- This topic has 12 replies, 2 voices, and was last updated 6 years, 9 months ago by
.
-
Posts
-
Hello,
Windows 10
U2412M x2 (different manufacturing dates)
Nvidia gtx 670
i1 Display proMonths ago I was advised by my peers to invest in color calibration as I might get into printing, so I ended up getting an i1 display pro from x-rite and decided to pair it with displaycal/argyllcms as that seems to be preferred over the company software by many people. Unfortunately I also have a consumer nvidia card (670) which apparently is pretty bad for calibration because nvidia blocks 10 bit color output for professional applications for consumer cards.
When I calibrated months ago I was just using all default settings in the gamma 2.2 setting which resulted in banding (Googling around it seems that it’s because of my GPU? I read that if I had an AMD card instead this wouldn’t happen). I read setting tone curve to As Measured instead of Gamma 2.2 can prevent this and this seemed to work or at least lessened the effect, so I just did that and used those profiles for a few months. Using Tools -> Report -> Report on Uncalibrated Display Device shows my gamma as around 1.98 for one monitor and 2.04 or so for the other while both are trying to hit 6500k+79-80cdms , so that could be it too.
A week ago I noticed a greyscale image with srgb profile attached having some coloration (See attached image) which is really odd because removing the image’s srgb profile also removes the minor color saturation. I used a window color picker (http://annystudio.com/software/colorpicker/) and an image editor w/ out of window color picker to check this. I thought it was because of firefox or chrome, but after reading stuff like http://www.gballard.net/windows_srgb/ I figure I would try to see if it was the monitor profile, and after updating displaycal/argyll and recalibrating it seems to be the case or is at least influencing this (before this recalibration the srgb image’s grays were much more saturated in color, the rightmost square before white for example was quite blue; r=222,g=230,b=232).
after recalibration for both Monitor #1 and #2 and checking grayscale image w/ a color picker:
Firefox rgb values from leftmost box to right: [0,0,0] [35,35,34][60,59,60] [82,82,83] [106,106,107] [132,131,132] [160,159,160] [190,190,191] [223,222,223] [255,255,255]
Firefox image srgb profile removed: [0,0,0] [28,28,28] [57,57,57] [85,85,85] [113,113,113] [142,142,142] [170,170,170] [198,198,198] [227,227,227] [255,255,255]chrome rgb values from leftmost box to right: [10,10,10] [30,30,30] [61,61,61] [82,82,82] [106,106,106] [132,132,132] [160,160,160] [190,190,190] [223,223,223] [255,255,255]
chrome image srgb profile removed: [0,0,0] [28,28,28] [57,57,57] [85,85,85] [113,113,113] [142,142,142] [170,170,170] [198,198,198] [227,227,227] [255,255,255]- So what accounts for this difference in rgb values? I figure attaching a srgb icc profile to an image can change its overall colors (..right?), but gray scale should be gray scale regardless of the image’s attached color profile right? For anyone here with properly calibrated gear, is there this difference on your screens as well?
- Given my circumstance and current hardware, was what I did the best I could do in terms of calibration while wanting to avoid banding? (Default displaycal settings with tone curve set to As Measured instead of Gamma 2.2, everything else the same; during interactive display adjustment get rgb bars close to the center as possible while trying to hit 6500k temperature, brightness @ ~79-80 cdms).
- What am I actually losing out by using As Measured in tone curve instead of Gamma 2.2, assuming my monitor’s native gamma is lower than 2.2? Everyone seems to heavily advise doing Gamma 2.2 but that doesn’t seem to be an option for me due to the monitor’s gamma and my gpu.
- I read some people opt to use “Single Curve + Matrix” (with black point compensation disabled?) instead of the default “XYZ LUT + Matrix” under Profiling w/ advanced options enabled, to avoid banding, discoloration and other problems. Why does this potentially lower the chance of such problems like banding and other things? It seems the only difference is the number of testcharts? Do I need to Reset Video Card Gamma Table if I want to try Single Curve + Matrix if I had previously used the default XYZ LUT + Matrix?
- Also, when trying to use the Verification function I get the following error (See attached image). Why is this? Also seems to happen when trying file -> create profile from measurement data.
