OK folks, here we go... Let's call this In-Depth Investigation Round 1
I have carried out a load of testing including measurements to find out what the hell is going on, using a brand new JVC RS3000/NX9, March 2019 build, with firmware 2.04; and here's what I have discovered.
Using a 1% ADL test pattern I succeeded in replicating the DI yellowing that has been reported by some of you fine gentlemen; wherein, this is SDR - IRIS 0 - HIGH LAMP - WITHOUT DI
And this is the corresponding measurement of the white, where I had accurately calibrated the projector hence the perfect white balance:
But this is what happens when the DI is enabled, being SDR - IRIS 0 - HIGH LAMP - WITH DI ENABLED
As you can see there is very noticable yellowing when the DI is enabled.
Switching between AUTO1 and AUTO2 there is absolutely no noticable difference. The yellowing is exactly the same, as is shown by the respective measurements.
This is SDR - IRIS 0 - HIGH LAMP - WITH DI SET TO AUTO1
And this is SDR - IRIS 0 - HIGH LAMP - WITH DI SET TO AUTO2
Therefore, from this we can conclude that there really is absolutely no difference between AUTO1 and AUTO2 settings for the DI with respect to this phenomenon.
Interestingly, although the appearance is yellowing there is in fact more green push than red; and the overall shift is very significant, skewing the white balance completely out of whack to 17.4% dE inaccuracy. Not good.
So then I measured the same, but with respect to LOW LAMP instead of HIGH LAMP, and here are the results... Firstly, SDR - IRIS 0 - LOW LAMP - WITHOUT DI:
where again I accurately calibrated this so perfect grayscale:
And then SDR - IRIS 0 - LOW LAMP - WITH DI ENABLED:
Hence, LOW LAMP is very similar to HIGH LAMP in this regard, with slightly less red push than HIGH LAMP, but skewing the white balance to 18% dE inaccuracy; so very similar.
Being solution oriented I then went about evaluating what settings changes improves the issue. In short, I tried absolutely everything, and the only thing that makes a difference is closing down the lamp iris. So, I investigated including measurements with respect to what's what in this regard and here are the results:
IRIS 7 - SDR - WITH DI ENABLED:
IRIS 8 - SDR - WITH DI ENABLED:
IRIS 9 - SDR - WITH DI ENABLED:
IRIS 10 - SDR - WITH DI ENABLED:
Therefore, closing down the IRIS does improve matters, wherein the dE inaccuracy improves from 17.4 dE to 16.4 dE (IRIS 7), 14.0 dE (IRIS 8), 10.8 dE (IRIS 9), and 5.0 dE (IRIS 10). Where I consider 5.0 dE to be acceptable, in that this will no longer look yellow, but white, as you can yourself see here in this photo of SDR - IRIS 10 - WITH DI ENABLED:
Looks considerably better than THIS doesn't it?:
Consequently, the good news is that as far as SDR is concerned, there is a workaround, namely set the LAMP IRIS to 10 - 15 and you won't be seeing any DI yellowing
Serendipitously, in setting up the projector to achieve 14 - 18 fL peak image luminance with SDR I would be setting the LAMP IRIS to within this range anyway.
Either way, I consider this to be good news, because it means that with SDR one can most certainly use the DI and avoid the DI yellowing issue by simply setting the LAMP IRIS to 10 - 15 before enabling the DI by setting it to either AUTO1 or AUTO2.
Of course, I will need to confirm that other units behave the same way as this one, where as it just so happens I have a second brand new JVC RS3000/NX9 here which I will shortly similarly evaluate and will confirm back whether or not it does indeed behave the same way.
OK, that's SDR, so let's move onto HDR
Well, here's where things get even more interesting...
...Because here's LUCY:
HDR - IRIS 0 - WITHOUT DI:
HDR - IRIS 0 - WITH DI ENABLED:
And here's the corresponding measurements of the white:
WITH DI ENABLED:
Yup. NO YELLOWING ! In fact, there's no significant skewing of the white balance whatsoever
Digging deeper, here's what happens with 1% ADL:
HDR - 1% ADL - IRIS 0 - WITHOUT DI ENABLED:
HDR - 1% ADL - IRIS 0 - WITH DI ENABLED:
So, with 1% ADL with the IRIS set to 0 wide open there is some color shift, but it's nothing like as bad as with SDR, being only circa 8% dE inaccuracy as compared with SDR's circa 18% dE. Also, interestingly the shift is not towards YELLOW but towards CYAN, the consequence of which is that it is far less noticable, as you can clearly see from the photo above of HDR - 1% ADL - IRIS 0 - WITH DI ENABLED
Strangely, when closing down the LAMP IRIS, unlike with SDR, this color shift remains essentially constant until when then IRIS is set to 10 or higher.
Here's IRIS 9:
And here's IRIS 10:
Wherein, it drops below 5 dE inaccuracy at IRIS 10 the same as with SDR.
Therefore, with HDR, with torture test content such as LUCY, there is absolutely zero colour shifting. With 1% ADL test patterns there is some colour shifting but towards CYAN, not YELLOW, and comparatively only slightly to the extent that I think most people would be hard pressed to notice it. This can be improved by closing the LAMP IRIS down to 10, however, because we need higher light output with HDR I can’t see many people actually doing this. All things considered, personally I would choose to leave the IRIS wide open set to 0 and live with the slight CYAN push of only 8% dE in the rare instances that it actually manifests.
So that’s the what’s what in this regard with respect to this particular JVC RS3000/NX9 unit. Like I said, I will check out another unit and see how the performance compares and will report back my findings. I also have two shiny new JVC RS1000/N5s here as well and I will test these as well.