JAVS V3 CURVES
Alright guys, here are the v3 curves. I have made a zip file containing all three versions now for ease of use. You guys are free to compare them and pick your poison. The v3 curves are a little more radical which I will explain why.
Arve Tool Versions. - Preferred method
JVC Autocal Software Versions
Those of you who are too scared of the Arve Tool, which is by far the preferred method to use these curves, here is a version of the v3 curves which can be loaded using the JVC Autocal software into one of your custom Gamma slots. Obviously they are slightly lower resolution that the Arve tools curves which are 256 point rather than these being 12. But they should be pretty much the same otherwise.
Here are both curves on an overlay, 1200 and 4000 nit.
I actually now set it to clip to 3977 nits, because typing any number above 3977 nits into the Arve tool results in a 4124nit curve I guess due to the slightly limited resolution of the curve creation, so, I would rather clip 23 nits than overshoot the target by 124 nits...
Here are the 4000nit curves compared
v3 vs v2
v3 vs v1
As it turns out, I had remembered back to when I was originally testing the Arve tool against the ST2084 built in gamma in the projector, and kept remembering that I actually preferred the fire in Batman V Superman with the built in ST2084 gamma. So I had a look at this again yesterday, and have now adapted the highlight rolloff in a big way to match the preference I had there, and I think I am really happy with this.
The problem is, we were going too bright generally with the Arve tool. Too bright everywhere above 100 nits. I was actually on the right track with the V1 curves and should have stuck with it.
My UHD Waveform analysis thread has done one thing for me, showed me where most of the real content that matters lies, and also showed me very clearly where all the highlight content begins from, it seems that over 100 nits we are well beginning to enter highlights in all content, really bright blue skies, reflections, fire, explosions etc...
For my 4000nit curve, I start adjusting this highlight rolloff at 150 nits, which is radical, we used to do this closer to 300, so, everything under 150 nits, is completely identical before This content UNDER 150nits, is why we all started using the Arve tool, we had pretty pathetic and unreliable shadow detail with the built in HDR gamma, now that's long been fixed, but in the process IMO, we went too hard with highlight detail, because we could, not giving it room to breathe.
Lets look at a couple waveforms and see first what's generally under 100nits, much more than you think!. First, look at how to read these waveforms.
This film never goes over 200 nits, but its interesting to see what lies under 100 nits here... I certainly had no problem with this films general brightness, again, this content will not be compromised or too dark with the new curves since I dont even touch under this point.
Not much general content above 150 nits there (between 512 and 640), only really content that started getting into highlights. So shadow detail is not going to be compromised with these curves.
Lets look at some really bright stuff, the stuff we want to adjust. Now I feel we were compressing all of this too much, I may be slightly throwing the ST2084 EOTF out the window here in regards to highlights, but I encourage you guys to test this content yourself if you have it. You need to fine true 4000nit content, Batman Vs Superman, Mad Max are two big ones.
This one is interesting, I did think this was much higher in nits, and found I couldn't get this to look good until I lowered where the rolloff begins.
Once that fire looked good to me, as you see above in earlier posts in my crappy samsung pics of fire, then this following stuff looked most excellent. This content hits hard against 4000 nits, and the following battle is just blinding HDR fest at clipping point.
When you test the scenes above yourself if you have them, compare the lightning bolts, and the fire, and the really bright things, you will see the v2 and v1 curves actually mush all this detail too much and its a bit too heavy handed, the v3 curves will allow you to see a more subtle highlight variation, detail INSIDE the fire etc.
Hell, maybe with content that goes over 1000 nits, but doesn't quite slam near 4000, you will want to use my V2 4000 nits curve, in shots where the content only reach 2000 nits for eg, that will indeed look brighter, but where it wont look good is that Batman vs Superman fire shot, which interestingly only goes up just over 1000 nits. So, I encourage testing here guys, hopefully some of you are confident enough now to try and compare a few curves, its all healthy exploration!