- This topic has 15 replies, 2 voices, and was last updated 7 years, 11 months ago by
.
-
Posts
-
Hello,
I am a new Displaycal user, and quite a noob in understanding the depths of display color calibration. I need a well calibrated display for photography work, and until now, have been using Spyder 5 Pro software. I became aware of Displaycal´s advantages, it seems like a magnificient software, if one knows how to use it´s potential, which I apparently don´t.
For some reason I seem to be unable to create a good calibration in the shadow area. I have a DELL U2715H display, which is attached via Displayport to NVIDIA GeForce GTX 970 GPU.
Attached are screenshots of my calibration setting (zipped), and also a htm-verification report. Looking at the report, the results seem very good. But when looking at variable LCD test imaged via browser (firefox) or Lightroom 5.7, there is noticeable reddish color cast in the shadows and near blacks, and black/white crushing. This is also confirmed by looking at the Calibration and tone response curves for the profile.
What am I doing wrong? I have tried all sorts of input variations in Displaycal, many sorts of profile types, and read through this forum. What ever my setting are, they always seem to result in this red color cast problem and crushing.
I´m using spyder 5 pro puck, and to my understanding, the corrections via mode selection in “display & instrument” are correct. The Spyder OEM software also creates a reddish (even more pronounced) color cast, but is does not crush the blacks or whites, to my understanding at least.
Any and all help would be much appreciated!
Mikko
Well, it seems that when I view my (attached) screenshots of the LCD test images on another computer and display there is no color cast in the shadows. I guess this is as expected, given the other computer+display is not using the same profile.
However, on the computer that the calibration was done in, and the profile used, there really is a distinctive reddish color cast and black crushing. The screenhots of calibration curves and tone response curves (from Displaycal, also attached) do show this issue I believe.
Mikko
Hi,
disable black point correction on the calibration tab (I would generally recommend to leave this option disabled) and re-do the calibration/profile. The Spyder5 seems to struggle to get accurate readings near black.
Thank you Florian for the answer. So Disabling the black point correlation is achieved by unchecking the “auto” option, and setting the “rate” slider to zero? Correct? what about the “Rate” value? Does it have any effect if “rate” is zero?
I will test this option again. I have in fact done it before, but checked the “auto” option because that seemed to result in more satifactory dE:s, but the color casts and crushing were still present.
Thanks again.
Mikko
Just unckecking “Auto” and setting the correction % to zero is enough, rate should stay unchanged at 4.
crushing were still present
Crushing could also mean a video <-> PC levels mismatch. You may want to check your graphics driver settings (“Desktop color” -> “Dynamic range” in the nVidia control panel), also see https://hub.displaycal.net/forums/topic/washed-out-screen/#post-2301
I have the Dynamic range set as “Full 0-255”, I thought this problem with would only affect HDMI-attached displays. Mine is attached from GPU Displayport to monitor’s MiniDisplayport. Also, when there is no calibration/icc profile used by Windows except the DELL U2715H default, I see near blacks perfectly fine with no color casts and perfect grey scale slides from 0 to 255.
But then I don´t know if the color and/or gamma is correct.
The greyscale/crushing/color cast problem only seems to come when using Displaycal or Datacolor software calibrations.
BTW, should the calibration be done in as dark as an environment as possible? No other light sources than the monitor? Can a bright ambient light like indirect sunlight screw up the results especially near the blacks, even if the puck is well placed flat on the monitor?
Thanks,
Mikko
Can a bright ambient light like indirect sunlight screw up the results especially near the blacks, even if the puck is well placed flat on the monitor?
It can influence the result. While it’s not necessary to profile in a completely darkened room (although it certainly doesn’t hurt either), high levels of ambient light (even if indirect) falling onto the monitor and probe during calibration/profiling should be avoided.
Hi Florian,
I ran the calibration/profiling again with black point correction off, as you advised. The calibration was done in a very dark room. The results did not change. Attached there are screenshots of the input settings, calibration results and the measurement report of a verification run (zipped). The issue is still present as can be seen from the tone response and calibration curves, and the resulting profile creates a strong red color cast and also strong black clipping in the darkest areas. Once again, a verification run of the calibration shows everything as more than OK, which is weird.
I also did several test runs with Datacolor software. It also creates a reddish cast in the shadows, but much less (if any) black clipping.
What in the world could be the issue here? I have checked and double checked the nVidia control panel, all seems to be well there. Also, with the calibration off, I can see rgb-values down to approx 2,2,2 differentiated from 0,0,0. So it is not the “rgb 16-235” problem.
I starting to feel a bit desperate. Any ideas forward with Displaycal?
Mikko
Well, I ran Displaycal again but this time with a Spyder 3 device. I also set White Level to “as measured”, previously I had that set to a slightly lower value than the actual to prevent black crush. That was the guidance in Displaycal manual. Also, I increased the number of profiling test charts to 1134, which is 2-3 times more than before.
The results seem very good now (data attached). There´s still a slight “bump” at the low end of the tone response curves, as seen in the screenshot. Is this normal and to be expected? The measurement report (also attached) looks very good also, but so it did for the bad profiles too. But now, when the profile is installed, test pictures and lightroom photos for example look very good, no colorization or crushing anymore.
So, do you think that this is because of:
-A defective Spyder 5 device?
– Setting the White Level to as measured?
-Increasing the number of test charts in profiling?
My guess would be a defective Spyder 5 puck. Even Spyder 5 software with Spyder 5 puck can not create a usable profile. Everything is reddish and there is crushing also in the deepest end of tones. With spyder 3 puck and datacolor software, a good looking calibration was done at the first run. With Spyder 3 and Displaycal however, I did the first run with white level set slightly below actual, and 400 or so test charts. Results were bad in the same manner as with the spyder 5 puck. Then I set the white level to as measured and increased the number of test charts to 1143, and results, as attached, look very good! So, this seems to indicate that it might not be just because of the Spyder 5 puck…puzzling. Any insights into this?
Thanks a lot for you help and support, and an incredible software!
My guess would be a defective Spyder 5 puck.
Could be. Do you still have warranty?
Fortunately yes, I just puchased the thing. I have started the process with Datacolor Europe, from where I bought it directly.
Do you think that the slight “bump” at the low end of the tone response curves of Spyder 3 calibration is normal and nothing to worry about?I did get a bad result with Spyder 3 initially also, before changing White Level to “as measured” and increased the number of test charts in profiling. Which one of these changes made the situation better? Do you have an opinion?
Do you think that the slight “bump” at the low end of the tone response curves of Spyder 3 calibration is normal and nothing to worry about?
No, that’s a typical gamma 2.2 curve shape.
I did get a bad result with Spyder 3 initially also, before changing White Level to “as measured” and increased the number of test charts in profiling. Which one of these changes made the situation better?
Unless you give me some specifics (what does “bad” mean?), it’s hard to comment.
