I have a JVC RS500 that is powered by my MadVR HTPC. I am wanting to create a 3DLUT for MadVR to use for Rec. 709.
I have an i1pro2 and an i1d3.
I ran a cube-based characterization with 7 cubes on both devices pointed at the same spot on the projector screen.
I exported both profiles as .bpd
I then ran a 21 cube characterization with the i1d3.
The weird thing that isn't making sense to me is that if I just make a 3DLUT out of the i1d3's characterization, the results are "ok," but if I apply probe matching, the 3DLUT is terrible, only using about 15% of the gamut even though the green and red are WAY beyond the Rec 709 standards (blue is pretty close to expected value).
I'm attaching my individual files in case that helps make sense of my conundrum.

Appreciate any tips/guidance.

Attached file: JVC RS500 - i1d3 - Rec709 - .zip

The i1Pro2 is not good at low-light readings.
So with a 7^3 profile there will be a lot of 'bad' readings.
You can see this if you look at the profile graphs, and if you make a LUT from the i1D3 Match and i1Pro2 Match profiles directly.
(Just look ta the difference in black readings between the two - 0.01 nits and 0.2 nits...)

Stick to the normal FCVM probe matching with the i1Pro2 and a display that has very low blacks/shadows.
The JVC also has very bad non-linear volumetric errors, which further compounds the issues... you can see this by comparing the profile to a colour space Extracted from the profile.

There is also a bad Drift reading in the 21^3 profile, so do not use that when making the LUT!
(Even though the profile name says 'No Drift' there is Drift data.)

Steve

Ok thanks! Couple follow up questions:

Is is there anything I can do about the nonlinear volumetric errors, or is that just a defect of the JVC projector? Or maybe a 3D lut correct for these errors?

And just to make sure I'm understanding I should make a 21 cube characterization with no drift selected or checked on the graphing box correct?

The LUT Calibration will fix the non-linear issues, as best they can be fixed.
Unfortunately, the colour issues with JVC projectors are well know, and JVC show no interest in fixing them.

And you should indeed use Drift when profiling!
I was pointing out there was a bad reading in the Drift data on your existing profile.
That may also mean there were bad reading is the actual profile data, but I didn't check for that.
(With the upcoming Point Edit for the higher license levels you will be able to easily correct such errors.)

Steve

Got it. Thanks. Can you tell me how you were able to see there was bad drift data in the 21^3 characterization? Do you just look at the CIE xy graph for irregularities?

I looked at the Drift graph...

Steve

Ok. Same setup as above. I did as you suggested and created profiles matching my i1d3 to my i1pro2 based solely on the RGBW in the probe matching dialog box. I've included both those, as well as final versions of the 21^3 cube characterization (with and without matching applied). The final 3DLUT that is created from the matched 21^3 profile is much better than previously (uses 98% of gamut instead of 35%), but I see a couple "smudges" (not sure what else to call them) in blue and orange area of the LUT.

What do these smudges indicate? That is, what--if anything--can I do to fix this error? I haven't tried applying this LUT to my madvr yet, but I'm guessing it wouldn't look great (maybe I'm wrong?).

Attached file: 2nd Attempt.zip

The profile has a lot of issues... you can see these by referencing the profile to a colour space extracted from itself.
These are then reflected in the LUT.
You can even see this by making a LUT targeting the 'Extracted' colour space.
If the uncalibrated display and the profile of it was 'good', the LUT would basically be perfect.

So, ignore any Probe Matching for now, and workout why the basic i1d3 profile is bad.
Is it really the projector setup pre-calibration, or is something going wrong during profiling?
Maybe more Extra Delay is needed?
Or possibly the i1d3 settings are not set correctly?

Steve

I think I have found another issue - the projector seems to be pre-calibrated (very badly) to a P3 colour space, not set to its native, uncalibrated, state.
That is not the main issue, and is not responsible for the bad readings, but it will make the calibration process more difficult that it should be, as you have a lot of compresses edge colour artefacts due to this bad pre-calibration.

Steve

Ok, thanks. I'll look into this further. I definitely attempted to follow the directions for 3D lut creation and attempted to find the best native color space but turned off all of the color correction, filters, etc. Then I did a few brightness and contrast test to set the brightness and contrast, and adjusted some of the RGB gain an offset to get a decent white balance, but other than that I feel like it's pretty uncalibrated.

Just to confirm the settings I ought to be using for the i1d3 with a JVC RS500:

I'm using a Pi4 to generate Ted's patches.
I'm reading off screen about 2-3 feet, slightly pointed at an upward angle to ensure no shadow interference
Integration Time = .75
Period
Int. Int. Checked
Extra delay = 1.00

Look good?

Yep looks ok - although the Extra Delay can likely be shorter.
Is there any light contamination in the room?

Steve

Steve,
Thanks for all the assistance; I think I'm getting a hang of things now. I've gone ahead and disassembled the JVC, cleaned out the optical block, lenses, mirrors, etc., and ran an Autocal to fix my gamma droop. Then I ran a 7-cube profile of Rec709 colorspace, which seemed to work. I then probe matched the i1d3 to my i1pro2. Here's my question: why does the 3dlut generated from the probe matched profile got a pink hue across the entire top portion (should be white)? The non-probe-matched 3dlut that I can generate doesn't have this, but does have some smudging near the top of yellow band.

I also wonder why I'm only getting 99% of the gamut coverage when all my primary and secondary colors are outside the Rec 709 color space, which I thought would ensure I got 100% coverage.

Attached file: 7Cube-Rec709 Profile with i1d3

If You look at the profile, and Enable/Disable Probe matching you can see the difference between the i1D3 and the i1Pro2 - especially in Red.
So the probe matching is doing exactly as it is being asked to do.

And if you look at the 3D Graphs, especially the Cube view, you can see there are volumetric colours that are not reaching Rec709.
Hence the projector cannot make full Rec709.

Steve

So does the red tint across the probe matched 3dlut indicate that the i1pro2 might have a calibration issue or is it that there's a red tint to the white light of the projector that the better probe is picking up?