Unforgiven 4K UHD (1992)

[Khronikos]

Originally Posted by philochs

Some of the main improvements that Dolby Vision on disk will definitely include over HDR10, that even everyone with a high nit HDR tv will visibly benefit from, are 12-bit color depth and 4:2:2 chroma subsampling.

12-bit color depth means absolutely no color banding, ever. HDR10 is 10-bit, and hence is going to sometimes cause minor banding. The 12-bits encoded in the disk will benefit current panels because 12-bit depth is above the threshold, and prevents any color banding, even when viewing on a 10-bit panel. This feature is enabled on both streaming DV content, and Dolby Vision that is encoded on disk.

Video compressed with chroma subsampling includes brightness information for every single pixel, but not for color information. Color information in 4:2:0 (HDR10), is only shared among every other pixel. Since DV on disk has 4:2:2, it means that Dolby Vision UHD BD content has full color resolution on the vertical lines, while matching the 50% of pixels that contain color information on the horizontal lines. The effect will be a noticeable increase in real color saturation within the fine details, but it will be most visible near the edges of sharp color transitions. 4:4:4 is the next step up from 4:2:2 DV, and it's uncompressed color, and could only be included in a future disk format. Note: streaming DV is only 4:2:0 still, so no one has got to see Dolby Vision like this before.

I am certain that 4:2:2 along with 12-bit color depth are going to mean a noticeable upgrade over HDR10 encodes. People who question this will be disproved in a few weeks when DV on disk launches.

If anything, 4:2:2 would have taken UHD Blu-ray version on Unforgiven to the next level. Let's hope they do release it again with Dolby Vision within the next 5 years. It's a deserving film.

It must be noted that there is also more room for abuse here. If people are using the same 66 GB discs with even more data to store that is NEVER a good thing for the bitrate.

With DV they NEED to use 100GB discs. There is no telling how much more data you have to store to produce good results. Maybe someone can do the math.... but we already have discs crammed on 66's with artifacting. Cramming even more data into that small of a disc is not going to help the transfer from every angle and in some ways could make it look worse.

In general I agree with you, but are you trying to compare streaming bitrates to full encodes on UHD? Seems like streaming is going to even have more issues with this. You don't seem to be too concerned about the bitrate, the things that actually matters when carrying all this data in the end.

[philochs]

Originally Posted by Khronikos

It must be noted that there is also more room for abuse here. If people are using the same 66 GB discs with even more data to store that is NEVER a good thing for the bitrate.

With DV they NEED to use 100GB discs. There is no telling how much more data you have to store to produce good results. Maybe someone can do the math.... but we already have discs crammed on 66's with artifacting. Cramming even more data into that small of a disc is not going to help the transfer from every angle and in some ways could make it look worse.

In general I agree with you, but are you trying to compare streaming bitrates to full encodes on UHD? Seems like streaming is going to even have more issues with this. You don't seem to be too concerned about the bitrate, the things that actually matters when carrying all this data in the end.

Nope, I wasn't ever comparing streaming bitrates to full encodes on UHD, and I certainly never gave the impression that I don't care about bitrates. Not sure where you got such an idea, you're talking to the guy who suggested an eventual 8K disk format with 300Mbps encodes. I'm all for 100GB disks being used on Dolby Vision titles. Heck, I wish they could upgrade the codec to AV1 eventually, as it's opensource and 25% more efficient than HEVC. But since it won't be ready in time, AV1 will be a streaming only codec. Since it is 25% more efficient, it will certainly help to blur the lines between streaming encodes and disk formats, a little more. Obviously, 100Mbps of HEVC is still going to be superior to a 15-20Mbps AV1 stream. The only reason I was even bringing up Dolby Vision streaming was to point out that it doesn't have 4:2:2 chroma subsampling like Dolby Vision on disk has.

[philochs]

Originally Posted by d514

I don't think a 12bit signal on a 10bit display is better that a 10 bit signal in general. The important question is: How was the 10bit signal on the usual HDR10 disc created? If it was created from a 12bit signal the HDR10 signal can have all the benefits a DV 12bit signal has - on a 10bit display. I highly doubt the downsampling/dithering on a consumer TV does a better job than the equipment in a professional studio.

