Blu-ray, HD-DVD & HD Broadcasts(H.264 & MPEG-2) Screenshots*BIG FILES*

[trbarry]

It occurs to me banding could also be created or increased slightly if an 8 bit conversion is done from YUV to RGB in the process of creating screen caps. For instance, AVISYNTH into VirtualDub could probably have this problem. I don't know how other software handles it.

- Tom

[stumlad]

A few more

[stumlad]

[stumlad]

[stumlad]

To be fair... Dragonheart looks good, but not all of it looks this good... there seems to be some dirt/scratches throughout... Overall I'm happy with it though.

What do you guys want to see next: Casino, Dawn of the Dead (new version), or Bruce Almighty

[bplewis24]

I'd like to see Casino.

Brandon

[Kram Sacul]

There are shots of Casino here.

Stumland, I'm going to see about matching your Dragonheart shots with captures from the dvd Should be quite a difference.

[bplewis24]

Originally Posted by Kram Sacul

There are shots of Casino here.

Umm, I'll go with Bruce Almighty then.

Brandon

[lgans316]

But PQ on Bruce Almighty HD DVD sucked. The BD import released by BVHE looked only slightly better than the HD DVD.

[stumlad]

I'll do 4-5 shots of Bruce Almighty. Definitely a below average title (so is Meet the Parents). I'll prob do Dawn of the Dead next. Are there any other Universal titles ..maybe the Bourne films?

Kram.. yes it would be nice to see the upgrade over the DVD (Dragonheart). If you want, I can edit my post to include your links... or however you want to do it.

[Kram Sacul]

I'm mostly interested in titles where there's no screen captures available anywhere but anything would be great.

I'll see about giving you the links to the Dragonheart captures so you could put them alongside your HD-DVD shots. I also have a non-anamorphic clip from one of the DTS Demo dvds that has some of the worst EE you'll ever see. It's the scene where Draco and Dennis Quaid's character are talking in the field and Draco is flying around. So if you want to post some captures and I'll match them.

Originally Posted by lgans316

But PQ on Bruce Almighty HD DVD sucked.

Exactly, but there should still be a visual record of it somewhere, especially if it has EE or DNR. Then the captures could be linked in the DNR/EE thread forever and even compared against the eventual Blu-ray versions when they come out.

[Vincent Pereira]

Originally Posted by lgans316

I found the PQ of Casino to be inconsistent due to the frequent flashing of dirt / white specks.

Like most Scorsese films, CASINO has a lot of opticals, which tend to have more speckles and dirt built into the image. Personally, I'd rather see some quick, one-frame flashes of speckles and dirt than DVNR artifacts.

Vincent

[Kram Sacul]

I'd also agree. The occasional black or white speckles is far less distracting and damaging than DNR or EE.

[House]

Dark City, 1080i vs BD (theatrical m2ts).

1080i;

BD;


1080i;

BD;


1080i;

BD;


1080i;

BD;


1080i;

BD;

[House]

cont.

1080i;

BD;


One interesting thing I discovered by accident, the theatrical vs director's cut of certain scenes that have been tweaked have quite variable PQ. In the theatrical cut, Jennifer comes out the shower and John speaks first, in fully lit area, then she walks towards him. In the director's cut, she gets startled first, sees him completely in shadow/blacked out, guesses it's him, and again walks towards him to sit down. Funny thing is that the walking towards bit is a completely different cut. Compare this;



With the one above from the theatrical. Apart from the dramatic PQ difference, it's also set up/framed a little differently (and she blinks at that exact moment in time where she doesn't in the other ). I'm not expert on the movie so no idea what the exact changes are to the director's cut, but I imagine those little things will be scattered throughout...

[paku]

Obviously the same old master was used. Sucks really; ten years later they decide to do a new Director's Cut but they don't bother to do a completely new, proper transfer. What a waste of an opportunity.

[msgohan]

Excellent comparison House, just the reason I love this forum.

A few mouseovers (the repositioned BD shots might look darker than they should in IE):
2 - 1080i vs BD repositioned
4 - 1080i vs BD repositioned
5 - 1080i vs BD repositioned

Seems the overall lack of detail is down to the master, but the BD adds color+contrast tweaking, DNR, and EE. Look at the wall screen-left of the guy in 4 (can't recall names at the moment). Surely VC-1 should replicate this better than the MPEG-2, and it does in many places. But there are some tiles clearly intact on the 1080i that are destroyed on the BD.

[Kram Sacul]

I had a feeling they used the same master. That means that no DNR-free master exists then. This is it. *sigh* It does look better than the broadcast version but the better compression just brings out the flaws. What was acceptable a few years ago looks like crap today.

