Jurassic Park Collection 4K UHD (1993-2015)

[movielover713]

So is the best buy steelbook stacked discs?

[Geoff D]

Originally Posted by pawel86ck

Your ZD9 is very cappable HDTV, it may not crush details in 1000 nits HDR mastered content (because it simply has enough nits) but there are also some UHD movies mastered at 4000 nits. I dont know if SONY ZD9 has similar feature like active HDR (feature that make dynamic metadata on the fly and because of that is able to adjust picture brightness to your particular display), because if jurassic park movies are mastered at 4000 nits, feature like that would help to resolve more shadow details.

1) The JP movies are all mastered to 1000 max/0.005 min.

2) Did you see my post above about the crush in Guardians of the Galaxy 2? I'm not imagining it there and I'm not imagining it on the JPs...

3) While some TV's may indeed appear to crush down shadow detail in HDR, bearing in mind how damaging the ABL can be to very low brightness scenes on an OLED (I suggest a quick look in the Unforgiven thread)...I don't have an LG OLED TV.

4) With the 0-10,000-nit ramp on Sony UHD discs I can see clear down to 0.001 nits.

[edit] You've clearly made your mind up about this issue but my TV isn't crushing anything in HDR, you can take that to the bank. FYI I've measured just under 2000 nits peak from my ZD9 and have two picture modes for HDR viewing, one for 1000-nit grades and one for 4000-nit grades to preserve the relevant amount of highlight information whilst lowering the brightness a tad.

[mzupeman]

Originally Posted by movielover713

So is the best buy steelbook stacked discs?

Don’t know yet.

[SynViks]

Originally Posted by Geoff D

And yet the grain in Nolan's UHD movies - from IP's all - doesn't look anything like as well resolved as that DOTD grab does, and while you're right about the anamorphosis squashing down the grain on Mr N's shows it doesn't make it invisible or conspicuously crawly all of a sudden

So here are a few factors at work when it comes to Nolan's transfers.

1: He likes his masters to look as close to a print as possible, and he has said he does this via A/Bing with a film element. This explains why he likes his raised blacks, goes light on the HDR, and doesn't rely on digital grading to get the look.

2: His films are shot on modern film stocks, which have a very fine grain (imagine it being finer than the DOTD still, which has very fine grain). He also shoots in anamorphic which is going to make the grain even finer vs something like Jurassic Park. We know he doesn't remove the grain since he doesn't grade and since it has a squashed shape.

I'd conjecture that because the grain was so fine, the encoder ate some of it, resulting in less visible grain and those crawling patterns you're reporting (I've not seen the UHDs in motion, and the screencaps are gone, so I can't comment more).

[pawel86ck]

Originally Posted by Geoff D

1) The JP movies are all mastered to 1000 max/0.005 min.

2) Did you see my post above about the crush in Guardians of the Galaxy 2? I'm not imagining it there and I'm not imagining it on the JPs...

3) While some TV's may indeed appear to crush down shadow detail in HDR, bearing in mind how damaging the ABL can be to very low brightness scenes on an OLED (I suggest a quick look in the Unforgiven thread)...I don't have an LG OLED TV.

4) With the 0-10,000-nit ramp on Sony UHD discs I can see clear down to 0.001 nits.

[edit] You've clearly made your mind up about this issue but my TV isn't crushing anything in HDR, you can take that to the bank. FYI I've measured just under 2000 nits peak from my ZD9 and have two picture modes for HDR viewing, one for 1000-nit grades and one for 4000-nit grades to preserve the relevant amount of highlight information whilst lowering the brightness a tad.

If your HDTV can display 4000 nits mastered contend correctly even without dynamic meta data, so can you explain to me whay something like dynamic data even exist? If some HDTV's like your can resolve everything without dynamic meta data, then this feature is useless?

[imsounoriginal]

Originally Posted by wildphantom

Watching Jurassic World now. Holy god!
The shot when the kid opens the shutters and we fly over the park in all its glory - indescribable until you see it with your own eyes. Wow.

SO excited to see what World looks like. Be still, my heart.

[Geoff D]

Originally Posted by SynViks

So here are a few factors at work when it comes to Nolan's transfers.

