- This topic has 23 replies, 4 voices, and was last updated 5 years, 2 months ago by
.
-
Posts
-
Hi again,
I recently asked here about colorimeter correction profiles and if it was ok to use one for another brand and model which uses the same panel.
I got the reply that it would only produce a slight difference in white point compared to the bundled white LED correction.Anyway, I have an Acer XV240YP (144hz IPS) and an i1 Display Pro with which I’ve tried several different settings in DisplayCAL and got quite interesting results. Initially I went with the sRGB preset since I almost can’t see the shadow detail if I go with 2.2. (I’m not in a dark room)
Initially I did two calibrations, one for work/web with about 65 cd/m2 (I sit quite close to the screen) and one for games at about 95 cd/m2.
The first one I ran at 65 cd/m2 (high speed, 175 patches) ended up with dE avg 0.07, max 1.67, RMS 0.19.
When I made the gaming calibration at 95 cd/m2 I decided to try and do a long calibration and check the difference.
I ran that one at medium with 4954 patches, the “very large” testchart. This gave me the result of a dE avg 0.01, max 1.71, RMS 0.07.I noticed banding in gradients with both, after which I found out about the registry hack for nvidia drivers. I have a GTX 1060 and currently driver 457.30.
(The latest driver has a severe flicker bug in a game I often play so I wait with updating it.)
The registry hack worked and reduced the banding though some of it is still visible. I made a scheduled task to apply it on every startup.Then I started searching around here for opinions about calibration speed and the amount of test patches. It seemed as if medium with 175 patches was considered the sweet spot. It made me feel as if I wasted my time doing that long calibration for what I guess would be an invisible difference to the eye.
By the way, what does the RMS value stand for or represent?
Yesterday during the day, with some slight sun outside meaning the room was a bit brighter than usual, I ran 6 different calibrations back to back for different settings. All on high speed and 175 patches. That RMS value increased after each. I also seem to have more banding issues on these 6 than the first 2.
Before the initial 2 calibrations I had the i1 Display Pro sitting on the screen while both warmed up for about an hour. Then I read somewhere that you’re supposed to let the i1 Display Pro warm up on its own, not being on the display. Which is true? For my 2nd round of calibrations I had them warm up with the colorimeter not sitting on the display.
Then it also came to my mind that I had dithering enabled in the registry, unlike the first calibrations. Would that affect those newer calibrations in a negative way? Should I disable dithering before calibrating?
And is it ok to run 6 calibrations back to back like I did there? Or is that RMS value rising each time an indicator that I shouldn’t?
My goal is to have one setting for work/web, one for games and one for movies. For movies I’ll most likely go with the Rec. 1886 preset.
So bottom line: What is the correct way to treat the display and colorimeter, is it ok to run calibrations after one another and what are truly the sweet spot settings? Also, should dithering only be enabled in registry when not calibrating?
Thank you so much in advance! And thanks for this amazing piece of software!
Oh, and what would be the best preset for re-calibrating my old Samsung 40″ CCFL LCD TV?
“LCD CCFL family” correction, right? (That would be done using HCFR)Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Then I started searching around here for opinions about calibration speed and the amount of test patches. It seemed as if medium with 175 patches was considered the sweet spot. It made me feel as if I wasted my time doing that long calibration for what I guess would be an invisible difference to the eye.
No.
Calibration is one thing: grey scale, “speed” is amount of patches & iterations for grey.
Profiling is another, it is to measure behavior after calibration, on a well behaved display after calibration 175 add enough grey patches to capture TRC in more detail (brightness 1xTRC, brightnes & color for each grey 3xTRC). That’s the reson you can name it sweet spot: matrix 1TRC = accurate grey brghtness, matrix 3TRC = same + accurate grey color info, XYZLUT = same but better volume mapping (at teh expense of some apps incompatibilities).
You can have long calibration (speed =slow), but profile at 175 patches.By the way, what does the RMS value stand for or represent?
Root mean square
https://en.wikipedia.org/wiki/Root_mean_square
https://en.wikipedia.org/wiki/Root-mean-square_deviation
Yesterday during the day, with some slight sun outside meaning the room was a bit brighter than usual, I ran 6 different calibrations back to back for different settings. All on high speed and 175 patches. That RMS value increased after each. I also seem to have more banding issues on these 6 than the first 2.
