Originally Posted by Supermans
Although I'd still like my original question in post 2345 to be answered. My new question is about VC-1 in general. Batman Begins for example was an early VC-1 encode made to fit onto a 30GB disc. If you were able to do the encoding for that movie over again to be placed on an HD-DVD51 for some special edition. Would the movie be able to achieve tier 0 quality using VC-1 because of a higher bitrate and newer methods or encoder versions? I know you guys have gone thru enhancements as you have mentioned and have improved upon the VC-1 codec since then. My other question would be if you had to do another re-encode of Batman Begins for a 30GB HD-DVD today. Would current technology be good enough using today's VC-1 encoder to make the film look better and be closer to tier 0 than it was? You may be thinking "Where am I going with this?" or "Why do I want to know this information?" It is simply because I still trust in you and believe when you say VC-1 is hands down better than all the rest you have seen when compared to other codecs when being compared to the original masters. Since I do believe you, I want to know how much better is Vc-1 on 50 or 51 GB's going to look in the future when compared to the same films done to fit on 30GB discs.
First thanks for the trust in our abilities wrt to VC-1 and kind hearted post
. Was nice to see in the sea of push backs otherwise. Second, I am glad you asked me about this movie because I know it well. I had watched it in the theater, and for once, I was most impressed with a sequel to a comic book -- unexpected story line, and great production overall. I had really enjoyed watching it on the big screen. Needless to say, when I heard it was coming to HD DVD, I was quite anxious to make sure it looked really good. I knew this would be a landmark title and we had to get it right, even beyond my own desires.
So I went to the guy who runs the codec team for me and put him on notice that this movie better look good and we should pull out all stops to achieve the absolute best quality possible even if it meant hand encoding every frame manually. To my surprise, he said the movie was already encoded and that the data rate was only 10 mbit/sec! I thought I heard him wrong. So I asked again and he repeated once more that it was just 10 mbit/sec. With a bit of trepidation, I asked how well it looked. He again surprised me by saying it looked great and that was the view of the compressionist who had encoded it also. OK, so I knew our codec had great performance but it was still hard to believe that we would be happy with the quality at that rate.
So I asked to see the check disc (preliminary production discs). The first scene I was looking for was the glacier pan. Not that it was hard to compress, as the pan is slow and VC-1 has no trouble tracking such large motion with very high efficiency (we have quarter pixel motion estimation which gives us much better ability to do this than the half pixel system used in MPEG-2). I wanted to see it because it was a grand scene in the theater and I was most impressed when that bright scene came up with all of its grandeur.
My suspicious were confirmed and the scene had no compression artifacts but look somewhat soft - no doubt that is what keeps it from tier 0 in people's books. While I was a bit disappointed in that, I could tell the softness was from the master as per my previous statement, it was not a hard scene to encode. So I had the team probe the post house and they confirmed that scene was just as soft in the original.
So I continued to watch the movie and I become progressively more impressed. My jaw dropped during some of the scenes where you could see detail that simply was not visible in SD, such as the dirt on the little soldier in the girl's hand. But the best was yet to to come. I am talking about the scene where the camera pulls away from the train and you see the cityscape. Man, that was so sharp I thought I would cut my hand if I touched the screen
I was really worried about the above segment and expecting to see artifacts. Reasons were many but most had to do with the fact that it was computer generated with no regards for who Nyquist was
. What I mean is that when you shoot a natural scene, the capturing equipment always rolls off the frequency response. This means that no edge is ever 100% sharp. In contrast, in digital domain, one can have a black pixel next to a white pixel which represents infinite frequency response (i.e. breaks the Nyquist theory of filtering things beyond the limits of the system).
You see the same impossibility on your computer screen all the time where a black pixel can be next to a white pixel and is represented perfectly on a digital display such as an LCD (the same on a CRT monitor would have less clarity, with the white a bit less white, and the black a little less black and possibly ringing from the video circuits being upset with an above spec signal bandwidth). Such was also the case with the city scene, with so many lines on the buildings, sharply defined, and moving in different directions. Yet the codec preserved that detail without seemingly even trying. It was job well done and no better proof of how well an advanced codec can deal with such a movie, yet hit a data rate that is the maximum of red laser DVD!
The final encode wound up being 12 mbit/sec btw as reported by cjplay here about a year ago (hence the reason I took liberty of mentioning the data rate here against our typical policy). They probably tuned it a bit more or simply bumped it up for good measure. But I didn't complain about that. After all, even at 12 mbit/sec, we are talking about the video only using 12 gigabytes of space! Put another way, you could put the movie and a lossy+TrueHD track and it would still fit on HD DVD-15!
So do I think we were space constrained in this movie? Not at all. HD DVD-30 was plenty to hold all the extras, PiP, etc. Where we data rate constrained? No. Other titles had come out with higher data rates still. There simply was no need to use higher rate for this title with VC-1.
What can we do today with the latest tools? For one, we could hit the same quality with much less work. Second, we could probably improve the micro-quality some. This is a term I just made up (
), to indicate the quality of individual frames improving even if a human can not perceive an improvement. Although you could argue that those of us who live and breath with this stuff, could see some marginal improvement, I doubt that even the best videophiles would be able to see any improvements. As such, I don't think there is any merit in re-encoding the movie. Rather, I like to see them re-shoot/re-master the glacier scene. That would make a much bigger difference than any re-encode.
Regardless of who wins the format war, your division will be working with VC-1 and 50-51GB's of space sooner rather than later so your input on this could bring some excitement back into the crowd.. And it really doesn't matter if we are on Blu-Ray's or HD-DVD side since whomever wins the format war will still benefit from VC-1 when it is used in the future by the victorious format..
The usage for TL-51 is really for people who think it is best to load up one disc, as opposed to use two, even if the latter is cheaper to make or has more consumer value (and I am talking about both formats here). And it is probably a bigger marketing tool than technical. This is why you don't see me talking about it all, unless asked. Current HD DVD is a well designed system. Its capacity is well matched with the advanced codecs being used on it. So we don't see much real need for extra capacity although we appreciate Toshiba, Memory-Tec, etc. continuing the pace of innovations.
On VC-1 being used equally in both formats, I am afraid that is not an outcome that I see, given where I sit. Most of the reasons are simple. The studios on HD DVD side have no technology arm which make encoders so they are much less biased by NIH (not invented here) syndrome. And of course, they are not annoyed at all that we choose to support HD DVD
, or hold a larger IP position in another codec. You see that the reality bears this out given the huge volume of VC-1 titles going through studios who publish in HD DVD, as opposed to other studios. BD companies are giving us another excellent reason to stay firmly with HD DVD on this front in their avoidance of VC-1 (not that more was needed, but here we are anyway).
I am waiting to catch my flight to Tokyo soon. Time permitting, I will expand more on this topic and history of VC-1/AVC and questions raised on my post last night.
Ja! (see you later in Japanese