- This topic has 9 replies, 3 voices, and was last updated 1 year, 2 months ago by
.
-
Posts
-
Can someone explain the difference between “Relative” and “Absolute” next to the gamma curve setting after enabling “Advanced options”? The documentation on DisplayCal didn’t really do it for me.
In AVS forum you’ll see graphical examples
https://www.avsforum.com/threads/madvr-argyllcms.1471169/page-208#post-40108362
It’s about how you deal with non ideal blackpoint (finite contrast). Relative (bending the gamma curve) is the usual approach for power law gamma calibration targets.I’ve read it, but I am not still not sure about the difference between Relative and Absolute.
Also, black output offset seems confusing to me as well.
Can someone explain these settings in simple terms?
As far as I know, generally the point of gamma is to give visually even progression from black to white. (Perceptual quantizer is something I would want to explore next, it seems to be a whole different beast)
Also, black output offset seems confusing to me as well.
Equivalence between “naming” in several apps in in that link too.
Can someone explain these settings in simple terms?
They are described in the simplest way
http://www.argyllcms.com/doc/collink.html#IbAs far as I know, generally the point of gamma is to give visually even progression from black to white. (Perceptual quantizer is something I would want to explore next, it seems to be a whole different beast)
Yes, but you have a finite contrast display, hence darkest patches cannot follow a gamma power law. How do you deal with it? Easiest way ys to scale it, to have a power law gamma “if you assume that non zero black is black”. That is what graphical examples show.
b) option is the typical approach for monitors.”
The lower case b and g flags use an effective power value specification, where the 50% input maps to the same output value as a pure power curve with a perfect black. These are the recommended flags to use, since this means that the overall contrast of the reproduction will not be affected so much by differences and variations in the black level, or how the black level is accounted for. The default effective power value is 2.2.
The upper case B and G flags use a technical power value specification, where the power value is the actual one that will be used. The end result will vary much more with the black level of the display though. The default technical power value is 2.4.
”
So in Argyll terms, effective means relative and technical means absolute. My understanding is that effective will make it so that at 50% input, a display will obey a pure power curve. But for the technical part, what does “power value” mean?
”
The second optional :p.p value is the proportion of black value that should be accounted for as output offset (the gamma value must also be specified after this). This defaults to 0.0 for the b and B flags, and 1.0 for the g and G flags, but by specifying it explicitly as a value between 0.0 and 1.0, a hybrid curve characteristic can be obtained. Setting a hybrid value is a way of maintaining black shadow tonality while being able to control how much detail is retained in those shadows. An appropriate value may depend on exactly how the video material was mastered.
”
I think I almost understand what black output offset means. To confirm, does “black shadow tonality” mean near-display-minimum shades of black? Moreover, does 100% output offset mean that near-black shadows will be clipped for non-zero black displays (everything other than OLED and miniLED)?
”
The lower case b and g flags use an effectivepower value specification, where the 50% input maps to the same output value as a pure power curve with a perfect black. These are the recommended flags to use, since this means that the overall contrast of the reproduction will not be affected so much by differences and variations in the black level, or how the black level is accounted for. The default effective power value is 2.2.
The upper case B and G flags use a technicalpower value specification, where the power value is the actual one that will be used. The end result will vary much more with the black level of the display though. The default technical power value is 2.4.
”
So in Argyll terms, effective means relative and technical means absolute. My understanding is that effective will make it so that at 50% input, a display will obey a pure power curve. But for the technical part, what does “power value” mean?
Grey gradient scaled to [0-1] interval will follow log (out) / log (in) = power value. “Gamma” definition.
”
The second optional :p.p value is the proportion of black value that should be accounted for as output offset (the gamma value must also be specified after this). This defaults to 0.0 for the b and B flags, and 1.0 for the g and G flags, but by specifying it explicitly as a value between 0.0 and 1.0, a hybrid curve characteristic can be obtained. Setting a hybrid value is a way of maintaining black shadow tonality while being able to control how much detail is retained in those shadows. An appropriate value may depend on exactly how the video material was mastered.
”
I think I almost understand what black output offset means. To confirm, does “black shadow tonality” mean near-display-minimum shades of black? Moreover, does 100% output offset mean that near-black shadows will be clipped for non-zero black displays (everything other than OLED and miniLED)?
Is about to keep some brightness distance between dark greys. Relative 100% would “bend” a power law gamma graph near blacks to avoid clipping, hence if the typical setup for monitors since color management will undo whatever calibration target you put.
You can see it in the sample graphs.
Is about to keep some brightness distance between dark greys. Relative 100% would “bend” a power law gamma graph near blacks to avoid clipping, hence if the typical setup for monitors since color management will undo whatever calibration target you put.
You can see it in the sample graphs.
I am not fully understanding the graphs attached here. What are the axes? And what do the measurement results mean?
If Relative is equivalent to bending a gamma graph, what is Absolute equivalent to?
x=RGB number sent to GPU (or monitor) scaled from 0 to 100% signal
y=”gamma” value2.4 power law gamma “bends” near black: dark patches bright a little more than requested to keep a small separation between greys
And yet, it is still not entirely clear in which cases it is necessary to use an absolute or relative gamma, and it is also not clear what percentage of offset to choose and what to be guided by.
Let’s say I want to work in a standard close to Netflix, (0.05cd black point, contrast 2000:1 and 100cd). I have an IPS WLED matrix with a black point of 0.11cd at 100cd and a contrast close to 1000:1. What should I choose to get closer to the standard without having a reference monitor?
When set to absolute, I get values close to 2.2 (when set to 2.4), but nice looking images. When choosing relative, with different black offsets, I get some kind of sharp transition from the minimum black level, with a drop in contrast in the shadows. It looks like some kind of clipping that you want to raise, but you can’t.
There is also some confusion with the 16-235 video level. If I’m on a PC with levels 0-256, do I need to switch to video levels or not (when viewing, creating, and exporting video content)?And yet, it is still not entirely clear in which cases it is necessary to use an absolute or relative gamma, and it is also not clear what percentage of offset to choose and what to be guided by.
Let’s say I want to work in a standard close to Netflix, (0.05cd black point, contrast 2000:1 and 100cd). I have an IPS WLED matrix with a black point of 0.11cd at 100cd and a contrast close to 1000:1. What should I choose to get closer to the standard without having a reference monitor?
When set to absolute, I get values close to 2.2 (when set to 2.4), but nice looking images. When choosing relative, with different black offsets, I get some kind of sharp transition from the minimum black level, with a drop in contrast in the shadows. It looks like some kind of clipping that you want to raise, but you can’t.I’d say that with a limited contrast display aim for relative 2.4
There is also some confusion with the 16-235 video level. If I’m on a PC with levels 0-256, do I need to switch to video levels or not (when viewing, creating, and exporting video content)?
Content is 16-235. Video player on a computer scales it to 0-255 unless you configured it to do not.
So, let’s say that you have a common IPS computer display connected to a common GPU (nvidia, AMD…. no decklink involved) and want to use as Resolve GUI or as a display to consume content with typical video players like MPC or even through madVR render.
GPU <-> display 0-255 (by default unless some nvidia weird behavior over HDMI)
Content 16-235 (you won’t do anything)
Video player confifured to use PClevels 0-255 (it will scale content for you).
So you always have it configured 0-255. You may want to set 16-235 if display (a TV in certain OD mode) expect data in 16-235.That’s for common GPUs, if you need to configure a decklink you’ll need to check your display capabilities first.
