- This topic has 19 replies, 3 voices, and was last updated 6 years ago by
.
-
Posts
-
This is my first use of DisplayCal.
I’m trying to calibrate a Panasonic TX-L37DT30E edge-LED IPS with a i1D3.
My first question is: which correction (spectral) should I use? I’ve seen 3 IPS (CCFL, RGBLED and WLED). Which one should I choose, if any? I suppose the only one not to choose should be CCFL…
Thank you
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Thank you.
I tried both with 16-235 levels, so the correct one is WLED, fine.
I got this result after calibration (after adjusting on TV Menu settings the white point with R+2 and G+2) + ICC profiling, in verification :
attached html file
So it seems quite good (maybe too good on DeltaE??), apart black that’s the limit of this panel.
But I checked also with Calman ColorChecker Free, just to compare the DeltaE and here I got relatively bad results (average Delta of 1,7 with at least 5x >3, Max @4.1),
quite different from DisplayCal (average DeltaE 0.36, max 2.15):
So, what’s going on? Which one is correct?
Does Calman Free use ICC profile created by Dispcal when checking the calibration?
How can I verify if a prog is using the correct Dispcal ICC profile?
I tried both with 16-235 levels
Contrast seems awfully low and black level extremely high, though. 16-235 does not seem to be the right choice.
Does Calman Free use ICC profile created by Dispcal when checking the calibration?
Good question.
How can I verify if a prog is using the correct Dispcal ICC profile?
Maybe try installing one of the test profiles (in DisplayCAL’s installation folder under “tests”) which will introduce a strong color cast via the calibration. If the cast is in the CalMAN ColorChecker patches when measuring them, you know the display profile isn’t used (the profile would cancel out the color cast otherwise).
I tried both with 16-235 levels
Contrast seems awfully low and black level extremely high, though. 16-235 does not seem to be the right choice.
Apart poor black, there is also this on the white of TV I read on a review: “For the seriously tech-inclined: the Panasonic 37DT30 discards some “above white” data at all times, with levels 246-255 being clipped into peak white. Adjusting the user andservice menu controls did not return full dynamic range. This is still acceptable.”
So, I thought it could be used with 16-235, but I will test it full range. Then I could make a calibration settings specific for video levels when playing video materials.
On Calman:
I’m sorry, I’m newbie on profiles: I right click on some of them and select “install” but nothing changed. I noticed only they were copied in the folder “C:\Windows\System32\spool\drivers\color” besides the other Displaycal created.
But the desktop, Photoshop and the rest is all with correct colors. When clicking on Install, should I SEE something changed on normal programs, or at least in those ICC complaint? I see all normal…
Could you give me exact procedure? TIA
I tried a full range calibration.
This time I also reset the tv settings and started from scratch: Profile Professional1, then I lowered the contrast from 30 to 12 to match the white level on Dispcal panel, and then the white point with R+1 G0 B-1.
From report I attach, it seems a better black level 0.579 cd/m² and better contrast 206.8:1. Is this good?
BUT I got a max DeltaE on black of 3.61 the other values seem quite good.
Anyway on this review:
you can see a pic of a quite complete gamut 709, whilst Dispcal seems to leave it more larger on green? Is it correct?
So, should I reach a better result with Dispcal?
For the seriously tech-inclined: the Panasonic 37DT30 discards some “above white” data at all times, with levels 246-255 being clipped into peak white.
It’s usually not an issue with actual content because that should not contain above video legal (i.e. 235) values. It may affect test patterns.
I’m sorry, I’m newbie on profiles: I right click on some of them and select “install” but nothing changed. I noticed only they were copied in the folder “C:\Windows\System32\ spool\drivers\color” besides the other Displaycal created.
Use the DisplayCAL profile loader tray application to associate the profile to the display.
From report I attach, it seems a better black level 0.579 cd/m² and better contrast 206.8:1. Is this good?
Not exactly good, but this TV seems to be rather contrast limited to begin with (it can only reach about 550:1 which is very low for a TV). Still, I’m inclined to think the black level on your TV is set incorrectly. You should use a PLUGE pattern (prior to calibration) to fix this.
I got a max DeltaE on black of 3.61
That’s normal for uncorrected black point (which will have a different hue than the rest of the grayscale). The alternative is to correct the black point, but this will reduce contrast even further without visual gain (hue shift at black usually isn’t very visible).
I used a pluge to lower the tv luminosity to lower the black level.
Now I got the attached result. Little better contrast about 300:1Thank you Florian, now I got the ICC profile thing.
I put a greenish ICC and loaded Calman: it did not change it, so I assume Calman doesn’t interfere with the DisplayCal correct profile loaded. But, I understood the ICC profile worked only on ICC compliant progs: how is it that I see this profile loaded everywhere, desktop etc. Infact I thought on non ICC prog, on Win, it should load only the gray correction, instead…
Anyway, assumed Calman leaves the profile intact, it remains strange it measures so different deltaE values, particularly on reds…:
With 0-255, Luminosity -13:I put a greenish ICC and loaded Calman: it did not change it, so I assume Calman doesn’t interfere with the DisplayCal correct profile loaded
It’s not so much about interference, but rather, is it actually using the profile. In a correctly color managed program, there will be no green color cast.
But, I understood the ICC profile worked only on ICC compliant progs: how is it that I see this profile loaded everywhere, desktop etc.
What you’re seeing is the effect of the 1D calibration (which introduces a strong color cast on purpose). In color managed programs, this strong color cast will not be there.
Anyway, assumed Calman leaves the profile intact, it remains strange it measures so different deltaE values, particularly on reds…
I don’t think CalMAN does anything with the profile. It probably just compares whatever was measured to a static target (which you can probably configure somewhere in CalMAN).
Thank you for the explanation.
So, if I understood well, with the correct DisplayCal ICC profile, Windows and also Calman load only the calibration part: it lacks the correction made on the profile part. So this explains Calman bigger DeltaE.
In other words it’s a middle way between measuring without any correction at all and applying only the calibration without the major part of the profile, needed to really lower the DeltaE. Is it right?A profile is not a correction. All it does is record the relation of device (RGB) values sent to the display, to the light emitted by the display (and subsequently measured by the instrument). You can compare it to a dictionary. Any form of “correction” or color management (= translation of color) only happens when you link two profiles together: The source profile tells the color management module (CMM) which (device-independent) color you want to have, and the CMM then uses the display profile to determine which combination of R, G and B it needs to send to the display to produce that color.
Ok.
I did also a MadVR 3D-LUT, and I got this (attached). I attach also the previous ICC (0-255 and 16-235) to compare.
I put also the zip file: do you notice some color clipping (where to look)?It’s good, BUT I see some worse deltaE regards the previous ICC profile. In particular with
Red = 2.11
in ICC was quite perfect:
Red = 0.02 !!Also other values are worse even if the overall average is perfect, like ICC one.
More, contrast lowered by 90 (maybe 16-235 vs 0-255?).Is it normal to have a 3D LUT worse than ICC? Can I improve something?
Note the difference of verifying a profile against the display response (roundtrip error/profile self check) vs. verifying with a simulation profile (i.e. a target you want reproduced using said profile). Your results look good either way.
Ok, I was a bit concerned about reds.
Infact I played a color clipping pattern and I can see gray rectangles (flashing) in all the colors (maybe difficult with Magenta also, but I can see them), but RED I see NOTHING… it seems all clipped above EDIT 230, others instead reach 235.
What do you think?