Speed high = less grey patches on calibration stage = what is not measured cannot be corrected.
Slider speed:
12 => 24 => 48 => 96 patches in an iterative process, if you move to slower you do all previous iterations (12+24+…)
Before the initial 2 calibrations I had the i1 Display Pro sitting on the screen while both warmed up for about an hour. Then I read somewhere that you’re supposed to let the i1 Display Pro warm up on its own, not being on the display. Which is true? For my 2nd round of calibrations I had them warm up with the colorimeter not sitting on the display.
That meant electrical warmup, just that, just a few minutes. No display is needed. Of course we talk about LCD panels not plasmas or oleds.
Then it also came to my mind that I had dithering enabled in the registry, unlike the first calibrations. Would that affect those newer calibrations in a negative way? Should I disable dithering before calibrating?
Display should be configured as you are going to use it, although Argyll may not notice dithering while calibrating since depending on calibration speed unditehred patches may fall under dE threshold for that iteration.
My goal is to have one setting for work/web, one for games and one for movies. For movies I’ll most likely go with the Rec. 1886 preset.
Do not even try Rec18886, is not meant for you. Too low contrast display.
For games you need an sRGB/Rec709 mode or use saturation controls (use HCFR for them).
For movies you can try games approach (with 2.2 or 2.4 gamma) or go full color management with madVR and LUT3D.
Oh, and what would be the best preset for re-calibrating my old Samsung 40″ CCFL LCD TV?
“LCD CCFL family” correction, right? (That would be done using HCFR)If it’s a sRGB/Rec709 only display, yes. (it should be, IDNK of widegamut TVs CCFL)
“For games you need an sRGB/Rec709” if you have a widegamut monitor, like those P3 displays 1xxHz for games what we see these days.
Do not even try Rec18886, is not meant for you. Too low contrast display.
The way I get it is Rec 1886 is gamma 2.4 but with a bump in the near-black area. Do you mean it’s for OLED displays only that can achieve actual zero black levels? I know it’s been around since 2011 and back then everyone recommended 2.2 as a standard and 2.4 for cinema or dark rooms.
For games you need an sRGB/Rec709 mode or use saturation controls (use HCFR for them).
My PC display is not wide gamut so I’m just aiming for sRGB/Rec709. As for saturation controls, my monitor has hue and saturation controls for each color but any changes to them would only affect 100% saturated colors in a positive way. After adjusting them, the 75%, 50% and 25% got sky-high dE values which must mean the hue and saturation controls on the monitor are useless. On my CCFL LCD TV I’m familiar with all its settings since I calibrated it one time 10 years ago using HCFR and got as near Rec 709 gamma 2.2 as I could, with all color hues very close to reference. That TV just lacks a little bit of red saturation is all.
Anyway, I know I’m only gonna use the low speed from now on to get as many measurements as possible. i1 Display Pro is quite fast anyway. And I guess my biggest mistake was to do the calibration during daytime.
Also, with “Calibration is one thing: grey scale,” alright, but the profiling process must correct hue, saturation and luminance of the primaries and secondaries, right? There is no way for me to measure my calibrated settings in a way I’m familiar with since if I try launching HCFR to do just that, DisplayCAL automatically disables the ICC profile and resets the video card gamma table. I guess I gotta bring out my old laptop sometime to measure the calibration. I’d love to see the values of all the color saturations and everything.
Rec1886: No, it modifies the whole curve. If you want ~2.4 ut works for OLED, plsma or high contrast VA. Otherwise choose 2.2 or 2.4 directly with relative.
Also, with “Calibration is one thing: grey scale,” alright, but the profiling process must correct hue, saturation and luminance of the primaries and secondaries, right? There is no way for me to measure my calibrated settings in a way I’m familiar with since if I try launching HCFR to do just that, DisplayCAL automatically disables the ICC profile and resets the video card gamma table. I guess I gotta bring out my old laptop sometime to measure the calibration. I’d love to see the values of all the color saturations and everything.
Profiling corrects nothing. It’s a taylor made suit, that’s all.
