BD or HD-DVD do you really care?
- Thread starter JoeSp
- Start date
- Latest activity Latest activity:
- Replies 71
- Views 8K
John is there not a point where reduction in the Mbit becomes detremintal to the current codecs being used? It seems to me that the real jump in PQ of DVD happened when that second layer started being used -- not better compression technque.
Then I can't help you. Having seen the results of the same title with an earlier generation and a later generation encoder and identical ABRs; it was no contest in PQ.
Was the 2nd layer helpful? Of course; but it's the encoder that makes the biggest difference.
WRT to HD encodes; in speaking with people that actually write and test the encoders; they are getting amazing results at ~15 Mb/second ABR and peaks around 20 Mb/second. This leaves up to 10 Mb/second for AUDIO and about half of that is needed for a primary language track in TrueHD and secondary languages in DD @ 448 Kbit/second.
Just a year ago they couldn't get the same results at ~20 Mbit/second ABR.
I know and converse with people that write AVC and VC-1 encoders.
Yes; that point is the point where the first bit of lossy compression is used.
Now; if what you mean to ask is this:
"Is there a point where it becomes visibly degraded?" that's a different matter.
Then it becomes a matter of what fashion is the drop in quality; and where does it occur. If you don't have the masters (D1s or D3s) then you don't really know what the original looked like. So all you can do is guess and look for visible artifacts in the image.
The goal for these codecs is to do the job with literally unnoticable defects when viewed at regular speed. No codec will be "perfect" if you freeze each frame.
Is there a point where this occurs yes; but that point has gotten lower in bit rate for MPEG-4/AVC/H.264 and VC-1 in the last year. VC-1 has seen the largest drop in required ABR; with about a 40-50% aggregate encoder gain in the last year. H.264 has only seen about a 20% improvement from last year. That's still quite a sizable improvement.
There is also the case of a pathological signal which can trip up any encoder and it doesn't have to be the same signal fed to all of the encoders. In this particular case; bandwidth out to the limit of the format won't help you.
Cheers,
Stacey Spears made this comment last week about x264 (Open Source implementation of the H.264 codec):
HD 1080p24 Challenge] MPEG2, VC-1 and H264 with real uncompressed source - Page 9 - Doom9's Forum
And although HD/BD compliant AVC encoding remains a bit of a mystery (NDA-ed specs),
x264 is best in class even when compared to commercial implementations.
When using "insane" parameters to extract every last bit of quality possible, x264 is faster than VC-1.
And if you believe doom9 quantitative methods used to compare codecs, x264 is better than VC-1, too.
Diogen.
Diogen:
That's why I appreciate Stacey's friendship -- he's honest even if the answer isn't what people want to hear.
Cheers,
That's why I appreciate Stacey's friendship -- he's honest even if the answer isn't what people want to hear.
Cheers,
I should point out that Stacey isn't a compressionist -- he actually works on the VC-1 encoder. He does however work with some of the best compressionists in the business 
I don't think he'll take lightly that there were better results from the comparison -- and knowing him he's already working on getting VC-1 better
I don't think he'll take lightly that there were better results from the comparison -- and knowing him he's already working on getting VC-1 better
He is the only one MS employee (at least that I know) that the doom9 gang respects and is not looking for political motives in his posts.
Do you also know Don Munsil? I loved reading his posts on AVS but his stint there was short...
Diogen.
I got the impression he was there from the very beginning, the famous first encodings at 7.7 Mbps that beat some of the 3x MPEG-2.
Diogen.
I know he has been working on VC-1 since before HD-DVD launched. I don't know how far back he goes beyond that as he used to work elsewhere at MS.
Cheers,
