Quantcast

Answered Can I clarify something about digital intermediates?

Vertigo1

Active Member
There's much talk about digital intermediates (DIs) in 4K Bluray circles of late, specifically how the vast majority of titles aren't "real 4K" but instead upscaled from a 2K DI. The reasons for this are obvious - it's much less demanding on equipment, storage, processing power and so forth to use a 2K DI than it is to use a 4K one.

Now all of the following is speculation on my part - I just want to know if I'm right or not.

2K refers to a resolution of 2048x1080 whereas 4K refers to 4096x2160. People have inferred from this that, as any 4K Bluray which uses a 2K DI has just been upscaled from 1080 lines, by definition it can't offer much more resolution than a regular Bluray would, as regular Blurays are capable of 1080 lines of resolution. Any actual benefits in terms of detail would be down to the studio doing a far more meticulous job of the upscale than your player would manage.

Is this actually true though?

Remember that virtually all films are actually around a 21:9 aspect ratio and are displayed letterboxed on our TVs. A regular Bluray contains a 16:9 image however, with the black bars encoded into the video stream, so in reality the move itself doesn't use the full 1080 vertical lines but nearer the 822 mark, the remaining lines are the black bars at the top and bottom.

Now the 2K DI from which the film is taken would contain an anamorphic version of the 21:9 source using the full resolution, so there's a full 1080 vertical lines of resolution there. When this is used to create a 16:9 Bluray, the image is letterboxed and vertical resolution is lost, but when the same process is used to create a 16:9 Bluray, there's twice the vertical resolution available so, even though the image has to be letterboxed, we still end up with about 1644 vertical lines in which to represent the image, so all of the original 2K DI's detail can be preserved.

So, in reality, a 4K Bluray taken from a 2K DI does offer increased resolution/detail over the equivalent Bluray because the latter has to compress the original 1080 lines into about 822 whereas the 4K Bluray does not.

Ironically, if this is actually the case, it follows that a 4K Bluray taken from a 4K DI is also not showing us the full originally captured quality as it'll suffer from the same issue as does a regular Bluray from a 2K DI, namely that the 2160 lines of the 4K DI will end up being compressed into about 1644 lines on the Bluray and the rest will be black bars.
 

Geoff_D

Distinguished Member
Sure, but one thing to bear in mind is that films are NOT finished "anamorphic" in the way you're thinking of. 1.85 movies are output for a 1998x1080 2K or 3996x2160 4K finish and 2.39 movies are output for a 2048x858 2K or 4096x1716 finish. There's no such thing as a full "21:9 anamorphic" final source master because there's no need for it, and certainly no provision for it in current digital cinema projection standards.

In other words the finished master usually retains its original aspect, it's not arbitrarily stretched or squashed into one container or another. While it's true that some DSMs (digital source masters) are rendered out at a taller than 'scope ratio e.g. 1.90 or 1.78 to take into account open-matte IMAX or TV versions for 'scope shows but still, that's actual picture information filling out the height, not an anamorphic squeeze. Sure, certain non-standard ratios e.g. the 2.20 of Tomorrowland or the 2.00 of Jurassic World have to be jury-rigged to fit those two theatrical DCP containers (1.85 and 2.39), but their finished masters don't reflect that as Jurassic World was finished at "2K plus" 2398x1556 resolution which is nearer 1.56.

I see what you're saying about the potential for uprezzed 2K movies to have more spatial resolution in the UHD versus the BD and that's simply because 2K into HD doesn't go, the 2048x858/1998x1080 image has to be adjusted to fit 1920x1080 which can be done by either scaling the image in the appropriate dimension(s) or by simply cropping the horizontal resolution down to fit. Cropping will give you a 1:1 pixel relationship between the HD and the 2K master BUT naturally you lose picture information and the ratio will be altered, e.g. 1.85 becomes 1.78 which isn't so bad, but 2.39 ends up nearer to 2.23 so the image will need to be cinched in further in the vertical dimension in order to maintain the ratio (this is before the black bars are baked in for home video purposes). Or you can scale, which will maintain the full image area inside the 16:9 frame size with the appropriate borders for 1.85 & 2.39 BUT you run the risk of inducing artefacts like aliasing which is why some filtering is applied to actively soften the image during this process.

UHD Blu-ray itself has to deal with these same things re: the 4K source masters because 4096x1716/3996x2160 into 3840x2160 16:9 doesn't go, but because the 4K source has (hopefully) been upscaled from the original 2K material (preferably the source DPX or MXF files) then yes: it should in theory carry across any detail that was lost when doing the 2K to HD conversion. But we have to bear in mind that the upscaling process is just that - another round of scaling - so even that has to be mindful to not induce any artefacts into the final result.

From my own experience with having compared 50+ UHDs to their Blu-ray counterparts I can honestly say that many of the 2K upscaled UHDs do indeed carry across more resolvable resolution than their Blu equivalent e.g. Magnificent Seven is a recent example, and even earlier titles like Pi, Lego Movie and San Andreas clearly pull down a touch more spatial detail. But this is not true of all movies, as some are the same as the Blu-ray (I wasn't impressed with X-Men First Class at all) and one has infamously been softened up the arsehole: Oblivion is demonstrably softer than the Blu-ray. Now, even these movies have other benefits like the extra highlight information that the wide dynamic range brings, but no way no how should they be softer than the Blu-ray.

HTH
 
Last edited:

stjernholm

Active Member
Geoff, world class explanation! Thanks.
 

Similar threads

Top Bottom