I apologize if my questions seem boneheaded, I’m trying my best to get my head around calibration in general but it is quite daunting and headache inducing. :\
Thank you
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Hi,
after reading stuff like http://www.gballard.net/windows_srgb/
The information given in that blog post seems questionable if not plain wrong. The introductory sentence contains the passage “eight Windows Vista computers were displaying tagged sRGB wrong in Photoshop and certain Web browsers, but correctly in Internet Explorer IE”, but Internet Explorer never supported correct color management to begin with (older versions didn’t do color management at all, newer versions started assuming sRGB for the display profile). I would ignore the blog post.
So what accounts for this difference in rgb values?
For color managed grays, the combination of output R, G and B is simply the mixture that is needed to produce a neutral (achromatic) color. Also note that color pickers do not take the videoLUT (1D calibration) into account.
Given my circumstance and current hardware, was what I did the best I could do in terms of calibration while wanting to avoid banding?
Yes, although banding is usually only a problem with artificial (noiseless) images, or images where the starting point is already quantized too much (e.g. badly exposed photos not processed in high bitdpeth RAW mode).
What am I actually losing out by using As Measured in tone curve
Non color managed content may not produce neutral grays.
I read some people opt to use “Single Curve + Matrix” (with black point compensation disabled?) instead of the default “XYZ LUT + Matrix” under Profiling w/ advanced options enabled, to avoid banding, discoloration and other problems. Why does this potentially lower the chance of such problems like banding and other things?
Not per se, but some (in my opinion broken) color managed applications may not be able to use the (more accurate) cLUT part of display profiles.
Also, when trying to use the Verification function I get the following error (See attached image). Why is this?
Make sure no external tools are interfering (e.g. virus scanner) and blocking access to files.
Thanks for the reply, Florian!
For color managed grays, the combination of output R, G and B is simply the mixture that is needed to produce a neutral (achromatic) color. Also note that color pickers do not take the videoLUT (1D calibration) into account.
Huh, I see. So it is actually not that unusual for a grayscale photograph/image to show grays slightly colored (r,g,b values not completely equal) when viewed in a color managed software in a calibrated workstation? Or grays having different r,g,b, values compared to when color management is off?
This is also the same as colored images right where the rgb numbers change from a color managed software to one where it is disabled? For example, sRGB tagged images lose overall color saturation in a color managed setting (from what I can see anyway), however in a non-calibrated workstation it appears the same whether color management is on or off?
Make sure no external tools are interfering (e.g. virus scanner) and blocking access to files.
I tried running displaycal as administrator and it still is giving me that error when trying to use the Verification function, I also only have the default Windows Defender Anti-Virus which I don’t think gave me any warnings or notifications. I can’t think of anything else that could be blocking it. I think the UAC thing in windows gives you a popup but I didn’t get one, and I think everything else in the Security Center is default. Is there anything else I can try?
If there is an error that’s supposed to show up the windows Event Viewer log, I’m not sure where to look for in there
Again, thanks for the help!
Huh, I see. So it is actually not that unusual for a grayscale photograph/image to show grays slightly colored (r,g,b values not completely equal) when viewed in a color managed software in a calibrated workstation?
Just because RGB is unequal doesn’t mean the grayscale is colored. It’s the other way around, neutral grayscale often requires unequal device RGB to account for the display device’s actual response characteristic.
This is also the same as colored images right where the rgb numbers change from a color managed software to one where it is disabled?
Yes.
For example, sRGB tagged images lose overall color saturation in a color managed setting
If the device you’re displaying the sRGB images on has a bigger gamut than sRGB, then they will appear oversaturated without color management (maybe only slightly if the monitor gamut is only slightly bigger).
Florian,
Thanks again for your response, I really appreciate it.
Just because RGB is unequal doesn’t mean the grayscale is colored. It’s the other way around, neutral grayscale often requires unequal device RGB to account for the display device’s actual response characteristic.
Ahhhhhh, reading this actually made it more sense now, I feel dumb haha. Thanks for this.
If the device you’re displaying the sRGB images on has a bigger gamut than sRGB, then they will appear oversaturated without color management (maybe only slightly if the monitor gamut is only slightly bigger).
Hmm, the difference still seems a bit much to me (in terms of brightness and color saturation) when looking at photographs/images made with sRGB profile in mind and comparing them with it off. But of course this could just be my perception playing tricks. I attached a gif of me toggling srgb on and off in a color managed software. Does this appear normal?
I also still could not determine why the Verification function still doesn’t work and gives me the same error message as before (see previous attached image). If anyone has any idea what could be causing this please let me know. Running Displaycal as admin doesn’t seem to do it, and I’m fairly sure this windows account is already the admin.
