Posts: 2,049
Threads: 56
Joined: 2016 Dec
Thanks: 161
Given 1008 thank(s) in 612 post(s)
I've been doing a lot of digging around with regards to these blanket tints that a lot of Blu Rays tend to receive, especially with regards to those which have a DCP cinema release which is unaffected despite coming from the same master/restoration (obvious examples are The Terminator and Aliens). In doing so I've come across various forum threads and articles regarding the DCI colour space P3 and it's white point which is noticably green. Just search for 'P3 white point' and a ton of stuff comes up. This thread in particular stood out:
http://www.liftgammagain.com/forum/index...oint.1326/
I'm guessing something is happening on the conversion to Rec709, ie the white point has not been compensated, resulting in green whites. Once the discussion starts going into xyz values my brain goes a bit mushy but I'm sure there are members here who can make more sense of this.
Posts: 1,108
Threads: 26
Joined: 2015 Jan
Thanks: 679
Given 304 thank(s) in 205 post(s)
2017-07-26, 11:10 PM
(This post was last modified: 2017-07-26, 11:19 PM by CSchmidlapp.)
I think it was you who suggested this in a previous threads.
In that the conversion from the more Recent Masters (4k generally) to Bluray was not being done correctly, in terms of colour conversions especially.
I think you were spot on then and even more so now.
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country: 
I still wonder if the people who worked on color correction regrading etc. have some vision problems, works with old CRTs, or new diplays not calibrated, or simply don't care about the whole thing, as long as they continue to be paid...
And, we should not forget about the magenta blanket - maybe it's related to some color space conversion as well...
Posts: 2,049
Threads: 56
Joined: 2016 Dec
Thanks: 161
Given 1008 thank(s) in 612 post(s)
It could explain why 50/50 green/magenta color mixes work so well if they are polar opposites.
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
(2017-07-26, 11:44 PM)zoidberg Wrote: It could explain why 50/50 green/magenta color mixes work so well if they are polar opposites.
I think it works well, but not because they are exact opposites, even if they should be in close opposite position, though...
Well, until now the approach of diminishing the green (or, better, teal) seems to work well, but I'm pretty sure that if we use the exact formula to convert DCI-P3 to BT709, the result would be better... or not? 
Posts: 2,049
Threads: 56
Joined: 2016 Dec
Thanks: 161
Given 1008 thank(s) in 612 post(s)
You could say that subtracting green is the same as adding magenta (red+blue) and vice versa.
Ideally we would be able to arrive at an avisynth script or possibly a LUT that would perform the exact correction so as to present what was intended rather than what might look subjectively 'right'.
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
(2017-07-27, 12:53 AM)zoidberg Wrote: You could say that subtracting green is the same as adding magenta (red+blue) and vice versa.
That would be true if the "green" version is full green (00ff00), and the "magenta" full magenta (ff00ff)... actually they vary, so you should find the closest blanket color to get rid of it - take a look here to have an idea of RGB color names.
Quote:Ideally we would be able to arrive at an avisynth script or possibly a LUT that would perform the exact correction so as to present what was intended rather than what might look subjectively 'right'.
Best solution would be to have an avisynth function AND a LUT to use with non-avisynth softwares.
I could live also with few ones that would correct the most common cases, more or less perfectly, tough.
Posts: 2,049
Threads: 56
Joined: 2016 Dec
Thanks: 161
Given 1008 thank(s) in 612 post(s)
I've been experimenting with RGBadjust in avisynth, I find that once the bulk of the blanket green is removed there is a lot of 'wiggle room' for fine tuning. So an approximate value may be enough in the short term. It could well be that different studios/mastering houses are interpreting the standards ever so slightly differently therefore requiring various adjustments. What I am finding is that one global correction can bring a whole film into line without any further adjustments.
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1466 thank(s) in 963 post(s)
Country:
(2017-07-27, 01:33 AM)zoidberg Wrote: What I am finding is that one global correction can bring a whole film into line without any further adjustments.
Yes, it's what I've done with my latest experiments... the problem is to find the right setting! 
deleted user
Unregistered
Thanks:
Given thank(s) in post(s)
Holy shit I love you guys haha. I just stumbled over this by accident while working on a grade and I think this IS in fact the answer.
I did a few tests on a BvS trailer with what I believed to be a "fixed" conversion and it in fact did neutralize the whites! Then I googled "blu rays green tint DCI-P3 whitepoint" and guess what was the first result? This forum post! Well granted, Google may already know I visit this site a lot, so that may be it.
Anyway, the key here is (in Adobe slang) the colorimetric intent. When converting into another colorspace, Adobe asks you whether you want Absolute or Relative Colorimetric. Turns out, Absolute means that the white point is kept. Meanwhile relative means that the white point is adapted to the target colorspace.
So: Take the green-tinted video in sRGB/Rec709 ... convert to DCI-P3 with ABSOLUTE colorimetric intent. Then convert back to sRGB with RELATIVE colorimetric intent. Voilla. 
This probably also explains why any attempts with curves never resulted in truly proper results ... those attempts work in the sRGB/Rec709 color space and this color space has different primaries (basically the parameters that define the precise color and intensity of the R, G and B components). Since the original discoloration comes from the DCI-P3 colorspace in a way, it probably has to be compensated there with the DCI-P3 primaries.... I'm just guessing though to be fair.
Check out these beauties:
http://www.framecompare.com/image-compar...n/DP777NNX
http://www.framecompare.com/image-compar...n/J299FNNU
Maybe it's just coincidence that it worked here and I'm being overexcited but I'm really happy with how these turned out!
P.S. Turns out it's not actually just a green tint, it's a tint both towards a warmer and greener whitepoint.
|