I'm not sure why you believe what you do, but for the record, 12-bit encoded video content is always 100% free of color banding, even when watching on a 10-bit panel. So yes, a 12-bit signal is a lot better than a 10-bit signal in general. All source material, all DI workflows used to produce UHD Blu-ray are either 12-bit or 16-bit color depth sources and are innately free of any color banding. If you dither them to 8-bit for Blu-ray it can introduce major color banding. If you dither to 10-bit for HDR-10, it can introduce minor color banding. if you keep 12-bit material at 12-bit or if you dither a 16-bit source only to 12bit, then the content itself no longer has color banding. In the case of 8-bit or 10-bit, color striping is inherent to the disk encode itself, nothing to do with a tv's processor.

Dude, when you watch Dolby Vision material on a Dolby Vision capable tv, your tv isn't dithering anything at all. You're simply watching an unadulterated 12-bit signal via a 10-bit panel. Your tv is doing 12-bit signal processing. It's a way of preventing color banding, period. Once 12-bit panels with 12-bit signal processing become standard, then 12-bit encodes will be even better, as you should then be able to see far more shades of colors as well.

[mzupeman]

Eh, I've heard that forcing 12 bit on a 10 bit display actually can get some banding to happen. You have any links that say otherwise?

[philochs]

Originally Posted by mzupeman

Eh, I've heard that forcing 12 bit on a 10 bit display actually can get some banding to happen. You have any links that say otherwise?

Wherever you heard that, it's definitely wrong. Yep, I do have a link which says otherwise. Page 9 of the Dolby Vision white paper includes that info, for instance...


https://www.dolby.com/us/en/technolo...hite-paper.pdf

[Sky_Captain]

It's not wrong, why else would Panasonic have bothered with that recent firmware update for their players, allowing a 10bit output?

[philochs]

Originally Posted by mzupeman

Eh, I've heard that forcing 12 bit on a 10 bit display actually can get some banding to happen. You have any links that say otherwise?

Think of Sony tv's 14-bit signal processing effect, 'smooth gradation'. You turn it on, it upconverts 8-bit and 10-bit material to 14-bit then displays that on a 10-bit panel specifically to break up any color striping. Some people leave that effect off always, as it could potentially mask some fine details, I suppose. However, DV content is 12-bit native so there is automatically not any color banding on the 12-bit encode content for your tv to have to worry about. The only way a tv can display color banding is if it's inherent to the source material you're watching.

[Sky_Captain]

Originally Posted by philochs

The only way a tv can display color banding is if it's inherent to the source material you're watching.

Nonsense.

[philochs]

Originally Posted by Sky_Captain

It's not wrong, why else would Panasonic have bothered with that recent firmware update for their players, allowing a 10bit output?

Yes, it's wrong to say that 12-bit encoded content could cause extra color banding on a 10-bit display, because there is a big difference between a UHD Blu-ray player that forces 12-bit output, and content that is actually encoded in 12-bit. The Panasonic players were previously forcing native 10-bit HDR10 disks to undergo some unnecessary extra processing, internally upconverting the color bit-depth, and then outputting in 12-bit. I can see why that would not be ideal, I read there were some issues with certain models of tv sets. Most purists would choose to leave extra processing like that off. When you're dealing with actual 12-bit Dolby Vision content however, you actually do want it output at 12-bit, as the DV capable tv/player/disk are all designed to display DV at 12-bit, and this prevents color banding, even on a 10-bit panel. Those DV capable tvs and players still output HDR10 only content at the proper 10-bit output.

When you're dealing with HDR10 10-bit content though, you want your UHD Blu-ray player at 10-bit output, because that's what it was designed for. Panasonic tried to get fancy and it didn't work right, so they released a fw update to fix 10-bit output for 10-bit encoded disks, simple as that. It really had nothing at all to do with 12-bit Dolby Vision.

[smackrabbit]

Originally Posted by philochs

Yes, it's wrong to say that 12-bit encoded content could cause extra color banding on a 10-bit display

No, you should never assume that the display in question is going to be able to handle the larger bit-depth correctly. In theory, yes, it should, but that also means that the TV has enough bit-depth in the CMS to properly handle 12-bit data, does YCbCr to RGB correctly at 12-bits, and doesn't introduce artifacts anywhere else in the processing.