Corerctions from X colorspace to display colospace is color management engine responsibility = none in games or non color managed apps.Displaycal measurement report can:
-verify if calibrated display matches its profile
-verify if calibrated display matches its profile and can be used to work with X colorspace image (simulation profile)
-verify if UNcalibrated display or HW calibrated display matches X colorspace(simulation profile & use as display profile)if you want to test if your grey calibrated display matches some colospace use HCFR (and make sure LUT is not disabled / clear LUT data is not enabled)
Profiling corrects nothing. It’s a taylor made suit, that’s all.
Corerctions from X colorspace to display colospace is color management engine responsibility = none in games or non color managed apps.I’ve probably not quite understood how ICC profiles work at all then, I guess. I thought they could at least correct color hues sent to the display, for example. I’d understand if they can’t correct luminance and saturation but at least hue?
Profile corrects nothing.
<<optional>> VCGT data contained in some of them corrects grey in 1D once loaded into GPU. This grey correction is global.
If VCGT says (RGB)
Grey 50: RGB 50->50, 50->49,50->48
Grey 150: RGB 150->151, 150->148, 150->148
etc, just 2 samples of VCGT
Then for RGB color (150, 50, 0)
150->151, 50->49, 0->0
Each RGB color takes 1D grey corrected values. That’s all.If you want more, a colro mangement engine is needed and apps must use it. Games don’t use it (unless Reshade stuff)
If you want more, a colro mangement engine is needed and apps must use it. Games don’t use it (unless Reshade stuff)
Thanks for all the useful info. I’ll look into getting 3D LUT for movies.
Weird that they don’t include support for 3D LUT in video card drivers then, so it would be applied globally.Another thing you could try is setting the tone curve to “as measured” when profiling. Assuming color managed apps work in a higher bit depth it will reduce banding. In theory less grey patches will be corrected so check with a verification report.
Weird that they don’t include support for 3D LUT in video card drivers then, so it would be applied globally.
Computational resources using GPGPU and in our political-ecofriendly world that means more energy. I do not have measured Watts running madVR but I assume high GPU usage.
Full dedicated HW with low energy goal like in high end displays implies more expensive cards (and I’m not sure about low energy goal as design target)A Lut-matrix-lut like some middle end monitors with HW calibration may be feasible for white, grey and basic colospace simulation. In fact AMD have that since 2005. They can emulate sRGB on the fly from EDID data from display. Unfortunately no user friendly way to upload your own data. If they had such generic vendor agnostic api you may have high bithdepth dithered calibration akin to HW calibration in Eizo CS. AFAIK no propietary APIs from some vendor will be inckuded in ArgyllCMS or DisplayCAL.
Computational resources using GPGPU and in our political-ecofriendly world that means more energy. I do not have measured Watts running madVR but I assume high GPU usage.
To be honest, PCs in general draw so much power these days that I doubt it would make things much worse. But of course the best thing would be full HLS controls on the displays. I can’t understand why the hue and saturation controls on mine ruin the lower saturations if fixing them at 100%. At 25% you can’t touch those settings at all. Shouldn’t they all just be going in a straight line from the greyscale to fully saturated? I have to keep trying some different settings. At least the hues on my display are not that far off really. The highest dE is on red which is a bit oversaturated.
Maybe it’s HCFR’s settings fault? The gamma curve was very wavy in the lower end before calibration and maybe that affects its measurements of the saturations.
Profile corrects nothing.
Sorry for continuing this but mind that even though I know some things, I’m quite the beginner when it comes to 3D LUT and I never truly learned what an ICC profile actually does. Can it be explained in an easier way?
I mean, for example: Some people get those DCI-P3 gamut IPS panels these days. As far as I know, DCI-P3 is a standard for cinema and covers the colors which 35mm film can capture. Most consumer content still uses the sRGB color space, so if what you say is true and you only get a more correct greyscale through VCGT calibration, won’t the ICC profile handle the primaries and secondaries? Or will such monitors just end up with a correct greyscale but way oversaturated colors?
