- This topic has 35 replies, 4 voices, and was last updated 1 year, 8 months ago by
.
-
Posts
-
Hi
In case this helpful for future reference, I have contacted ledoge about the ‘ calibrate gamma’ checkbox that we discussed and he replied with the following:
‘Yes, you should enable it. Without it, grayscale is left entirely uncorrected. In your case (assuming that you set the white point using the monitorˈs RGB controls, i.e. 100% white does not look red but only gray tones do), you’re seeing a tint because your monitor’s grayscale tracking is poor. With the gamma option enabled, the grayscale data from the profile is used to correct that and also make the luminance follow 2.2 relative gamma.
(the above explanation assumes that you’re seeing the same tint when not using novideo_srgb at all, otherwise there’s something else going on).’
I think it may do more than just grayscale calibration. It says “full LUT-Matrix-LUT calibration” but I’m not sure what that means. If I’m guessing, maybe a LUT for the grayscale and a Matrix (RGBCMY primaries and secondaries only) for the colors?
ICC profiles are also supported and can be used in two different ways. By default, only the primary coordinates from the ICC profile will be used in place of the values reported in the EDID. This is useful if you want to use a profile created by someone else without taking their gamma/grayscale balance data into account, as that can vary a lot between units.
If you enable the Calibrate gamma to checkbox, a full LUT-Matrix-LUT calibration will be applied. This is similar to the hardware calibration supported by some monitors and can be used to achieve great color and grayscale accuracy on well-behaved displays.
Ledoge also says:
To achieve optimal results, consider creating a custom testchart in DisplayCAL with a high number of neutral (grayscale) patches, such as 256. With that, a grayscale calibration (setting “Tone curve” to anything other than “As measured”) should be unnecessary unless your display lacks RGB gain controls, but can lead to better accuracy, especially on poorly behaved displays. The number of colored patches should not matter much.
Additionally, configuring DisplayCAL to generate a “Curves + matrix” profile with “Black point compensation” disabled should also result in a lower average error than using an XYZ LUT profile. This advice is based on what worked well for a handful of users, so if you have anything else to add, please let me know.
I think it may do more than just grayscale calibration.
No, mixing primaries require “degamma” to work with linear input, otherwise you won’t be able to mix primary colors. If you do a raw/simple sRGB emulation, like AMD driver, it will take nominal/expected gamma of display (let’s say 2.2 or sRGB) and use that gamma for degamma, then compute matrix mixing priaries to emukate sRGB, then reencode again using that gamma.
Applying grey calibration is just multiplying component by component VCGT calibration to some of the 1DLUTs in lut-matrix-lut. In HW this is usually done in latest lut but others may implement ion in the other one to ensure that amtrix is actually in displays’ linear gamma. IDNK which ine is uses in Ledoge’s app, source code will tell you that.