1: He likes his masters to look as close to a print as possible, and he has said he does this via A/Bing with a film element. This explains why he likes his raised blacks, goes light on the HDR, and doesn't rely on digital grading to get the look.

2: His films are shot on modern film stocks, which have a very fine grain (imagine it being finer than the DOTD still, which has very fine grain). He also shoots in anamorphic which is going to make the grain even finer vs something like Jurassic Park. We know he doesn't remove the grain since he doesn't grade and since it has a squashed shape.

I'd conjecture that because the grain was so fine, the encoder ate some of it, resulting in less visible grain and those crawling patterns you're reporting (I've not seen the UHDs in motion, and the screencaps are gone, so I can't comment more).

All of that I'm aware of re: Nolan's working practices, but the grain slowing to the point of appearing completely static in certain shots simply doesn't sound like anything that compression alone would result in. And one only has to look at Batman Begins' worst moments - which have nothing to do with any underlying film-related practices whatsover, nor are they a compression problem - to see this "management" at its worst, and I contend that this is also what's been applied to the Nolan UHDs in general, albeit in less severe form on other titles.

Again: I'm not saying I'm against the process itself, it's how it's done in certain cases that irks me. You keep mentioning Nolan's film-based credentials, of which we're all too aware, but there isn't a print of Batman Begins in the world that looks as fudged as the UHD does in certain parts.

[SynViks]

IIRC the ZD9 can hit a peak of about 1700 nits, but that's only within a 10% window; real world usage it will get much less. If you're doing 4k nit grades with no tone mapping or DV the highlights will be clipped.

[bookcase]

Originally Posted by movielover713

So is the best buy steelbook stacked discs?

I'm going to guess yes since Universal put out The Mummy collection steelbook and it was gobs of discs on two spindles.

[SynViks]

Originally Posted by Geoff D

All of that I'm aware of re: Nolan's working practices, but the grain slowing to the point of appearing completely static in certain shots simply doesn't sound like anything that compression alone would result in.

Of course it could. Part of the way the compression works is via prediction and transforming sampled blocks of data to other parts of the image or other frames if they are similar enough.

Basically, I'm conjecturing that the grain was so light that the encoder didn't "recognize it as being substantively different", and "saw it as territory to lower the bit rate", resulting in grain that could get stuck, crawl, etc between reference frames. Some encoders have options to control how the encoder responds to grain, and it sounds like they might have error-ed in that step.

[Geoff D]

Originally Posted by pawel86ck

If your HDTV can display 4000 nits mastered contend correctly even without dynamic meta data, so can you explain to me whay something like dynamic data even exist? If some HDTV's like your can resolve everything without dynamic meta data, then this feature is useless?

Dynamic metadata has one primary purpose, and always has: to ensure optimal downconversion on displays that cannot manage the full capability of the original mastering display. If we all had TVs that could hit 10,000 nits and all of P3 gamut there would literally be no need for dynamic metadata. But we don't, so we do.

As I said, on my 4000-nit mode I have to sacrifice some of the luminance to preserve all the highlight information, it still hits just under 2000 nits peak with a reduction in brightness across the mid range, not a major problem in a light controlled room and when displays still have plenty of light output in general (which was mentioned recently by Spectracal's Tyler Pruitt). Not a 1:1 representation of the master then, yet more than most HDR TVs can manage in the Wild Wild West that is static tone mapping.

[edit] Example: These SDR converted screen caps of Fantastic Beats (4000 max/ 0.005 min mastering as is typical for Warners) are clipping at 200 nits and 4000 nits respectively. 200 is nice and bright but destroys the highlights. 4000 gets you those bulbs back but kills the APL. Well, on my ZD9's 4000-nit settings I'm getting all the highlight detail from the 4000-nit cap but the APL is far closer to the 200-nit shot! My gamma is a bit higher than the 200-nit cap so it's a touch denser in the darker spots and doesn't look as flat as the 200 does.

So, imagine this level of overall picture brightness:



crossed with this amount of highlight retention:



and you can probably start to imagine why I speak about HDR with such authority/arrogance (delete as applicable).

