Originally Posted by odyssey The problem with getting an accurate intra-image CR measurement with CRT projectors is eliminating the effect of the room. This is hard to do and more of a challenge than single lens units. It becomes easier if you measure each primary individually using the same techniques as single lens units.
Yes I understand and you're right, this is something that has to be guarded against even with a single lens projector but it's much more difficult with a CRT because of the physical separation of each gun and also the glow of the phosphor that is induced by ambient light that is discussed later in this post.
I don’t know which test pattern you are currently using to measure intra-image CR,
The patterns are posted at the beginning of this thread. There are two sets of patterns the first set was developed by Wm Phelps and is designed to measure intra-image contrast using varying sized full white rectangles which sweep the APL region from very low to ANSI. I developed a second set designed to measure intra-image contrast using varying grey whites with a constant geometry. This second set allows direct comparison of projectors that use dynamic contrast mechanisms.
but I expect the performance of the G90 to be relatively poor. There are many internal reflections in even the best LC CRT projectors. The halation effect is strong and there is only so much you can do with anti-halation coatings. The phosphors emit light in all directions and the lenses have much lower contrast performance compared to the best DLP projector lenses.
Yes this is exactly what makes this experiment so interesting. It seems as though a CRT would have many problems with intra-image contrast, but real world empirical tests seem to suggest there is more going on. As an example if we take a high native digital like an RS1 and project something like a star field image on it and compare that to a DI equipped projector like the VW50 the RS1 has noticeably more contrast in that scene. The reason for this is because that image requires more intra-image contrast than the DI can provide. If we take that same image and display it on a RS1 and a CRT, the CRT seems more contrasty. This suggests to me that the CRT is delivering more than the intra-image dynamic range of a DI and also an RS1 with that particular scene which is on the order of many thousands to 1. So this is the heart of what I find most intriguing about this experiment.
The ANSI CR and MTF performance give a good indication of what will happen with an intra-image contrast pattern that more closely resembles typical content.
Yes but only at high APL. At low APL I think the CRT will perform much better and I think it will perform better than the best digitals. But the only way we will know for sure is if someone wants to invest the time and dedication into doing this work and why I really want to encourage Cliff to give it a shot.
One of the issues with getting a meaningful sequential CR measurement is defining the performance of grey scale tracking and low luminance detail that’s required. It’s very easy to degrade these and get a very high number with the G90. What is the requirement for D65 accuracy and at what luminance levels? What is the requirement for gamma starting with digital value 17?
Yes, I can see this easily happening with the off state measurement with a CRT. I think it's also easy as an example for someone without a gamma tweak to sacrifice greyscale detail for blacks. Because the off state of CRT is so good, I know it will have huge on/off numbers but this doesn't really mean that much except for black out scenes. For intra-image contrast at low APL though it's not so easy to tweak the CRT to get astronomical contrast readings which is another reason I'm most interested in those measurements. We know all about the weaknesses of CRT in MTF and ANSI cr but very little about it's strengths.
[font=Arial][size=3]The issue with ambient is not that the CR will be even higher if ambient is subtracted, but that the CR measurement is inaccurate. The main source of ambient is the screen. The entire screen is focused on the phosphor screen by the lens. You can see this clearly if you shine a light on the screen and look into the lenses.
Yup, I understand now. This is a similar effect to shining a laser at a phosphor or even a light bulb. For the measurements that I did in this thread, I placed a black sheet over the screen and also masked areas around the projector to eliminate reflected light off of the probe from hitting the projector and bouncing back into the probe. I can see how this can be even more important for CRT measurements.
The intra-image patterns used in this thread will likely have enough white so that the black reading will be many multiples above the ambient so that ambient will probably be less of a factor although it still should be carefully monitored and guarded against.
but I am not interested in making these measurements. I don’t use the G90 for movies and I have a much better overall projector for that. Also, I will be replacing the G90 soon.
Even if you're not interested in using your G90 anymore, you'll have to admit that it's still an interesting exercise in seeing how the best attribute of a G90 compares to a digital (and arguably the worst attribute of digital). I never planned on using a DI equipped projector but went through a lot of time and effort just to understanding better how it compares with native projectors (and hopefully other dynamic technologies in the future, especially iris equipped LCD's). You have the knowledge and the equipment, it sure would be interesting to see your results if you decide to change your mind and spend a day or so on this