Starship Troopers 4K UHD (1997) (20th Anniversary Edition)

[TbeRw01]

Originally Posted by andreasy969

maybe TbeRw01 can also say if it's per frame or scene whatsoever)

DV metadata can be frame by frame but of all the metadata I checked (a lot), it's always shot by shot metadata and on average 2000-4000 shots depending on the movie length (and style).

[andreasy969]

Originally Posted by lgans316

Fair point but once again, I would still go by how it looks on the display equipment.

Fair enough, and I guess we can all agree that DV appears to do proper/better tone mapping, but the opacity of this stuff only got bigger and with DV you never really know what you're actually looking at. Even more so than with HDR10 IMO. And I don't like.

[MisterXDTV]

Originally Posted by KMFDMvsEnya

Does the DV layer include 12bits of amazah colours?
Or is it just static metadata that happens to tame the Sony Light Cannon Beast?

Perhaps the 'improved colors' stems from decreasing the excessively elevated APL which frankly makes the image look rather washed out to my eyes.

It seems as though 12bit DV layers can cause discolorations macroblocking in the base HDR10 layer.

MEL DV like this one doesn't have 12 bit color...

It's a 10-bit encode

[MisterXDTV]

Originally Posted by TbeRw01

well the DV metadatas are inaccurate. This is not how the average DV plot should looks like. To me, it looks like the studios went cheap and decided to take shortcuts and set a static L1/L2 instead of grading DV shot by shot properly.

When they grade DV properly, they run the Dolby algorithm to generate L1 metadata which measure and maps all the pixels of the actual HDR grade. It creates a min_pq, max_pq and avg_pq for EACH SHOT. L1 should never lie to the TV like it does in this BD. The brightness adaptation to different displays is done in the L2 trim passes, not L1 which is what the graph plot is.

Then they do the time-consuming part of the DV creation: the L2 trim passes where they have to go shot by shot again and adjust everything for different brightness outputs (usually 100/600/1000nits trims). This is the metadata your TV use to adapt the content to its internal capabilities.

In other words, using static L1/L2 metadata makes no sense to me and the only reason I see is to save time a money.

I know they shouldn't be static in DV... but still there's a 2000 nit difference in MaxFALL between HDR10 and Dolby Vision metadata

A massive difference!

I don't understand such a choice. They could have "copied" the metadata for the HDR10 main track

[andreasy969]

Originally Posted by MisterXDTV

I know they shouldn't be static in DV... but still there's a 2000 nit difference in MaxFALL between HDR10 and Dolby Vision metadata

A massive difference!

I don't understand such a choice. They could have "copied" the metadata for the HDR10 main track

IMO the "problem" is not so much the difference with the meta data, but the DV meta data being simply wrong. If (!) that's done on purpose, the sole reason I can imagine why one would do that, is because one knows that this particular title will provide the best results with DV that way today. But what about tomorrow? Hence me being confused. IMO it's basically the same (but not the same league) as stating Rec.709 when it's 2020. Which should cause trouble instead of providing the better results.

The "dynamic" DV meta data of this title actually does the exact opposite of what MEL DV is supposed to do.

[MisterXDTV]

Originally Posted by andreasy969

IMO the "problem" is not so much the difference with the meta data, but the DV meta data being simply wrong. If (!) that's done on purpose, the sole reason I can imagine why one would do that, is because one knows that this particular title will provide the best results with DV that way today. But what about tomorrow? Hence me being confused. IMO it's basically the same (but not the same league) as stating Rec.709 when it's 2020. Which should cause trouble instead of providing the better results.

I don't think that matters a lot: I doubt we will see 4000 nits TVs anytime soon, especially considering the power required to reach that mark...

[andreasy969]

Originally Posted by MisterXDTV

I don't think that matters a lot: I doubt we will see 4000 nits TVs anytime soon, especially considering the power required to reach that mark...

Likely, yes, but we do have 1000 nits already. And I'd expect subpar results from the DV engine when being fed a 650 nits max info instead of 1000 when the actual content is 4000.

If (!) on purpose, I'd call it a total hack.

[TbeRw01]

Originally Posted by MisterXDTV

I know they shouldn't be static in DV... but still
I don't understand such a choice.

I dont understand as well...

More info about the DV RPU in this release:
There are 1964 shots with identical L1 metadata and it only has a 100nits trim pass (no 600 nor 1000nits trim pass)

see: https://justpaste.it/8l4oy
madVR brightness measurements OLD vs NEW BD: https://screenshotcomparison.com/comparison/29342

Summary:
  Frames: 186264
  Profile: 7 (MEL)
  DM version: 1 (CM v2.9)
  Scene/shot count: 1964

[Rollo Tomassi]

Originally Posted by lgans316

The differences doesn't look this drastic on my TVs but I did notice the colours looking a tad better and highlights a tad less blown out on the DV stream.

Which is all I wished for with this new release! I guess, Champagne would be in order!
Cheers!

[Brian81]

So the compression is better (I see now that I'm looking on a laptop screen and not zooming on a phone). So what is this about the DV layer? Is it wrong?

[KMFDMvsEnya]

Originally Posted by Brian81

So the compression is better (I see now that I'm looking on a laptop screen and not zooming on a phone). So what is this about the DV layer? Is it wrong?

Seems like what is being observed is that the DV layer is static metadata which is lower than the HDR10 base layer that kind of tames the nuked levels on DV enabled displays but SOL for non-DV displays.

An inelegant Band-Aid solution for something that should have been readdressed in the HDR mastering for this new release. Whether it was intentional or due to inattentiveness we can only speculate at this point.

[Warm Gun]

Originally Posted by sleeman

After watching the new disc it's very harsh to watch the old one at 4000 nits brightness.

For a minute there, I thought your OLED could do 4000 nits and was like...



"I didn't know that men could build such things."

[sleeman]

Originally Posted by Warm Gun

For a minute there, I thought your OLED could do 4000 nits and was like...

Yeah, my C9 can't go that high, however, I still find the old disc excruciatingly bright.

[Vriess]

Sounds like Sony really did this one dirty if theres no dynamic tonemapping. Is this a trend with Sony or only this release?

Still, the DV looks soooooo much better than the HDR 10 on this. It really looks quite good.

[Geoff D]

Originally Posted by andreasy969

Why do you expect that lower (wrong) meta data values will resolve more highlight detail? Which was your issue with this disc HDR wise IIRC. I'd expect the opposite if I feed too low values, i.e. it clipping too soon. Unless maybe such basically false flags result in the DV doing something that's neither here nor there. But you wouldn't want that would you?

Just trying to understand the logic.

EDIT: Other than me trying to make sense out of everything, I'm also looking forward to hear what you'll have to say about the practical results.

I'm hoping that it'll reduce everything on-screen to the 650 nit mark, that's not borne from logic but a misunderstanding of what's happening here! We've yet to hear from anyone saying that the DV presentation is blown out AF though, same with As Good As It Gets as that's virtually static in the DV layer according to matey's special graphs and no-one's reported that looking nuked either despite its DV metadata being capped at 459 nits while the content regularly exceeds 1000 nits.

I think my ST steely is with the local courier so it may even show up tomorrow.

Originally Posted by TbeRw01

DV metadata can be frame by frame but of all the metadata I checked (a lot), it's always shot by shot metadata and on average 2000-4000 shots depending on the movie length (and style).

Yeah, people talk up the frame by frame aspect but it's not especially necessary, I think even Dobly's own documentation says to set it shot by shot and finesse it from there.

Originally Posted by andreasy969

IMO the "problem" is not so much the difference with the meta data, but the DV meta data being simply wrong. If (!) that's done on purpose, the sole reason I can imagine why one would do that, is because one knows that this particular title will provide the best results with DV that way today. But what about tomorrow? Hence me being confused. IMO it's basically the same (but not the same league) as stating Rec.709 when it's 2020. Which should cause trouble instead of providing the better results.

The "dynamic" DV meta data of this title actually does the exact opposite of what MEL DV is supposed to do.

