Originally Posted by CINERAMAX
We loaded an S shape gamma (based on a 2.2) and voila DEPTH OUT OF THE WAZOO.
It isn't my goal to rain on your parade, but I think what you found is more proof of why high on/off CR matters, and looks to me like you basically robbed Peter to pay Paul. That is, you had a certain amount to spend between 0 IRE and 100 IRE (like 6000:1 on/off CR) and decided to spend it in the mid-range. I would like to see the gamma curve to see where the bends are. If you are running Blu-ray, DVD, or other consumer sources like that, then they weren't mastered for S-shaped gamma curves. But you can use one and make many images look better and have more depth, because those images would now have more CR. But, it would steal CR from some other images. Kind of like SOWK finding that going to a high gamma number with his Marantz 11S2 added a lot of depth to many brighter mixed images, but then made many dark images look worse than a 2.2 gamma. With the S-shaped curve, somewhat similar to spending the vast majority of $6k between 10 IRE and 90 IRE, but then not having much left over for CRs between things outside that range.
Picture the curve that was used to master the material on a CRT monitor. There is some debate amongst experts, but it would be in the range of a 2.2 to 2.5 gamma. Charles Poynton told a friend of mine around 2.4 or 2.45 (I don't remember which). Now if you take that gamma curve and pull 10 IRE down toward 0 IRE and 90 IRE up toward 100 IRE you will end up with less CR in images with 10 IRE and under or with 90 IRE and up. So, less shadow detail and less highlight detail. I wouldn't expect good things overall in dark images. Say the 2nd half of chapter 7 of the 2nd Alien vs Predator
movie. The black level is already raised from having 6k:1 on/off CR or less and now the 10 IRE level has been lowered even more by using an S-shaped gamma curve and borrowing from down there, crushing the range between 0 IRE and 10 IRE. Or maybe it was things under 20 IRE or 30 IRE where CR was borrowed from. Either way, it has to be borrowed from someplace for you to apply it elsewhere in the video encoding range.
I'm guessing you'll argue against this, but please try it if you can. Find some of those dark images like I just mentioned and try them with a 2.2, a 2.4, and your S-shaped gamma curve.
If you had tons of on/off CR you could still have the CR you want between 90 IRE and 10 IRE (or 70 IRE and 30 IRE) and still have a lot left over to use between 0 IRE and 10 IRE (or 0 IRE and 30 IRE). I know you told me recently that I would find that on/off CR was unimportant once it got over the 3500:1 range, but I think you should see why it still matters by the tests you are running and by then looking at some dark images where everything is under 10 IRE (but not just a blackout) and see what is going on down there. If you want the depth you are seeing with your S-shaped curve and also want the dark images to look as good as possible, you should want much higher on/off CRs so that you can have that CR in the mid-range and also at the low end.
As an example, I would expect Wormtongue to look more like a head on a black hole in the scene I described in the More Test Scenes
by going to an S-shaped gamma curve compared to the numbered gamma curve it is based on. Although I would like to see the 2 curves together to see where the deviations were made.