- This topic has 42 replies, 4 voices, and was last updated 4 months ago by
.
-
Posts
-
Self report is without measurements. Measureemnt report with no special validation configuration is against device profile.
It seems that [display + decklink] (as a set) is behaving in a weird way that cannot be captured with a matrx profile, so you’ll have to choose a XYZLUT (a full 3D mesh capturing display response = number of nodes in such mesh is aprox equal to the number of nodes in cube edge… it scales very fast in stotal node size = patch number size).
Once you get a match in display vs display profile you can make a LUT3D to simulate Rec709, not before.Also… display+decklink is so bad behaved (behavior so poorly predicted by a matrix profile) on isolated primary channel ramps from 0 to 255 … that I’ll test it connected to your macbook’s GPU, just to check if it is display the one behaving so wild.
It may be monitor’s limited 16-235 HDMI input not working properly, it may be HDMI input even un full range but over minidisplayport/displayport in full range mode work as intended (can be matrix profiled and get a profile that matches display).
It may be your decklink IO not properly configured…This has to be tested by you. I mean you can profile with an XYZLUT profile and get an match, but I would test why it is behaving so bad on pure primaries black to priarue ramp and pluggig to a computer an testing HDMI limited vs full vs displayport may get you a hint if what is causing this. But if you test on a computer make sure you have no limited vs full range issues in GPU configuration (maybe it’s easier to do in windows)
Given this situation profile display without calibration and verify it against itself (no simulation profile)
As you may have already seen, I have made other tests/calibrations but I suppose those are wrong.
Could you please elaborate on the sentence above ?
I think I haven’t understood correctly what I should do from the last step that said was “good”.Thank you again Vincent !
The grey ramp seemed good. 000 -> 0 255 0 ramp and 000 -> 255 0 0 ramp are not, aven even profiling to a matrix (idealized response ) with actual primaries it does not fix it… so an idealized matrix cannot describe the behavior of teh whole system [display HDMI limited range + decklink limited range]
… hence trye a non idealized aproach, a 3D mesh with several nodes (XYZLUT profile)I have connected my monitor to and HDMI port on my MacBook Pro without the I/O device and I have made calibration profiling within DisplayCal using the same settings as used before : Rec.709, Gamma 2.4
You can see the result attached below with a better scores.
According to what you said, this looks like Resolve and/or the I/O device are interfering in some way, right ?
I have attached my Resolve settings and the I/O device settings used previously. I don’t see any other setting at the moment.Thanks a lot !
Should I set my monitor on 0-255 ?
It is currently set on “auto-detect” so I suppose it is detecting what is coming from the video signal and adapting its setting to it ?In Resolve, the I/O device is set to Video (16-235).
No, what I’m saying is that it may be wroth the time finding which element is causing black to primariy green / red to behave so badly that a matrix porfile cannot predict its behavior)
A XYZLUT shoudl be able to capture non ideal behaviors, so display can be predicted by a profile and you can attempt to make a LUT3D which will work as inteded
… but WHY is [display HDMI limited range + decklink limited range] behaving so bad n black to primaries ramp? IDNK, i have no access to your system.
It MAY BE (as a guess) that limited range HDMI input is not working as expected…. if that is happeing plugged to macbook as common display through usbc-/tunderbolt to display port input and profiled in a simple way should be able to be described by a matrix profile (that would be the test)
It may be that 2.4 gamma preset in monitor is causing the issue, so it will behave as bad with Displayport and working as a common monitor pluged to macbook. .. but using default setting on gamma (2.2. or whatever), then green or red ramp from black to 255 wont behave so bad (that would be the test)
It may be something different from these samples. Without actual access to the system is guessing, you’d have to test this.
If i were you i’ll try to find it, even with just setting an XYZLUT on your current configuration & then profiling my problems go away.I have connected my monitor to and HDMI port on my MacBook Pro without the I/O device and I have made calibration profiling within DisplayCal using the same settings as used before : Rec.709, Gamma 2.4
You can see the result attached below with a better scores.
According to what you said, this looks like Resolve and/or the I/O device are interfering in some way, right ?
I have attached my Resolve settings and the I/O device settings used previously. I don’t see any other setting at the moment.Thanks a lot !
I do not see messurement report… so i cannot check if black to green/black to red ramp is behaving properly
Sorry, I forgot to send you the measurement report.
MacBook Pro – HDMI – BenQ PD2700U
Profile self check AE*76: average 0.76, maximum 2.00
You can find the profile settings, profile self check, verification settings and measurement report attached below.After that, I have tried making a XYZLUT profile with the UltraStudio box linked, as you suggested :
MacBook Pro – Thunderbolt 3 – UltraStudio 4K Mini (via Resolve) – HDMI – BenQ PD2700U
Profile self check AE*76: average 0.33, maximum 2.29
You can find the profile settings, profile self check, verification settings and measurement report attached below.From both of those tests, can you find any useful data or make any conclusion ?
Let me know if there is anything I have done wrong.
I can provide any information or detail that would be needed.Thank you so much for your time !
Sorry, I forgot to send you the measurement report.
MacBook Pro – HDMI – BenQ PD2700U
Profile self check AE*76: average 0.76, maximum 2.00
You can find the profile settings, profile self check, verification settings and measurement report attached below.Did you use limited range or full? since it was plugged as a computer display. Maybe it’s HDMI limited input that was causing the irregular behavior or green and red primary ramps from black to 100% signal that we saw on previous reports.