There are lots of things that TVs should do correctly but do not, so assuming they'll automatically handle additional bits correctly (for all we know, they'll just truncate the extra bits and process it as 10-bits for a 10-bit panel) is something you can't do.

[philochs]

Originally Posted by smackrabbit

No, you should never assume that the display in question is going to be able to handle the larger bit-depth correctly. In theory, yes, it should, but that also means that the TV has enough bit-depth in the CMS to properly handle 12-bit data, does YCbCr to RGB correctly at 12-bits, and doesn't introduce artifacts anywhere else in the processing.

There are lots of things that TVs should do correctly but do not, so assuming they'll automatically handle additional bits correctly (for all we know, they'll just truncate the extra bits and process it as 10-bits for a 10-bit panel) is something you can't do.

Actually, Dolby Vision capable tv sets with 10-bit panels have no issues rendering 12-bit encoded Dolby Vision content, and they never truncate the extra bits. In fact, that's one of the key things that makes them Dolby Vision capable tvs. If your 'display in question' is not a Dolby Vision set, then no, it won't display anything in 12-bit properly, but you'll just be stuck with the base layer HDR10 anyway.

Streaming and forthcoming disk encoded Dolby Vision content does not ever have color banding issues, period. HDR10 content can have visible color banding, whether on disk or streaming, as it's been dithered to 10-bit color depth from the uncompressed source. I have no idea what is so confusing to you guys about this. I've made it simple as humanly possible, but the mis-info keeps coming back up.

[Geoff D]

Originally Posted by philochs

I'm not sure why you believe what you do, but for the record, 12-bit encoded video content is always 100% free of color banding, even when watching on a 10-bit panel. So yes, a 12-bit signal is a lot better than a 10-bit signal in general. All source material, all DI workflows used to produce UHD Blu-ray are either 12-bit or 16-bit color depth sources and are innately free of any color banding. If you dither them to 8-bit for Blu-ray it can introduce major color banding. If you dither to 10-bit for HDR-10, it can introduce minor color banding. if you keep 12-bit material at 12-bit or if you dither a 16-bit source only to 12bit, then the content itself no longer has color banding. In the case of 8-bit or 10-bit, color striping is inherent to the disk encode itself, nothing to do with a tv's processor.

Dude, when you watch Dolby Vision material on a Dolby Vision capable tv, your tv isn't dithering anything at all. You're simply watching an unadulterated 12-bit signal via a 10-bit panel. Your tv is doing 12-bit signal processing. It's a way of preventing color banding, period. Once 12-bit panels with 12-bit signal processing become standard, then 12-bit encodes will be even better, as you should then be able to see far more shades of colors as well.

That's a negatory on "all source material" being 12 or 16-bit. The latter is a rarity and although 12-bit is more common nowadays a LOT of DIs are comprised of 10-bit DPX files, bearing in mind that the process is a good 15 years old already. And 10-bit is still very much in use today, I read the AC article on a very recent movie (can't remember which, if I find it I'll edit this post) and it specifically mentioned 10-bit mastering.

BUT at least the 12-bit DV encode of that content will basically be a 'supersampled' version and should keep banding at bay, as the same is true of the 12-bit XYZ gamut that such 10-bit P3 content is mapped to when being prepared for a theatrical DCDM.

[kobeson]

Unforgiven...........

[reanimator]

Originally Posted by kobeson

Unforgiven...........

You're in the wrong thread for that.

[AlexIlDottore]

I'm learning a lot here, so keep up the tech talk, please

[StingingVelvet]

Watched this in full today. My thoughts more or less match up with my brief scan on release day. It is definitely a bit dimmer, but it's not that huge a difference. The upgrade is subtle, but noticeable. One of the most obvious upgrade moments is the opening and ending cabin at sunset, which looks much more crisp, and the sunset looks much more refined and real, on the UHD. Overall it's a much more refined image with more natural colors and crisper images, but it's not a night and day thing. I think it's probably roughly similar to Goodfellas, so you should know how you felt about that.

One negative though is it does seem mildly sharpened, or at least there is some mild EE style effects here and there. Nowhere near as sharpened as the original blu, but something's there. There are a few errors with it too, not just the windmill but there's one earlier when Munny is herding hogs and the kid shows up. There's some blatant EE issue on a post behind Eastwood. Still, in motion at 24fps I don't think it's that common or noticeable, but it is unfortunate. I also had a few moments where the image got noticeably dimmer all of a sudden in the middle of a shot, but that could be my smart LED being weird or some other equipment issue. Don't want to say that's a disc problem unless others notice it.

