I promised I'd provide some more information and thoughts on the data gathered last night, so here it is. Since the table of data is on the previous page, I'm including it here again for easy reference:
And to give an idea of the test setup, here is a photo of the colourimeter in place. I tried to keep the colourimeter as horizontal as possible to avoid any influences that might be caused by any 'viewing cone' being present.
First off, I'm very impressed by what is shown here. SF mixes do seem to perform exactly as intended - as ambient light is increased, more of the black level is maintained, without sacrificing the white point. All of this is done with very little change to the light response; gamma is preserved, colour saturations preserved, the image looks great! This is a remarkable achievement; PB_Maxxx and Mississippi man should be very proud of their work.
Obviously there is a trade off, or the suggestion to simply use high value SF blends would always be given. I only have data from two levels of SF, so my conclusions here are more hypothesis than fact. It seems that as you increase the value of the SF blend, the peak brightness takes a very slight hit, while the black level is even better maintained. The trade off here is increased sparkles, and this really should come as no surprise. The darker grey improves black, and the increased reflective components compensates in order to retain peak brightness. The more reflective components, the greater the sparkles.
As a more subjective test, I also got my wife to view some different pieces of content in darkness, and with ambient light. She preferred the overall look of the SF 4.0 blend, and I have to admit that with ambient light present, this was clearly the better performing test panel, both subjectively and with measured data. The difference between 3.0 and 4.0 is very subtle, with only a slight improvement in dark areas. Our eyes are more sensitive to this however, so perceptually the difference is larger than the data would suggest. The BOC with ambient light was barely watchable, so both SF blends win out here of course.
Testing in a dark environment with white walls was much more challenging. It seems that the difference between the SF blends becomes more obvious as the ambient light level is increased, resulting in an almost identical response between the 3.0 and 4.0 blends in the dark. I personally found the sparkly nature of the 3.0 blend distracting, and the 4.0 blend objectionable in the dark (they were both great with ambient light), and this may be because of my relatively close viewing distance (~90" / 7' 6"). I'd be very interested to see if moving to a lower SF blend maintains the ANSI contrast ratio improvement, and reduces the sparkly appearance.
I've received a PM asking for more information on any colour shifts created as a result of moving to SF, and I'm happy to provide additional information here as well. I do first need to point out that when using a colourimeter rather than a spectrophotometer, the accuracy of the colour measurements is dependent on the spectral output of the projector, and a conversion matrix or spectral sample should be created with a spectrophotometer before truly depending on the data from the colourimeter. Luminance is less affected by this, so I focussed on that for the data shared so far. The data below does have value, but only as relative, rather than absolute. With that said:
BOC RGB Levels
SF v2.5.2 3.0 RGB Levels
SF v2.5.2 4.0 RGB Levels
As you can see, there is a very slight blue push, which in practice raises the white point by about 400o
K. This could easily be corrected with the projector white balance controls, so I'm not very concerned about this.