After that, I have tried making a XYZLUT profile with the UltraStudio box linked, as you suggested :
MacBook Pro – Thunderbolt 3 – UltraStudio 4K Mini (via Resolve) – HDMI – BenQ PD2700U
Profile self check AE*76: average 0.33, maximum 2.29
You can find the profile settings, profile self check, verification settings and measurement report attached below.From both of those tests, can you find any useful data or make any conclusion ?
Let me know if there is anything I have done wrong.
I can provide any information or detail that would be needed.Thank you so much for your time !
That profile matches display (It captures all irregular behavior from [display +decklink] system working as a single part), although there are some outliers (IDNK the cause, RTC from display? IDNK . I did not “see” it)
Now run LUT3D creator, DisplayCAL-3DLUT-maker.
Source colorspace: Rec709 g 2.4 blck out 100%
Destination profile: this custom XYZLUT profile.
Iridas cube format. There is no VCGT ofr a decklink monitor, so…
Copy .cube format to Resolve folder and use it, fro example:
If there is no LUT3D active you can use device link (ICC equivalent of a LUT3D) in your measurement report verification.Usually people do not need this workaround like you did, but in your situation [PD2700U + UltraStudio 4K Mini] as a system, as a whole, as a “black box”, had a very weird behavior in “black to 100% saturation primaries” ramp… hence it could not be captured with a matrix profile.
But when plugged standalone to computer… it could be described with a matrix single curve without issue as you see.
If when plugged to computer through HDMI you use HDMI full range, my guess, and it’s a guess since i have no acces to your HW, is that HDMI limited input in your PD2700U was acting a bit weird…
Capturing weird behavior thorugh densely populated 3D mesh is XZYLUT expertise…Anyway, make LUT3D , load it to Resolve when you have all validated and make some visual test with “gradients” in video footage. It display behavior was very weird you may see some banding actifacts (due to mesh resolution & truncation errors in calculation) that will not show vy playig that video file on your mackbook.
It should not… but these irregular behavior soemtimes are difficult to capture.Did you use limited range or full? since it was plugged as a computer display. Maybe it’s HDMI limited input that was causing the irregular behavior or green and red primary ramps from black to 100% signal that we saw on previous reports.
I don’t know since it was set to auto in the BenQ and DisplayCal settings…
That profile matches display (It captures all irregular behavior from [display +decklink] system working as a single part), although there are some outliers (IDNK the cause, RTC from display? IDNK . I did not “see” it)
Now run LUT3D creator, DisplayCAL-3DLUT-maker.
Source colorspace: Rec709 g 2.4 blck out 100%
Destination profile: this custom XYZLUT profile.
Iridas cube format. There is no VCGT ofr a decklink monitor, so…
Copy .cube format to Resolve folder and use it:So I have exported a 3D LUT with the following settings : cf. screenshot attached
Usually people do not need this workaround like you did, but in your situation [PD2700U + UltraStudio 4K Mini] as a system, as a black box, had a very weird behavior in “0 to 255 primaries” ramp… hence it could not be captured with a matrix profile.
But wehn plugged standalone to computer… it could be described with a matrix single curve without issue as you see.
If when plugged to computer through HDMI you use HDMI full range, my guess, and it’s a guess since i have no acces to your HW, is that HDMI limited input in your PD2700U was acting a bit weird…
Capturing weird behavior thorugh densely populated 3D mesh is XZYLUT expertise…Indeed, this is weird.
Do you think it’d be important to dig a bit more on this issue ?
I have provided you all the informations I have and, as far as my limited knowledge goes, I don’t really know what to do next about this issue.
Maybe checking the [MacBook Pro – HDMI – BenQ PD2700U] video range can help ?
If so, what should I test or set to get to a conclusion to move further ?Thank you again Vincent, sincerely !
Anyway, make LUT3D , load it to Resolve when you have all validated and make some visual test with “gradients” in video footage. It display behavior was very weird you may see some banding actifacts (due to mesh resolution & truncation errors in calculation) that will not show vy playig that video file on your mackbook.
It should not… but these irregular behavior soemtimes are difficult to capture.I have made a 3D LUT and verified it (measurement report attached below).
After applying it to the video monitor output within Resolve, I don’t see any banding or weird behaviour at first glance.By the way, correct me if I am wrong but the DisplayCal preset setting for Resolve is also using XYZ LUT + Matrix for profiling.
However, the result is way off compared to the current “manual” XYZ LUT + Matrix profile calibration…Ok, there is some weird behavior near full saturation patches, as in profile. If profile does not match display (like those outliers near full saturation of one channel), LUT3D accuracy (which depends on same profile accuracy) will fail in the same places.
I wonder if it is due to bad behavior of your monitor HDMI limited input scaled in a bad way by monitor firmware to full renge bafore sen data to display panel.
IDNK if you can test this with a macbook. On windows you could try with HCFR and tweaking HDMI output configuration … or with a raspi pattern generator at full range sending it video range encoded patches (hence raspberry pi acting only as a full range passthrough: monitor limited <- raspipi full <- computer app limited).
I mean maybe that PD2700U is not suitable to be used with limited range… but you need to test it. If that happened IDNK if your UltraStudio 4K Mini can do the job of scalling or let resolve do this. Ask in LGG forums about its capabilities.Do you think that the fact of launching the calibration via a Resolve project that is already set up with nodes and effects may interfere in the patches and overall calibration ?
Maybe should I test profiling again from a new empty project ?
Do you think that the fact of launching the calibration via a Resolve project that is already set up with nodes and effects may interfere in the patches and overall calibration ?
Maybe should I test profiling again from a new empty project ?
I did not know that you have done profiling that way. All off or new project.
What you need to profile is Rec709 content directly mapped non color managed to display response.
Your effects apply to content (encoded in content), not to visualization.
