- This topic has 17 replies, 3 voices, and was last updated 2 weeks, 6 days ago by
.
-
Posts
-
Hey! I’m doing an ultra low budget grading setup. Creating a 3D LUT for Resolve.
Have an ASUS ProArt PA247CV hooked up to a BMD UltraStudio Monitor 3G. Using a Calibrite Display Pro HL for my colorimeter. I’m decently familiar with DisplayCAL and understand 3D LUT creation.
What I can’t sort out is the low gamut coverage at the end of the calibration of the ASUS. It’s always in the low 80% range. I’ve done the following:
Calibrated with Spectral: White LCD Family (AC, LG, Samsung).
Calibrated with downloaded CCSS and CCMX files for the ASUS.
Tried various modes on the ASUS like Rec.709 Gamma 2.4 mode, Standard Mode and even sRGB mode. I obviously thought Rec.709 Gamma 2.4 was the way to go, so began with that. I get chromaticity coordinates dialed in really well during the interactive part.
Again, every scenario has me in the low 80% gamut coverage. Is there a test I should do after the final prompt?
Thanks for any clarification!
Calibrite Display Pro HL on Amazon Calibrite Display SL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Video card? Check your driver settings. OS? Your color profile should be none but Windows will default to srgb. Disable Windows calibration. Try another tester. HCFR or Lightspace will show the cie charts and can see the gamut coverage.
Hey Ben…GPU has nothing to do with it. I’m using a Blackmagic UltraStudio Monitor 3G. So the GPU is completely bypassed through Resolve and over the BMD hardware.
For anyone looking at this (thank you!) here’s part of the report I get:
Hey Ben…GPU has nothing to do with it. I’m using a Blackmagic UltraStudio Monitor 3G. So the GPU is completely bypassed through Resolve and over the BMD hardware.
GPU video level vs app vs display mismatch (IDNK which one of them, check one by one)… just look that all three RGB primaries are desaturated in the CIE a*b* plot.
Also 500:1 contrast confirm this.Vincent! Thanks for taking time to look at this.
Some of this is over my head, so I’m not questioning your thoughts. Just clarifying for own own sake:
Won’t GPU be totally out since I’m hooked up to the UltraStudio Monitor 3G (similar to a Decklink card)?
I’m looking at the a*b* plot report, but here’s the thing…I’m a very familiar with post, but not display calibration. Been reading and reading and testing and know way more now. But a report like this is still like another language. So all of that to say, I have no idea how to tell what my RGB primaries are doing.
When I’m at the interactive display part, according to DisplayCAL I have everything really close at that point. Contrast on the display is left at the default. I don’t mess with that. Just brightness, Gain and Offset.
This display (ASUS ProArt) is supposed to have a 1000:1 contrast ratio.
So…I just finished a 7 hour detailed scan. Then did a report…if you don’t mind, here’s the link:
https://kylerboudreau.com/film/new-report.html
And here are the screen shots of my settings:
Vincent! Thanks for taking time to look at this.
Some of this is over my head, so I’m not questioning your thoughts. Just clarifying for own own sake:
Won’t GPU be totally out since I’m hooked up to the UltraStudio Monitor 3G (similar to a Decklink card)?
Read my first message this way: somwhere in the whole pipeline, from software to monitor, somebody is telling “this is legal levels”, while monitor expects full range.
As said before you’ll have to check one by one till you find teh culprit. Reading contrast (nits ewhite / nits black) will be the easiest and shortest way to testLUT3D IRIDAS “.cube” should be Full to Full. Monitor is full range (contrast test). After all this is fixed monitor would be >100nits
Test contrast again it shoudl be 800-1000:1 not 500:1Also you should not use any CCMX unless you made them:
“i1 DisplayPro, ColorMunki Display & ASUSTek COMPUTER INC PA247CV by 4KM (i1 Pro 2).ccmx”
If you did not make that CCMX you shoudl stick to suitable CCSS corrections which are portable between i1d3 units, CCMXs are notVincent…now I understand what you’re saying. THANK YOU.
Makes sense. So I’ll check the pipeline. One question:
I’ve read other places that I should be using legal range for my UltraStudio Monitor 3G to display config. However, I can switch it all over to full if that’s better. I guess my real question is this: If my pipeline is legal range all the way through is that cool, or should I be full data all of the way through in your opinion?
And good to know on the CCMX. Didn’t realize that. There is a CCSS file out there for my display. Should I go with that or just use White LED?
Hey Vincent, doing a full data path now from Resolve to Monitor and then checking this in both tabs on DisplayCAL too.
Question: How did you know from what I was showing you that it was asking for full? I’d just like to understand the reports a little more.
Thanks!
Where did you get device link profile? Thats says it is legal range. I do not recall after reading your monitor manual that it is legal or full. Most monitors are full range.
I know nothing about BMD UltraStudio Monitor 3G. I knew the report had to be good srgb coverage and a bad setting causes it to not be. Level mismatch throws everything off.
Thanks Ben. The UltraStudio 3G hooks up to usb-c port and bypasses GPU. DaVinci Resolve sees it and passes video through there out to display.
I’m flipping everything to full so the entire pipeline for sure matches, and trying it again now. Will report back. Much appreciated man!
Content is legal. Your monitor may or may not accept legal, it should be in the specs. Usually it’s unlikely that an entry sRGB monitor will accept legal RGB +24p… but who knows, check. Hence before going to GPU app, still in app logic, by config or guess it will scale contrent to output. It’s a direct scale legal -> full (cliping BTW) so no info is lost.
A typical test is to load as media in your viewer/editor some sample video with bars near black 16 and near 235 white. There should be no clipping 16+ and -235. Also there should be no different content bellow 16 and over 235.
AVSForum should have several vidoe samples for Rec709 16-235. Load ot into your project and check this. Then you can run DisplayCAL profiling + LUT3D + verification.Question: How did you know from what I was showing you that it was asking for full? I’d just like to understand the reports a little more.
a ) CIE a*b* plot showed colored dots (measured) “inside” predicted/expected white holes => measured dots show desaturation from expected values in all RGB boundaries, so this is lileky be caused by 235 limit on a display expecting 255range. Go to the end of your report you’ll see CIE a*b* plot. a*b* plota re used because “2d” distance is more realted to human vision color distance that CIE xy 2D distance which is significatively different.
b) Contrast. Monitor is full range to sending data in 16-235 range will cripple contrast-> 500:1 while display is expected to be 1000:1
This may not hold on mediocre monitors like Benq SW line where due to extremely poor QC in panels and agresive uniformity compensation (so they can be sold without their native severe green-pink tint on sides) it destroys contrast. But for a common While LED display IPS… very unlikely that it is showing 500:1.Thanks Vincent! So I just followed your advice and did a full data range pipeline from Resolve to Display to DisplayCAL and also used the Spectral: White LED and did the calibration again.
And FINALLY got a higher gamut coverage. It’s not as high as I’d like, but I’m guessing this low cost ASUS PA247CV is part of that problem. And just for kicks, I set everything back to Limited again and yeah…the calibration never gets above 83% or less. But full got me to a consistent 88% coverage.