Anyway, if there were a $10 BD edition out there I think a lot of people might have been better off with it, but there isn't. Either way though, if you love the look of film and subtle upgrades in that area I think this is very pleasing to the eye, if a little dim here and there.

[zmarty]

Originally Posted by philochs

Wherever you heard that, it's definitely wrong. Yep, I do have a link which says otherwise. Page 9 of the Dolby Vision white paper includes that info, for instance...


https://www.dolby.com/us/en/technolo...hite-paper.pdf

Maybe I am reading this wrong, but the link you provided contradicts your point. They are showing there that a 10 bit display does show banding? The point of that graph is to show that 12 bit is much better than 10 bit.

[mzupeman]

Originally Posted by zmarty

Maybe I am reading this wrong, but the link you provided contradicts your point. They are showing there that a 10 bit display does show banding? The point of that graph is to show that 12 bit is much better than 10 bit.

Yeah. Nowhere on that page does it say what he claims it says. And with that, I'll just ignore what he had to say from now on.

[Khronikos]

Originally Posted by Khronikos

It must be noted that there is also more room for abuse here. If people are using the same 66 GB discs with even more data to store that is NEVER a good thing for the bitrate.

With DV they NEED to use 100GB discs. There is no telling how much more data you have to store to produce good results. Maybe someone can do the math.... but we already have discs crammed on 66's with artifacting. Cramming even more data into that small of a disc is not going to help the transfer from every angle and in some ways could make it look worse.

In general I agree with you, but are you trying to compare streaming bitrates to full encodes on UHD? Seems like streaming is going to even have more issues with this. You don't seem to be too concerned about the bitrate, the things that actually matters when carrying all this data in the end.

"I am certain that 4:2:2 along with 12-bit color depth are going to mean a noticeable upgrade over HDR10 encodes. People who question this will be disproved in a few weeks when DV on disk launches."

In particular, I think you need to push on the brakes a bit bud. You are talking about pushing an extra amount of data that is NOT insignificant into the same space as the more compressed versions. This is not something that just magically looks better. Many people may not even be able to tell the difference outside of some minor banding improvements.

Yes, banding may be improved, but we, logically and empirically, have no clue what the actual transfers will look like in truth. We simply have never seen this amount of data crammed into a 66GB or 100GB disc yet. And let's not pretend this is 4:4:4 now. We are nowhere near that level of quality that people would actually notice en masse IMO.

I'm not saying you are pushing an agenda or anything, I'm saying it is definitely best to wait and see. 66GB is too small for a lot of 4K films IMO. And god knows what kind of filtering they are using behind the scenes anyway. We don't know a whole lot about how 4K discs are authored to be talking with authority. Some discs look better than others. Some discs look terrible. No one is sure quite why at this point.

And now we are going to suddenly be pushing 4:2:2 and 12-bit into the encodes in the same space? I can't see this occurring without major problems for some films with 66GB discs for sure. We'll see what they do.

[Geoff D]

Originally Posted by Geoff D

That's a negatory on "all source material" being 12 or 16-bit. The latter is a rarity and although 12-bit is more common nowadays a LOT of DIs are comprised of 10-bit DPX files, bearing in mind that the process is a good 15 years old already. And 10-bit is still very much in use today, I read the AC article on a very recent movie (can't remember which, if I find it I'll edit this post) and it specifically mentioned 10-bit mastering.

BUT at least the 12-bit DV encode of that content will basically be a 'supersampled' version and should keep banding at bay, as the same is true of the 12-bit XYZ gamut that such 10-bit P3 content is mapped to when being prepared for a theatrical DCDM.

To save me editing here's the relevant quote from AC June '15 p.65 about Avengers Age of Ultron:

DI producer Michael Dillon relates that the Technicolor Los Angeles crew deBayered to 3414x2196 DPX 10-bit files using Colorfront Transkoder and delivered the plates to the visual-effects vendors, who scaled and center-extracted to a DI working resolution of 2158x1214 (1.77:1).