I thought the ICC profile was what dialed in the colors in supported apps, so to speak. And Windows itself supports ICC profiles although with less bitdepth than DisplayCAL profile loader, or something like that. Can’t the ICC profiles even fix the hue of colors? I just don’t get it. If there is no way to limit your monitor to the sRGB gamut then why even make DCI-P3 panels and advertise them as “gaming monitors?” I know some have an “sRGB mode,” mine has one and all it does it lock the backlight to a luminance of about 250 cd/m2 which would basically burn out my eyes but it doesn’t move the primaries and secondaries into the sRGB gamut at all, I checked using HCFR. (I found the setting to prevent HCFR from messing with the VCGT too.)
I’m sorry for the confusion. I just don’t get the point with wide gamut monitors being marketed for things they aren’t meant to if there is no way to “swap between gamuts” on them so to speak, not even through calibration/ICC.
Profile corrects nothing.
Sorry for continuing this but mind that even though I know some things, I’m quite the beginner when it comes to 3D LUT and I never truly learned what an ICC profile actually does. Can it be explained in an easier way?
I mean, for example: Some people get those DCI-P3 gamut IPS panels these days. As far as I know, DCI-P3 is a standard for cinema and covers the colors which 35mm film can capture. Most consumer content still uses the sRGB color space, so if what you say is true and you only get a more correct greyscale through VCGT calibration, won’t the ICC profile handle the primaries and secondaries? Or will such monitors just end up with a correct greyscale but way oversaturated colors?
All oversaturated unless:
-such monitor has an sRGB mode
-such monitor has RGB/RGBCMY saturation controls and you modify those values with HCFR to match sRGB/Rec709
-you have an AMD card to enable sRGB emulation based on EDID data
-you use some app that supports color management : Lightroom, Photoshop, madVR compatible video players or if properly configured/simple profiles MS Edge and FirefoxI thought the ICC profile was what dialed in the colors in supported apps, so to speak.
Correct. SO is just a notification publisher
And Windows itself supports ICC profiles although with less bitdepth than DisplayCAL profile loader, or something like that. Can’t the ICC profiles even fix the hue of colors? I just don’t get it.
No. Windows acts like a notice board. If apps ask for profile, windows returns it, but app is responsible to do things with it, not OS.
If there is no way to limit your monitor to the sRGB gamut then why even make DCI-P3 panels and advertise them as “gaming monitors?” I know some have an “sRGB mode,” mine has one and all it does it lock the backlight to a luminance of about 250 cd/m2 which would basically burn out my eyes but it doesn’t move the primaries and secondaries into the sRGB gamut at all, I checked using HCFR. (I found the setting to prevent HCFR from messing with the VCGT too.)
Explained above, check in “games” OSD modes for RGBCMY saturation controls and modify them with HCFR feedback measurements
I’m sorry for the confusion. I just don’t get the point with wide gamut monitors being marketed for things they aren’t meant to if there is no way to “swap between gamuts” on them so to speak, not even through calibration/ICC.
Surprise! Yes, is a nonsense selling WG monitors for gamming without an sRGB emulation or RGBCMY saturation controls.
-you have an AMD card to enable sRGB emulation based on EDID data
Sadly I have an Nvidia. At the time I upgraded a few years ago they were way ahead of AMD in performance.
Surprise! Yes, is a nonsense selling WG monitors for gamming without an sRGB emulation or RGBCMY saturation controls.
I’m glad I didn’t fall for the marketing bluff. Almost everywhere the most recommended budget monitor is AOC G2U, either 24″ or 27″, and the info about it says “126% sRGB.” I can only imagine how awful that must look. Is that the value that’s called “volume” in DisplayCAL?
-you have an AMD card to enable sRGB emulation based on EDID data
Sadly I have an Nvidia. At the time I upgraded a few years ago they were way ahead of AMD in performance.
Surprise! Yes, is a nonsense selling WG monitors for gamming without an sRGB emulation or RGBCMY saturation controls.
I’m glad I didn’t fall for the marketing bluff. Almost everywhere the most recommended budget monitor is AOC G2U, either 24″ or 27″, and the info about it says “126% sRGB.” I can only imagine how awful that must look. Is that the value that’s called “volume” in DisplayCAL?
Yes. It may not look awful if it is a P3 only screen (no AdobeRGB green) and have saturation controls. Check that AOC user manual to see OSD options.
