Hi all,
I would like to better understand what is going on with my EOTF Diff at the high end on this JVC DLA RS540 and if it's something to be concerned with. LUT is uploaded to a Radiance PRO.

On Diff EOTF, everything looks good (I think) except when we start to go over 80% towards white where things get a little wonky with blue. Clip looks ok though (again, I think). Gamma point here is also smooth, though there is a spike at > 95%. I thought I read an explanation for this once on AVSForum, but I can't seem to find it now.

Is this something to be concerned with?

RS540 Diff EOTF

RS540 Clip

Disregard this. There was just a bit of excess blue at these points causing my gamma point blue to drop here and manifesting on the diff chart this way.

Indeed - if you look at the RGB Balance graph you would see the white point colour temp would be wrong there too.

Steve

I made a few changes and new results are attached. This was more academic than anything as I am not really sure it makes a difference on actual content. I just wanted to get familiar with the graphs as they are a bit more detailed than LightSpace.

Full transparency, I had to lower blue manually in the Radiance Pro at 90 & 95 which I know is frowned upon (changing the LUTs manually after they are loaded). While not a best practice, I don't see any adverse effects on my verification runs.

You would need to post more data for any valid assessment to be made - CIE graphs, etc...
(Or ideally, the actual profile itself)

Steve

Thank you Steve.

I was just fixating on the Diff EOTF because of how much it deviated near 100%. I reloaded the LUT the other night and did a 1000 pt. verification (attached). Would love your feedback.

My grayscale could be a bit better below 30%, and my RGB separation is doing something near white, but overall I think I am happy with these results. I would like to post them on my website as an example of what the Radiance + Colourspace can do for a front projector (which normally can be very challenging).

Thanks again!
John

Attached file: 1000 pt. Verification RS540

Hi John - the results are not bad, based on the overall points measurements.
But, there are some rather large gamut excursions that I would like to understand.
(They are still mostly with acceptable tolerance...)
It could be general instability in the projector?
You should include 'Drift' when profiling, so you get an idea about that, and so the data can be used when generating a LUT.

There is a 'lift' in the gamma around 75%
And the RGB Separation is likely associated with the blue gamma issue in highlights?
Also, the 'large blue' 'back-light' contamination will cause issue as you get closer to the black level.
(You'll need to use a large patch set when doing the initial pre-calibration profile.)

Basically, with ColourSpace I would like to see better...

Steve

Thanks Steve.

I am running a fresh profile this weekend. So I will let you know.

And yes, even after a good 1 hour warm up the projector drifts to red over the next 45 minutes. So at 1:45 there is still some stabilization. I love projectors and the JVC line, but calibrating them is like herding cats.

I will keep trying.

Have added more info to my post...

Steve

I usually do a 17^3 to match the Radiance Pro capabilities. You would recommend higher?

Yes indeed, as the profiling size has nothing to do with the generated LUT size.

Steve

Hi Steve, so here is my latest run. I think there are some improvements.

First some background which I probably should have provided before....

Projector is a JVC DLA-RS540 with about 500 hours on the lamp. Projector is warmed up for a minimum of 1 hour before any profiling or verification.
Profiling meter is a Klein K10A probe matched through CS to a Jeti Specbos 1211
I am using one of the JVC filter modes (Reference mode in this case) to get a wider gamut that I can calibrate down to Rec. 709.
Final LUT is uploaded to a Radiance Pro. I am also using the Radiance to generate YCbCr 422 patches for profiling.
I used Map Space in this case for the LUT, though Peak Chroma produces almost identical results.

I ran the 21 cube profile as you suggested with a drift of 50.

Attached is the 1000pt. verification for this run as well as my grey only verification.

The Clip, EOTF, Diff EOTF and RGB Separation all look improved. RGB balance is nicer as well. I am still seeing points out of gamut around green and along green-cyan, but I think this is just a result of how this filter mode calibrates. I have always noticed some oddities with green so perhaps it is related. For example, no matter how precise my probe match is, I will start to see my K10 measurements on green deviate from the Jeti as I go even just slightly lower in luminance. I don't see this on Red or Blue when verifying the probe match.

Thanks for taking the time to look at this. I want to post these results to the web site I am building, so I would like to know how you would grade this calibration from a professional standpoint.

Best,
John

Attached file: RS540 Profiles

Hi John - you have not included the original 21^3 pre-cal profile...
But the verification looks ok!

Steve

Sorry Steve, here it is. I mistakenly called it "21-pt." instead of cube.

Thanks for looking at this!

Attached file: RS540 21 Cube Profile

Adding a Grey Large verification that I did last night. I really like using the Grey Large with the Radiance because it helps to show how well the Radiance 1D LUT performs. Since most of the points in this profile are "between" the adjustable points in the Radiance, it shows how nicely the parametric greyscale/gamma is working in the Radiance.

Attached file: RS540 Grey Large