- This topic has 89 replies, 5 voices, and was last updated 5 years, 2 months ago by
.
-
Posts
-
Plug LG to PC through the same input as decklink (but to common GPU) and configure video range and color format in the same way it is through decklink. Then calibrate as normal desktop display. Make LUT3D (VCGT on), plug through decklink, measurement report, check if that solves your issues.
=> that means ensure 96+48+24+12 patches iteration for VCGT.
Ok Vincent, I’m going to try it, thank you!
So, in DisplayCal, first I have to calibrate without creation of a LUT3D (practically as I’ve already done with the displayport input), and then make a LUT3D out of the profile I’ve just created?
Do I have to worry about any specific settings (knowing that later I will make a LUT3D with that profile)? Because in the wide gamut displayport calibration I’ve chosen these settings:
– Output levels: Full Range (as on LG OSD)
– Correction: LCD GB-r-LED IPS (Dell U2413)
– Interactive display adjustment enabled
– White point: x 0.3127 y 0.3290
– White level: 100 cd/m2
– Black level: as measured
– Tone curve: gamma 2.4 / relative
– Black level output compensation: 100%
– Black level correction: 0%
– Rate: 4
– Speed: slow
– Profile type: XYZ LUT + matrix
– Default rendering intent: relative colorimetric
– BPC enabled
– Quality: high
– Testchart: auto-optimized
– Patches: 2600Can I go with the same settings and then make a LUT3D out of it?
And if that solve the problem, does this mean every time I have to calibrate for resolve/decklink, I will have to do all this rigmarole (instead of using the displaycal-resolve calman connection)?
And if that solve the problem, does this mean every time I have to calibrate for resolve/decklink, I will have to do all this rigmarole (instead of using the displaycal-resolve calman connection)?
But it does not work, by your own test… hence you have DisplayCAL only.
Regarding settins BPC=on may result in inaccurate near black. Also you do not need such high number of patches for a fast test. Just ensure 11-point cube lut and perfect grayscale.
Okok!
But why isn’t the resolve calman connection working as it should? Is it a bug (by displaycal or resolve)?
Most likely due to usless low quality display with serious issues in grey. Cubic root of 2600 ~ 14 steps. It’s like you calibrated at fastest speed although this assumes uniform patch location across cube.
So try to get smooth neutral grey with full iterations first, then profiling at whatever X number patch size you like most.
Regarding settins BPC=on may result in inaccurate near black.
What contrast ratio is enough to switch it off, so no visible shadow clipping appear, – your opinion?
Profile != LUT3D
Profile can be accurate, including actual black levels, then when making LUT3D choose black output offset in the same way you can choose what to do with white point (relative/abs) or out of gamut colors. But… he can try.
I’ve found out that on LG monitors the black level settings in the OSD is related to levels, high black level is full range, and low black level is video levels. But at this point that confuse me even more. How do I have to set the chain? If I set all to full range or all to video levels (OSD, resolve, displaycal), I have a LUT3D’s verification with very low contrast (300:1) and high black level (0.33 cd/m2), while if I mix it up (full range on displaycal and video levels in the other two), I have the full contrast (800/900:1) and deepest black level possible (0.11 cd/m2), why is that? How am I supposed to set the chain?
From what I’ve read in this thread: https://hub.displaycal.net/forums/topic/data-or-video-levels-davinci-resolve-calibration/ what I’ve chosen in the beginning was fine: full on displaycal and limited in monitor OSD and resolve.
Plug LG to PC through the same input as decklink (but to common GPU) and configure video range and color format in the same way it is through decklink. Then calibrate as normal desktop display. Make LUT3D (VCGT on), plug through decklink, measurement report, check if that solves your issues.
=> that means ensure 96+48+24+12 patches iteration for VCGT.
I’ve test this way and it lead to nothing (see attach)… Really I have no more ideas.
What do you think I should do?
Yeah they don’t match, in order to have white point 0.3127 0.3290 with the LG connected to the decklink I have to set certain RGB values in the OSD, while with the LG connected to GPU (same cable, same LG HDMI input) I have to set different RGB values in the OSD, otherwise during interactive display adjustment the results are different. I have to ignore it and go with the RGB values as they were set with the decklink connected? Also, do I have to choose particular settings in the nvidia control panel? What is the procedure?
Maybe you have not set on “common GPU” control panel the same settings as decklink is sending
Yes, I should. In the nvidia control panel (I have a 1070) I’ve set RGB format (the only option available on the 1070, but the decklink should be the same because I use the HDMI connection, not the SDI one), and 16-235 both in the “change resolution” and “adjust video color settings” tabs (since in davinci I’ve set video levels in video monitoring settings).
The only thing I can’t match I think is the bit depth, decklink has 10bit, 1070 has only 8bit. But you know, that’s exactly the reason why I purchased a decklink.
