Color wheels - just a little bit more
Illumination or gasoline? You decide.
Hopefully this encapsulates what the issues are.
Oliver Peters - oliverpeters.com
[Oliver Peters] "Illumination or gasoline? You decide."
Thanks for taking the time to write this article. I see it as illuminating not fire bombing.
It's obvious there's a "problem" in how the wheels work in 709. If they were designed for HDR, then they should only work when the Project is in an HDR color space. The same goes for the color board and 709.
As a result of these discussions, I have definitely looked in to HDR more than I ever have. With Bill Davis's permission, hopefully it's OK to think aloud on this forum, even if all the answer's aren't available.
When using HDR in Lumetri (in Pr or Ae) there's a button to click when moving to HDR space. There is also a scale to choose in the scopes when moving to HDR. If you don't choose this new scale, the brightest whites appear to clip, but in actuality they don't. Here's two pictures of the same ramp, just using different waveform options. The "8bit" option appears to clip, but look at the difference in the scopes in HDR.
HDR is an interesting situation. It seems that 10,000 nits, or the most HDR of the HDR formats is the new ceiling, which is 100 times brighter, at least by scale on the waveform, than 709 or 100 IRE (or 1 in the terms that we have been discussing in recent threads). Soon we will need sunglasses just to watch TV.
But the best and brightest (ha!) of monitors gets to about 4000 nits, and the best consumer HDR monitors get to about 1000 nits. Many consumer HDR sets only hit 300 nits (3 times brighter than a normal HDTV). So it seems that the stage has been set, it's just a matter of time before technology catches up. What is hardest about this is that so far, every display is different. It seems like this would make it very difficult to grade in HDR since there's such a wide discrepancy in display options, meaning some HDR sets get much much brighter than others. It seems this creates an even bigger minefield of options for consumers, and will open up the door to weirder looking images as people won't know how to setup the tv correctly (or simply don't have the same capabilities despite them all being "HDR TVs", as Stu Maschwitz talked about), and it makes delivery options for content creators much more complicated. Add to that Alexis Van Hurkman's discussion about the Automatic Brightness Limiting circuits that will be governed in to TV sets so they don't consume too much power. The circuit will attempt to limit the amount of brightness on a scale (for instance, it will allow 2% of the overall image to reach peak brightness) which seems to suggest that the "advantages" of HDR may be overridden by a circuit to limit power consumption. What will that do to the look of the image?
Despite all of the HDR options, the ever increasing pace of emerging technology, I think that that the biggest question still remains, why don't the new FCPX Color Wheels work in boring ole 709 color space? 👍👹😳
Again, thanks for writing and taking the time.
[Jeremy Garchow] "It seems like this would make it very difficult to grade in HDR since there's such a wide discrepancy in display options, meaning some HDR sets get much much brighter than others."
My understanding of this is that there is tone mapping that happens on the display side based on metadata in the file. If you grade your video on a 1,000nit monitor there is a flag in the file that it's graded for 1,000 nits. If you play it back on a 300 nit monitor the display sees "This is designed for 1,000 nits, adjust accordingly" and your 300 nit display will remap the image to fit, as best it can, the 1,000 nit grade into the 300 nit space.
That's my very basic understanding of how the wide discrepancies between HDR monitors is being handled. And of course, like in all situations like this - trust a calibrated monitor in your suite and let the chips fall where they may when your video is released into the wild.
[Michael Hancock] "That's my very basic understanding of how the wide discrepancies between HDR monitors is being handled. And of course, like in all situations like this - trust a calibrated monitor in your suite and let the chips fall where they may when your video is released into the wild.
No question. It would seem that it would be even worse than 709, though as the variables are higher.
[Michael Hancock] "My understanding of this is that there is tone mapping that happens on the display side based on metadata in the file. If you grade your video on a 1,000nit monitor there is a flag in the file that it's graded for 1,000 nits."
That's kind of how I understand it too, but it still seems like you'd need to know the final output, and most times the tone mapping is used to transform from one color space to another. But you are right, that it seems there is some sort of curve function built in to the spec in order to account for the different brightness levels. Still, is all of that functionality created equal in all devices? :)
[Jeremy Garchow] "Still, is all of that functionality created equal in all devices? ☺
LOL. That's the beauty of standards - there are so many to choose from!
I find it semi-weird that these programs even offer an HDR mode as the actual reference monitors required to do this sort of work cost like 30 thousand euros eg. the Sony BWM-X300. Then there are people with some random "HDR"-televisions "grading" in "HDR" and wondering why it doesn't look good.
The FCPX documentation seems to only address HDR for YouTube. My guess is that YouTube and the iPhone X/iPad Pro is their real target. Dolby Vision (PQ) is there because they could. That's doesn't mean it's right, though.
Until someone does real public testing with a PQ-rated system, we are all just guessing. I've done some test sequences in PQ and ironically these display just fine in 709 (although there are some color issues, particularly reds). I believe that's because PQ is a layered system, so you don't see the extra highlight range until viewed on a PQ display.
However, my issue with the FCPX PQ implementation is that it doesn't just extend the highlights. It extends everything in PQ, which to my understanding (still limited at this point) isn't the right approach. But as a friend of my said, "Isn't HDR just the new 'vivid' mode on a TV?"
Oliver Peters - oliverpeters.com
[Tero Ahlfors] "I find it semi-weird that these programs even offer an HDR mode as the actual reference monitors required to do this sort of work cost like 30 thousand euros"
Someone has to be the first to introduce a new feature that may or may not ultimately become considered standard or essential. But more often than not, software development is driven by marketing considerations rather than by feature utility. I'm not singling out Apple in this, most companies are guilty of it in some form. Bug fixes and productivity enhancements don't rate highly in terms of marketing appeal. It's much easier to tout "cool new feature X," regardless of its ultimate relevance to the majority of customers.
This doesn't only apply to software. In an attempt to boost sales, electronics manufacturers regularly push new features that are ultimately not improvements, but mere gimmicks. Examples include 3D TVs and curved OLED screens. 3D is all but dead, and curved OLEDs have largely been rejected. Marketing is driven by sexy sounding features and numbers. Megahertz, megapixels, resolution specs, bogus contrast ratio ratings, 0-60 times, horsepower ratings, etc. The usefulness of those features and specifications is irrelevant - their purpose in marketing is to sell more products.