The good thing about Sony's approach is that it maps the actual incoming signal itself to what the TV is set up to receive rather than using the fixed metadata to apply an arbitrary remapping that may be very badly suited to the content, e.g. something which barely hits 200 nits MaxCLL but which is flagged with 10,000-nit Mastering Display metadata! (US disc of BR2049, in case you were wondering.)

A dynamic system would be able to preserve all of the following: the full amount of my TV's peak brightness, the mid-range brightness AND all the highlight information from scene to scene on a 4000-nit encode (or rather a 4000-nit mastering that actually reaches those kinds of numbers). I am awaiting the day when I can put this to the test but that's contingent on having a UHD disc player that can work with the Sony DV profile, and there are none as yet.

[Geoff D]

Originally Posted by SynViks

Of course it could. Part of the way the compression works is via prediction and transforming sampled blocks of data to other parts of the image or other frames if they are similar enough.

Basically, I'm conjecturing that the grain was so light that the encoder didn't "recognize it", and "saw it as territory to lower the bit rate", resulting in grain that could get stuck, crawl, etc between reference frames. Some encoders have options to control how the encoder responds to grain, and it sounds like they might have error-ed in that step.

Funny how that step got error-ed across multiple Nolan movies while other Warners UHDs before and since get on just fine with resolving even the lightest grain fields (or indeed none at all) without turning them into frozen grain, you know? The outlier is of course Nolan using different materials for transfer than the original negatives but even so, using an IP doesn't make the grain disappear, if anything it makes it look a touch more dupey and chunky (as can clearly be seen in that DOTD scanner grab), something that a lossy interframe codec would shirley hang more of its bits on, not less? And, again, what happened to various portions of Batman Begins is not simply a compression issue, but is a symptom of the processing done to these movies turned all the way up to 11 IMO.

[pawel86ck]

Thanks for detailed explanation Geoff D. Because JP movies are mastered at 1000 nits, you should see all the details on your HDTV anyway, so if you see crushed blacks in CGI scenes, then I believe you

[Geoff D]

No worries, I'm always up for a bit of lively debate. Not too lively mind you, my 'directness' has caused a few thread clean-ups of late and I'm sure that the mods are watching very closely!

[Pyoko]

Originally Posted by pawel86ck

But LG active HDR feature make HDR10 content bright as in Dolby Vision mode and you can see shadow details that you could not see before without active HDR enabled. So these shadow details were there before in HDR content, and it was simply HDTV that could not resolve all shadow details without dynamic meta data.

This has not so much to do with dynamic metadata or the nit capability of the set as it has to do with LG's static tonemapping being garbage. Instead of just rolling off the highlights that are outside the TV's capability, LG aggressively lowers the brightness of the entire range in order to save as much highlight information as possible, so dark parts become even darker and sometimes the image becomes dim to the point there's barely any HDR effect left.

There would no need for LG's active HDR or Dolby Vision in bringing back shadow detail if they had a static tonemapping curve that didn't touch the shadow detail in the first place. Part of me almost suspects they've done it this way in order to make their active HDR and Dolby Vision seem better than it really is.

By the way the active HDR is not trustworthy either, as I've seen it brighten shadow detail to the point where posterized parts of the image that were clearly meant to be only barely seen start to pop out. When I have the active HDR enabled I lower the black level by one point because of this (as well as lower the contrast on the output from my Oppo because sometimes the active HDR gets too bright and starts clipping like crazy instead, and there's no good way to fix it on the set itself. The OLEDs are great for the most part and fantastic value, but they have more than their fair share of quirks. )

[Pyoko]

Originally Posted by SynViks

Of course it could. Part of the way the compression works is via prediction and transforming sampled blocks of data to other parts of the image or other frames if they are similar enough.

Basically, I'm conjecturing that the grain was so light that the encoder didn't "recognize it as being substantively different", and "saw it as territory to lower the bit rate", resulting in grain that could get stuck, crawl, etc between reference frames. Some encoders have options to control how the encoder responds to grain, and it sounds like they might have error-ed in that step.