I saw this in the theater back in 1998 and once at a special Q&A screening a few years after with writer/director Alex Proyas and cowriter David Goyer in attendence. I remember the print being dark and slightly grainy with fantastic textures. This isn't it.

[msgohan]

Originally Posted by Kram Sacul

I had a feeling they used the same master. That means that no DNR-free master exists then. This is it. *sigh*

It's tough to tell due to the blocking but it seems like the 1080i might have LESS DNR at least. How bout them wall tiles? Surely the lines weren't removed by the VC-1 encoder.

Also it seems to have less EE judging from the man's outline in 4 and the side of William Hurt's face. Unless the MPEG-2 is somehow softened a bit.

[Kram Sacul]

I would say it's softened. If I remember correctly and House has the same capture the mpeg-2 broadcast is only about 6GB in size. There's some really bad blocking during the end battle.

[dvdmike007]

If its the same master how did they match the new elements to it ?
There are a few diff shots as shown here

http://www.dvdbeaver.com/film2/DVDRe...ty_blu-ray.htm

[Mr. Hanky]

I propose another avenue to take this topic is to profile, say, the top 3 lowest bitrate vc-1 titles against the top 3 highest bitrate vc-1 titles. I believe we have the resources, by now, to query what the top 3 and bottom 3 will yield. Then we reference if screenshots already exist for movie x, and if not, we add them to this topic. (One caveat is if the 3 lowest or highest happen to all be CG flicks. If that is the case, then we stipulate that one live-action title that fits closest to the criteria should substitute one of the 3.) The nominal/avg bitrate for the entire movie should be indicated, as well as the specific bitrate occurring in the selected scene, if available. (Let's try to keep the scene selections to include something happening on screen, not just blank black screens or static credits )

I think this should give a reasonable envelope of performance of how vc-1 is getting utilized in real encodes.

[stumlad]

Originally Posted by Mr. Hanky

I propose another avenue to take this topic is to profile, say, the top 3 lowest bitrate vc-1 titles against the top 3 highest bitrate vc-1 titles. I believe we have the resources, by now, to query what the top 3 and bottom 3 will yield. Then we reference if screenshots already exist for movie x, and if not, we add them to this topic. (One caveat is if the 3 lowest or highest happen to all be CG flicks. If that is the case, then we stipulate that one live-action title that fits closest to the criteria should substitute one of the 3.) The nominal/avg bitrate for the entire movie should be indicated, as well as the specific bitrate occurring in the selected scene, if available. (Let's try to keep the scene selections to include something happening on screen, not just blank black screens or static credits )

I think this should give a reasonable envelope of performance of how vc-1 is getting utilized in real encodes.

Some of the best of the higher bit-rate live action VC-1. Shoot Em Up, The Orphanage, Becoming Jane (never saw this one though)

Best of the lower- King Kong, Troy DC, UK Island (havent seen), Harry Potter

King Kong beats em all

[bplewis24]

Originally Posted by Mr. Hanky

I propose another avenue to take this topic is to profile, say, the top 3 lowest bitrate vc-1 titles against the top 3 highest bitrate vc-1 titles. I believe we have the resources, by now, to query what the top 3 and bottom 3 will yield. Then we reference if screenshots already exist for movie x, and if not, we add them to this topic. (One caveat is if the 3 lowest or highest happen to all be CG flicks. If that is the case, then we stipulate that one live-action title that fits closest to the criteria should substitute one of the 3.) The nominal/avg bitrate for the entire movie should be indicated, as well as the specific bitrate occurring in the selected scene, if available. (Let's try to keep the scene selections to include something happening on screen, not just blank black screens or static credits )

I think this should give a reasonable envelope of performance of how vc-1 is getting utilized in real encodes.

I'd just like to ask what purpose this is to serve? The only conclusion we could possibly draw from the low/high bitrate movie doesn't really apply to the corresponding high/low bitrate movie since it's another film.

If we're strictly looking for compression artifacts I could understand that, though.

Brandon

[stumlad]

Originally Posted by bplewis24

I'd just like to ask what purpose this is to serve? The only conclusion we could possibly draw from the low/high bitrate movie doesn't really apply to the corresponding high/low bitrate movie since it's another film.

If we're strictly looking for compression artifacts I could understand that, though.

Brandon

I agree that that the test doesnt serve much... but when you look at the best looking low bit rate titles and best looking high bit-rate titles for VC-1, the higher bit rate titles dont necessarily look better... they come from different sources though so it's not a great test.

