Mb/s transfer is a slight improvement and there's no sensible reason for a commercial release to limit themselves to BD25, but it's nice to know that even a freeware encoder using bitrates that'd fit a 2 hour film on a single layered disc are still "good enough" for everyday use. Keep in mind the above examples aren't even x264 working at its full potential.
So, to any readers with videophiliac leanings, what do you think? Using this test and the following samples as a rough measure, is 25 Mb/s on naturally grainy material "good enough", or do you think the inevitable compromises at single layer bitrates are unacceptable even with a high quality encoder? Would you pay $30 for a BD50 and maxed out bitrates over $20 for a single layer transfer of this quality, or do you think this is the point where it ceased to be an issue? I promise I'm not fishing for compliments - x264 author Laurent Aimar did all the legwork here, not me - but I am legitimately curious what anyone reading this thinks about the above experiment.
No comments:
Post a Comment