Originally Posted by EvLee
That's a curious spectrum. The 2020 blue is defined to be 467 nm. Their readings show them going even lower with a peak down at 448 nm. That would produce a more purplish-blue and also result in some of the 2020 gamut being cut off. It also increases the potential blue light hazard. Any idea if those blue LEDs are the same that go into the televisions, or is it possible they went with a lower wavelength (higher energy) to increase emission and make a more impactful demo?
Full disclosure: I work for Nanosys and am the guy holding the spectrometer in Scott's photos above. I’m also new to this community and I hope you guys find it appropriate for me to contribute here.
Sharp eye indeed- you are absolutely right about blue the wavelength in the spectrum and I think I can help explain. There are really two things going on here:
1) This little demo backlight system is not representative of a rec.2020 TV backlight.
The backlight in question is about 450nm but it happens to be from a tablet display that used that wavelength. While you are correct that choosing a shorter wavelength could ever so slightly improve efficiency and therefore brightness, the effect would be really subtle. This backlight just happened to be a convenient platform on which to build a neat demo. The film here is also a fairly random “cutting room floor” scrap sample (although the bottles are correct) so please don't read too much into the precise wavelengths shown here.
I’ve attached a real spectrum here from the backlight of an approx 90% coverage rec.2020 at D65 white LCD that we’ve shipped for accurate reference (taken with a PR
-655 in our lab).
2) Funny enough, your comment is still spot on because we often find that different wavelengths help us to achieve maximum coverage for a given gamut in an LCD display.
Why would we do this? The short answer is that it turns out that the blue (467nm) and green (532) wavelengths for rec.2020 are extremely close to each other. When you pair this with a typical LCD color filter, you end up with a significant amount of “leakage” between the sub pixels. This means that when you open up the blue sub pixel, you’ve got some of that short-wavelength green leaking through, which pulls the blue primary off of its target. This is also gets to CineTechGeek’s comment about narrower primaries– the rec.2020 primaries are in fact so close that even an ideal laser light source doesn’t actually buy you much more coverage. As a result, we find that we're able to actually maximize front of screen rec.2020 coverage by pushing those two primaries apart. See attached CIE plot showing gamut coverage we've achieved with this kind of optimization.
Can we do better? Yep. You have to keep in mind that, in the “85-90% rec.2020 coverage” devices that are shipping today with our QD technology, we’re being asked to work with relatively loose color filters that were originally designed to reproduce rec.709 with a white LED light source! When we are allowed to choose the off-the-shelf color filter we like best we can push it to 93.7% coverage today and we’re working with a color filter maker on a slightly tweaked blue formulation that would take you to 96-97% coverage.
Getting beyond that is really tough for any display technology. One quick reason as to why- triangles are really finicky. The primaries are already on the locus so you can't make deeper primaries and calibrate back to 2020 in order to guarantee coverage with MP manufacturing tolerances the way you can with other, smaller gamuts. One small shift in any primary and you've lost a big chunk of you're coverage.
Our CEO gave a more thorough discussion on this topic at a conference last year if you’d like to do a deeper dive. You'll find the most relevant portion starting at about 13:26: