Originally Posted by Kris Deering
I do not use the dynamic iris either. It does help contrast, but since I am always evaluating different things (mainly tone mapping lately), I don't want to mistake DI artifacts for something else. For my NX9, I am getting about 30K:1 running wide open with the filter in place for my 140" scope screen (I get about 101 nits full field and 110 nits when measured with a 10% window), but I have a DCR lens.
I think there have been some changes to the filter over the life of the new line (or variances in production). When I got the first NX9 in for review, I was measuring about a 5-7% loss in light with it in line (I praised this in the review). I was also seeing nearly perfect coverage of P3. Since I am a calibrator I have measured A LOT of these since they've come out (and I own a NX9 that is NOT the unit I reviewed). Now I am consistently seeing about a 10-15% loss in light with the filter in place and nearly every unit I measure has a hue shift at green that limits P3 coverage to around 96% (the green shift moves cyan in as well). I have measured ONE exception in the last six months! Earlier units didn't seem to have this issue (they would ship with a good green point but red would be pulled in, but this could be fixed with the JVC calibration software).
I've inquired about this with JVC and have given them quite a few screen grabs of the green hue shift. Not much info has come back about it. I've also requested they add a PROFILE OFF WITH FILTER IN PLACE color mode as I can get the entire gamut back when I do the workaround with a PC to temporarily make this work. I know Manni has been asking for this for A LONG time now and found a work around to make it happen with the Vertex, but I don't have one and don't want to add another HDMI box into my video chain at the moment. That may change if this is never addressed by JVC.
I haven’t tested as many units as you did, but in my experience the brightness loss due to the filter is smaller in high lamp than in low lamp, so the small numbers you mention would fit high lamp, but not low lamp use. I posted all the measurements in the calibration thread. You mention that the new units you evaluated had a filter, so not an explanation, but I think it’s important to clarify this so that those with an early unit using low lamp don’t feel robbed. 5-7% loss only ever applied to high lamp.
Also, the minor red deficiency in P3 at 100% sat in my case (it’s an early unit) isn’t corrected with the Autocal. As long as the unit is properly calibrated, it doesn’t really matter with actual content as very little content reaches the edge of P3.
The profile off with filter on (it’s not the Vertex but Vertex2, Diva and Maestro, details in my sig) helps me get 100% cover in rec-709, but I “only” get 97% of P3, which is fine.
However, besides possibly getting a wider gamut cover, the biggest advantage of profile off when you use a 3D LUT is that there is no gamut compression towards 100% sat, so you can use a 3D LUT even with BT2020 without seeing a lot of posterization at the edge of the gamut. This is by far the biggest advantage, especially if your native gamut covers more than P3.
However, the native gamut (profile off with filter on) tracks P3, not BT2020, so you will need a very large LUT (17x17x17 at minimum, preferably 21x21x21) to make it work.
I’m getting excellent results with profile off / filter on with large LUTs and as a result have stopped using the Autocal (profile off / high bright can’t be autocaled). The Autocal is great when there is no LUT available for some sources, which was my case for all my non-HTPC sources, but the Envy has changed all that.
If you run large verification patches on a JVC color profile, you will see some significant errors due to this gamut compression. JVC needs to fix this, and to deliver a profile off with filter profile ASAP.
workaround is great, but it adds a delay of 5-7 seconds every time the user mode changes (I still use -12 for SDR and iris open for HDR, even if I now use profile off filter on for both), so it’s not ideal. But if you’re using large LUTs, it’s the way to go with the JVCs.
Also, users of the NX5/RS1000 without a filter should definitely use profile off if they use a LUT, as they won’t have the same issue with the filter. The results should be significantly better than with any of the factory profiles, especially with BT2020, though in their case there is no point having a BT2020 calibration (assuming they use a Radiance, Envy or madVR) as they won’t get a larger gamut that way.
For those of you who didn’t understand a word of the above, no worries, the Autocal still works great and gives excellent results if you’re not using a 3D LUT, so keep using it as usual.
I’m only posting this to make sure that JVC delivers a profile off with filter mode in the next f/w revision. We’ve been asking this for ages and there is no reason not to deliver it. Or they need to fix their factory profile and make sure that they are always slightly oversaturated and don’t have that gamut compression at the edge, which makes it difficult to calibrate using 3D LUTs without causing artifacts such as posterization. Some software deal with this better than others, but it’s a cumbersome process.