I'm playing with the new JVC color profile off with the filter in place (wide mode) and the display is not responding as if the color profile is in native mode at all. I have large RGB separation that starts around 30% and gets worse as it gets closer to white. Is anyone having success with profile off in wide mode? If so, what settings are you using.

There are also three custom tone maps that you can load with the JVC software. The NX9 does not seem to have a gamma off mode so I loaded in gammas of 2.3, 2.4, and 2.5. 2.5 seems to be the closest to 2.4 so I have been using it.

Any ideas or hints would be great.

Thanks,
craigr

If the RGB separation is bad, are you sure of what the pattern generator is sending? Is it possible what you're sending is tagged as HDR and / or BT2020, and what you are seeing is the result of the tone mapping causing RGB separation issues?

It might be worth speaking with Kris Deering though as he did report what appeared to be some brokenness in the implementation of profile off filter on when he tried to use on on NX9:
https://www.avsforum.com/threads/official-jvc-rs3000-nx9-jvc-rs2000-nx7-n7-jvc-rs1000-nx5-n5-owners-thread.3038288/page-1669#post-60261335

I use profile off on my X7900 with the filter via a bit of a hack, it works well on that and is the only consistent way I've found of getting good 3DLUTs on the X7900 with the filter in place.

Thanks for your response. Unfortunately, this has nothing to do with tone mapping. I use both the Six-G and Lumagen pattern generators. I know how to disable tone mapping to send a 2.4 gamma curve to allow the Lumagen to do all tone mapping after calibration. It's always easy to make sure that there is no tone mapping too; just run a quick grey profile and look at the gamma curve.

Kris is a good friend of mine and we talk on the phone often, sometimes multiple times daily. He's one of the few other calibrators in the USA who I recommend, but he's not doing LUT calibration on JVC projectors. In his opinion, the JVC's calibrate well enough on their own, and since he uses Calman he has no way to do LUT's that have undersaturated color gamuts. On this point he and I highly disagree.

Your point of using the bit hack to apply the filter to the native JVC mode would be an interesting experiment on the NX9. I haven't played with the native mode without filter much on the NX9, so I am not sure if it's truly native or not (I tried it with the bit hack about a year ago and still did not get good results). I don't think it's really native on the NX9/7, but I will check. The trouble is that I have to set these up for clients, so using the filter hack is not a good option for me as a pro to use on client machines; it's just too complicated. However, on the NX9/7's I usually use a custom profile I load that I call CIRec709F. It's basically a BT.2020 profile with the filter engaged and then I use the Lumagen to dial in a nice Rec709 LUT to get great results for Rec709. Without the filter, the NX9/7 is almost always undersaturated in green on Rec709

Aside from just the bad RGB separation there is also terrible nonlinearity in the gamut that can not occur in nature. There is still an active CMS. Also, attached to this post is a photo I took of the screen. It shows me feeding the projector YCbCr 422 with the projector forced to YCbCr444. As you can see the grey steps pattern is really messed up. There should absolutely not be any blue steps in the pattern; all the grey steps should be the same color.

Because of this, I have been considering modifying the NX9/7's so that the filter is simply always engaged and never gets moved out of the light path. That way I could use the native mode with no filter and just always have the filter engaged. This way my clients wouldn't have to worry about the hack.

I was already planning to test Native mode with no filter, but your experience has certainly pushed me along in that direction. I will give it a try today.

Thanks and kind regards,
craigr

JVC forced space

CIR Engineering
I guess the thousand dollar question is do you see perfect RGB separation when using profile off without the filter engaged? If you do, and the mere act of enabling the profile off filter on option that is built-in causes the RGB separation to break (all other things being equal) then it's fair to conclude the mode is broken.

I came up with the hack method of enabling the filter on the X7900, and I agree it's likely not really a good option for use outside of an environment where the owner/operator knows what they're doing. But it would be, if nothing else, in an investigative mode, be an interesting data point. If you get the expected results in the projector with profile off, filter off, and then profile off, filter off - but with the filter cunningly slid in manually - then clearly there is some internal projector CMS disaster going on, and maybe JVC could address.

I think JVC did make some changes to the ordering of where "profile off" sits in the projector. On the X7900, the gamma autocal for droop that you may have done have no effect on Profile Off, but I was informed by someone (can't remember who) that Autocal does now correct the gamma droop even in profile off mode. Which would be annoying if there is no longer a way to get to the raw panel gamma over HDMI, and might well hint that it is possible other aspects of the autocal banks (which I guess are populated at the factory with the factory calibration) are still somehow in play when using Profile off on these projectors.

Colour autocal on these has now 4 banks (all the combos filter on / off and lamp hi / lo). I wonder if these are still in play somehow in profile off filter on.

I run profile off filter on for both REC709 and REC2020 (well, SDRP3 as I use in the Lumagen). And it works great to sort out the lack of REC709 green that these seem to have.

I think the LUT / No-LUT on these is a bit of a split decision. On the one hand, you can't argue with the absolute accuracy of a good 3DLUT on these, I've had unfeasibly low errors on large verification profiles. On the other hand, the built in CMS works via continuous transformations of the pixel values, and if you use a Lumagen as your LUT holder there are always it seems some unavoidable cyclic matters which you'll observe in one or other of RGB ramps after the LUT is applied. I always come down on the side of running with the LUT, as I prefer the accuracy and control, and being able to re-target the display using the profile data. Plus I'm yet to see any of the things that you can note in the pure ramps in actual content. But I can see it can be argued either way.

Well, it looks like there is just no native mode in the NX9. Here are my RGB separation and gamma graphs. It seems that the gamma is OK, but how does the RGB separation compare with your JVC? This is profile off without the filter. The LUT that is generated from the data is also 'interesting.' Here is a LUT from a 17^3. It can't look like that without a CMS being applied

I've been talking with Rob Budde of JVC and he agrees that there is an inconsistency. His advice was to redo the autocal so that the projector can setup it's default gamma. Not surprisingly this didn't really help in a 'profile off' setup.

Sigh...

craigr

Here is a LUT made with the regular old concactination method using a desaturation matrix. It's so frustrating to have to do that and lose some saturation!

craigr

CIR Engineering
The old faithful X7900 in profile off with filter forced into place, and also without the filter, gives a basically perfectly straight RGB separation graph. The only real difference with the filter in place is the increased gamut.
Looks like profile off is broken to me if that really is profile off without the filter.

Sigh indeed.

I've never liked that concatenation method for the reason you mention - you have to bring the gamut in way too much to get rid of all the artifacts, and unfortunately without it the LUTs are shocking due to the way the output gamut compression disagrees with the LS/CS engine. That's why I sought out how to get the filter loaded with profile off on the the X7900.

Very annoying if profile off is that broken on the N series as I had been eyeing them for an upgrade at some point.