As far as tomorrow is concerned, if the magical unicorn 4000-nit TV were to appear and ST's HDR technically didn't need any tone mapping then one could just turn the DV off. This is a MEL release (as are all Sony DV discs) so there's no 12-bit extension data that's being lost, the sole benefit of DV in this case is the assistance with tone mapping so if a TV could do it natively then DV would not be required.

In your subsequent reply to MisterXDTV you're absolutely logically correct (as always) that it's not all or nothing, there are 1000-nit displays (and even 1800-nit ones like I has right here) that could take advantage of a properly audited suite of dynamic metadata for a Light Cannon™️ grade such as this, but as internal tone mapping continues to improve with each product generation then that could also make this fake DV layer surplus to requirements. Me though, I'd prolly still keep it at 650 nits because I really don't like Sony's grades all that much. Sometimes it works for the content, don't get me wrongo, but in the case of ST it's kinda hideous IMO.

[andreasy969]

Originally Posted by Geoff D

[Show spoiler]I'm hoping that it'll reduce everything on-screen to the 650 nit mark, that's not borne from logic but a misunderstanding of what's happening here! We've yet to hear from anyone saying that the DV presentation is blown out AF though, same with As Good As It Gets as that's virtually static in the DV layer according to matey's special graphs and no-one's reported that looking nuked either despite its DV metadata being capped at 459 nits while the content regularly exceeds 1000 nits.

I think my ST steely is with the local courier so it may even show up tomorrow.



Yeah, people talk up the frame by frame aspect but it's not especially necessary, I think even Dobly's own documentation says to set it shot by shot and finesse it from there.



As far as tomorrow is concerned, if the magical unicorn 4000-nit TV were to appear and ST's HDR technically didn't need any tone mapping then one could just turn the DV off. This is a MEL release (as are all Sony DV discs) so there's no 12-bit extension data that's being lost, the sole benefit of DV in this case is the assistance with tone mapping so if a TV could do it natively then DV would not be required.

In your subsequent reply to MisterXDTV you're absolutely logically correct (as always) that it's not all or nothing, there are 1000-nit displays (and even 1800-nit ones like I has right here) that could take advantage of a properly audited suite of dynamic metadata for a Light Cannon™️ grade such as this, but as internal tone mapping continues to improve with each product generation then that could also make this fake DV layer surplus to requirements. Me though, I'd prolly still keep it at 650 nits because I really don't like Sony's grades all that much. Sometimes it works for the content, don't get me wrongo, but in the case of ST it's kinda hideous IMO.

The thing is that I actually fail to see how feeding a way too low max cannot confuse and will only result in subpar results with every current display. Because we must not confuse these values with what the diplay can do, i.e. it's not target display nits. EDIT: And if it were, it'd still make no sense.

So I really don't get it. Unless it's not being used in the first place and the DV just does better dynamic tone mapping. That would also explain why they aren't just setting the static HDR10 meta data accordingly.

re. As Good As It Gets: Didn’t take a look yet, i.e. don't know if it's a light cannon anyway.

[Rollo Tomassi]

Judging by the caps shared on this thread and the ones at Caps-a-holic,looking at grain refinement the encode is better, highlights are slightly less overblown. So overall, this should be an upgrade IMO.
But: why am I still seeing (slight) edge enhancement on this cap. Sure, it's a lot more subdued on the 25th (Edit: not 30th, duh! ) anniversary 4K than the halos on the blu and I probably won't even see it from the viewing distance to my telly, but still...

https://caps-a-holic.com/c.php?a=3&x...&l=1&i=20&go=1

[Ben_UK]

Can someone decipher the DV chats into plain english? From what I can deduce the DV HDR is static, is that correct? So its garbage?

[spiltmilk]

Originally Posted by Ben_UK

Can someone decipher the DV chats into plain english? From what I can deduce the DV HDR is static, is that correct? So its garbage?

That seems to be the case. Not necessarily garbage in practice though, since it most likely tames some of the Lightcannon™ grading of the previous release.

I have ordered it even though I find the steelbook hideous.

[Russhole]

Happy 25th Anniversary!