Also, I looked up my monitor’s panel (LM240WU8-SLA2): http://www.panelook.com/LM240WU8-SLA2_LG%20Display_24.0_LCM_overview_13818.html
Does this need the “White LED IPS (WLED AC LG Samsung)” spectral correction? Referring to the “4 strings WLED.” Or is that option generally for higher end IPS?
Thanks again!
I attached a gif of me toggling srgb on and off in a color managed software. Does this appear normal?
Yes it does.
Does this need the “White LED IPS (WLED AC LG Samsung)” spectral correction? Referring to the “4 strings WLED.”
Yes.
Florian,
Thanks again. I also managed to make the Verification function work after many, many hours of playing around with it: renaming it from the default name made it output the html file without giving me an error. I am not sure why the default html file name causes an error. Hopefully someone will find this helpful
For the verification I just did the defaults with simulation profile checked with “sRGB IEC61966-2.1.” I unchecked “Use simulation profile as target profile” and Tone Curve was default at “Apply black output offset (100%)”. Do these results look okay, or does something look off? This is also after recalibrating with the Spectral correction LCD White LED IPS.
I might give Gamma 2.2 under Tone Curve a try again when I get back my i1 display (lending it to someone who is also learning color management for a day or so). Although it sounds like a bad idea since both my monitors are barely or not even hitting ~2.0 gamma as their native, do you think it is worth it over the As Measured option? The reason I’m asking is because some of the software I work with has no color management and I’m hoping to lower the difference between them and color managed software that are using sRGB work space.
Do these results look okay
Looking good.
do you think it is worth it over the As Measured option? The reason I’m asking is because some of the software I work with has no color management and I’m hoping to lower the difference between them and color managed software that are using sRGB work space.
If you want to match sRGB material (atleast in terms of tone response) between color managed and non-color managed apps, use the sRGB curve for calibration.
Florian,
Thanks again.
If you want to match sRGB material (atleast in terms of tone response) between color managed and non-color managed apps, use the sRGB curve for calibration.
Ahh, I completely forgot about this other option! I’ll definitely try this out when I get my i1 Pro back. Hopefully the banding won’t be too bad otherwise I’ll probably stick to native gamma.
Also, I realized the reason my monitor’s native gamma has dropped from it’s marketed 2.2 could be because of their age and usage. If you look at reviews of my monitor many report around ~2.2 when it was released years ago. Is adjusting the monitor’s contrast setting able to compensate for this similar to how increasing brightness and the RGB settings can compensate for color/brightness drifting as LCD monitors age? Or is it something only video card LUT tables can compensate for?
Another thing, (I apologize, I cannot find the modify/edit post button): does the windows’ own calibration tool (I think called “Display Color Calibration”) generate its own icc or modify the graphics card LUT? I noticed it can adjust gamma settings. I assume displaycal removes and overwrite these settings when it calibrate/profiles? No one seems to mention this program anywhere when using displaycal/i1profiler/etc, so I assume it’s a bad idea to play around with this or is completely redundant with Displaycal?
does the windows’ own calibration tool (I think called “Display Color Calibration”) generate its own icc or modify the graphics card LUT?
Yes, it can generate very simple, not at all accurate profiles that contain calibration information that only alters color temperature and gamma.
assume displaycal removes and overwrite these settings when it calibrate/profiles?
When the profile is installed, yes.
[…] is completely redundant with Displaycal?
Yes.
Florian,
Thanks! that’s good to know; I’ll definitely stay away from windows’ calib tool then.
About my contrast question, my friend tested this with the i1 display pro and said that contrast seems to affect gamma at least on his viewsonic LCD monitor (I think he said higher contrast = lower gamma reading in displaycal’s uncalibrated report & lower contrast = higher gamma). This is weird because all I read is that contrast setting does not affect gamma at all.
I read that many people say to try not to touch the contrast setting in LCD monitors, but which would be less destructive to color accuracy: just letting displaycal change gamma through tone curve setting with a less than ideal hardware setup (like 6-bit+dithering monitor 2.0 native gamma & consumer gpu) or adjusting monitor contrast to get closer to gamma target and -then- let Displaycal do its thing? Does monitor age actually affect/lower gamma & contrast? (Like how rgb & brightness drifts in LCD monitors with age? in which case shouldn’t contrast be adjusted as it ages then?)