I think Nature's Journey, the Mummies, and a few other titles that have been encoded for both formats have shown that increasing the average bit-rate by 10mbps does extremely little (this does not mean I want them to keep bit-rates low, but it does go down as further evidence to what amir has been saying all along).

[Mr. Hanky]

It is all conjecture, really, until we actually profile the "top 3" and "bottom 3" and work from there. Certainly, there will be examples of well done encodes at even low bitrates, and by all means, they deserve recognition.

To be clear, I did NOT say "best looking low bit rate titles". I said to simply pick the 3 lowest bitrate movies and the 3 highest bitrate movies, sorted purely on native bitrate, regardless of if they are AAA productions or not. Let the pieces fall where they may, and let's see what it yields.

[stumlad]

Originally Posted by Mr. Hanky

It is all conjecture, really, until we actually profile the "top 3" and "bottom 3" and work from there. Certainly, there will be examples of well done encodes at even low bitrates, and by all means, they deserve recognition.

To be clear, I did NOT say "best looking low bit rate titles". I said to simply pick the 3 lowest bitrate movies and the 3 highest bitrate movies, sorted purely on native bitrate, regardless of if they are AAA productions or not. Let the pieces fall where they may, and let's see what it yields.

I chose the arguably best looking high and best look low bit rate titles... and that in itself was still an apple to oranges comparison. If we do a test like you're saying, it would be completely useless.

We already have comparisons with the same movie, same master, etc where the only differences are bit-rates... why would we test anything else? Even those tests aren't fair... Here's why. The Mummy BD is appearing to have the slightest bit of DNR added to it. If it was the lower bit-rate encode that had the DNR, everyone would be saying how the higher bit-rate shows more fine object detail (attributing the smoothing to the lower-bit rate encode). Even a test like that isn't truly apple to apple, but you want to compare 6 completely different titles?

[Mr. Hanky]

No, not comparing title to title. Just bringing 3 on the bottom and 3 on the top into the fray, so we can see. Not everything has to be a rigid comparison of A vs. B. I'm more intending to convey- what are the titles, here they are to see, is there anything left to discuss with what we see? There is plenty to learn from such a query. Not only can we judge how good or bad they look, but also speculate on the presence/absence of dnr/smoothing/various other techniques that may/may not have made the low bitrate encodes possible as they are. It just might be interesting to see how they match up to their broadcast counterparts, where available (there's your comparison ).

Mind you again, I'm saying to pick the 3 lowest bitrate titles and 3 highest bitrate titles, regardless of if they are the best example of their genre. What are they, and how do they look? If they are already present in this topic, then we can back reference to those shots, as needed.

Who knows, maybe the vc-1 encodes will occupy a fairly confined range, and have a similarly consistent level of quality as a result of that confined range. We'll never know, unless a study is directed in this particular manner.

[kschmit2]

A very low bitrate example would probably be Rennaissance HD DVD.

Average bitrate for the video elementary stream is around 10.7 Mbps.

PQ is virtually perfect though.

Review here: http://hddvd.highdefdigest.com/900/r...ce2006_de.html

[Mr. Hanky]

What happen to that topic with the table of movie titles with associated specs, I wonder?... We should be able to pick out examples by codec and video bitrate pretty easily, once we got that. I guess being able to re-sort the data would drop the answers right out, so we'll attack that once/if we get there.


EDIT:
Here's what I come up with using the Bene Table :

(EU) Testosterone___________vbr 10 Mb/s vc-1
(have no idea what this is, so maybe skip it for something more mainstream?)
Mutiny on the Bounty________vbr 12 Mb/s vc-1
(perhaps it is better to skip this in favor of a more contemporary movie to fill the one live action slot?)

(EU) Lady in the Water_______vbr 11 Mb/s vc-1
Ant Bully__________________vbr 13 Mb/s vc-1
Polar Express______________vbr 14 Mb/s vc-1

(there were some exclusions, for obvious reasons where a certain movie title is already known for its "issues" or different format, altogether- namely, Full Metal Jacket and South Park)

Blazing Saddles____________vbr 24 Mb/s vc-1
(EU) Lethal Weapon_________vbr 24 Mb/s vc-1
Reds_____________________vbr 24 Mb/s vc-1

Perhaps, there are other suggestions for substitutions? Feel free to weigh in.

EDIT: There would be an additional perk to comparing Reds (if info is accurate), since there seems to be an mpeg-2 version of it on BR @ 27 Mb/s. (Just a suggestion, but I fully understand if this is something to be avoided, as well)