That's not the case with the Nolans, 100% sure of it. That kind of temporal retention in compression is typically accompanied by severe artifacts (as you say it's in order to save bits) and rarely retains fine spatial detail that cleanly. There's nothing of the sort there, in fact the compression is fantastic. The effect is just way too constant and consistent (both across the image and throughout the movies) to be an accident, and compression can't explain why some shots that look exactly the same as surrounding shots in a sequence would suddenly not be affected in the slightest (the answer is of course human error, a selection mistake or perhaps copy-paste mishap and the effect wasn't applied to a few shots.)

All of the Nolans except maybe Dunkirk and The Prestige (haven't watched it yet) are DNR'd to various extents.

[HeavyHitter]

Originally Posted by Pyoko

This has not so much to do with dynamic metadata or the nit capability of the set as it has to do with LG's static tonemapping being garbage. Instead of just rolling off the highlights that are outside the TV's capability, LG aggressively lowers the brightness of the entire range in order to save as much highlight information as possible, so dark parts become even darker and sometimes the image becomes dim to the point there's barely any HDR effect left.

There would no need for LG's active HDR or Dolby Vision in bringing back shadow detail if they had a static tonemapping curve that didn't touch the shadow detail in the first place. Part of me almost suspects they've done it this way in order to make their active HDR and Dolby Vision seem better than it really is.

By the way the active HDR is not trustworthy either, as I've seen it brighten shadow detail to the point where posterized parts of the image that were clearly meant to be only barely seen start to pop out. When I have the active HDR enabled I lower the black level by one point because of this (as well as lower the contrast on the output from my Oppo because sometimes the active HDR gets too bright and starts clipping like crazy instead, and there's no good way to fix it on the set itself. The OLEDs are great for the most part and fantastic value, but they have more than their fair share of quirks. )

Curious how the 2018 LG OLEDs handle this as I've heard there are some notable improvements.

[JG-havkroll]

Originally Posted by Joshjp

Ohh nice..its a good price to...wish the 4k version had that.

Same. Missed opportunity. Oh well.

[Pyoko]

Originally Posted by SynViks

As for grain in general, I feel like it's helpful to take a look at this still from the Dawn of the Dead IP, which, aside from having a simple balancing, is straight from the scanner with no "DNR":

[Show spoiler]

Note that the grain isn't as contrasted as the Sony grain. It's very fine, and is very noticeable in the chroma as well as the luma (whereas Sony grain seems far more visible in the chroma). Obviously the rules by which the grain manifests are governed partly by the stock used...

It's important to remember that, when compressing a grained image, the HEVC encoder will smudge the grain a little bit and potentially lessen the chroma component since the chroma resolution is HD.

That's why I feel the Jurassic Park transfer is a good example of proper grain management on UHD:
-It's very fine
-It's very visible in the chroma component
-It doesn't look like noise

Obviously if the grain were finer or more resolute I'd prefer that, but there are compromises that have to be accepted with compressed video.

It depends on the film stock but I imagine it could also be affected slightly by the scanner used, depending on the sensor and resolution.

But this is a pretty nice comparison actually because I believe the grain in that frame is probably pretty close to what JP has, or maybe should have had, in that the grain has very prominent dye blobs in the blue/yellow channel, with the luminance and red/green channels being fainter. But that frame still shows a beautiful, pixel-fine luminance grain which is largely, sometimes completely, absent in JP, which is mostly showing the softer blue/yellow blobs.

Part of this is compression for sure, it's not particularly great, but my theory is that they've applied some DNR to the finest luminance frequencies and you can catch hints of it in some of the smeared detail and frayed edges. I've seen enough of Universal's DNR to see the signs, and this looks like it, only more subtle and applied a bit differently. This would also make sense as the parts of the grain that are there really do look untouched for the most part, there's very little temporal slowdown or trailing or anything like that.

And I would definitely call the grain in that frame Sony-level, but it's also significantly nicer than what's on display on the JP UHD in my opinion. I don't think the compression excuse is valid either, just look at what Universal managed with The Mummy.

[Pyoko]

Originally Posted by HeavyHitter

Curious how the 2018 LG OLEDs handle this as I've heard there are some notable improvements.

To be perfectly honest I've purposely avoided reading anything about the 8-series in order to avoid any stupid thoughts of upgrading after only a year... Though from the pre-release information it seemed to be more of a "tick" year for their OLEDs.