AVS Forum banner
1 - 5 of 5 Posts

·
Registered
Joined
·
1,373 Posts
Discussion Starter · #1 ·
Well, have to cancel my tee time this morning because of snow and freezing rain. Here is a question for the expert and eagle-eyed:


Without using a measuring device or A/B comparison, how accurately can you tell ansi or on/off CR by just eyeballing? Within 100, 1000, or not able to tell at all?
 

·
Registered
Joined
·
1,289 Posts
Contrast perception seems to be logarithmic, that is to say it's easy to detect a doubling in contrast, rather than standard linear changes.


So going from


1000:1 to 2000:1 is noticeable, then:

2000:1 to 4000:1

4000:1 to 8000:1

8000:1 to 15000:1


Obviously I picked random starting / doubling points, so you would perceive a 5000:1 pj to be noticeably better than a 2000:1 projector, but the difference between a 4000:1 and 5000:1 pj would be more subtle (not undetectable though, with familiar scenes, equal lumens and a side by side viewing.)


So a jump in contrast from 1000:1 to 2000:1 is more noticeable than from 14000:1 to 15000:1.


When you ask how well can you eyeball CR, on familiar scenes I would be able to place it in its general tier (1000-2000, 2000-4000, 4000-8000,8000-16000) reasonably well in a familiar environment with familiar scenes and some high APL to fade to black and hold type sequences.


Intrascene CR is harder but I do find it noticeably when you go from a high ANSI DLP to a LCOS type display like the Pearl. The ANSI seems 2x to 3x in certain scenes. In other scenes the difference is hard to detect. There are certain mid to high APL scenes with lots of dark colors where the pop of a high ANSI pj is detectable. I would say to within +/-100:1 since ANSI doesn't go too high on PJs.
 

·
Registered
Joined
·
6,092 Posts
rickster,


Eyeballing CR is done more on a logiritmic scale not an absolute scale. This means accuracy is best described as a +/-XX% number not a +/-XXXX:1 number


For low contrast projectors (say 500:1), I am personally probably accurate to about +/-20% (400:1 to 600:1). For high contrast projectors (5000:1) due to the deeper blacks my eye accuracy gets worse to about +/- 30% (3500:1 to 6500:1)


I am pretty keen on determining contrast ratio since for a while, say 5 years ago, it was by far the limiting factor for projectors. There was a time on this forum when many would claim that black level was the only thing that mattered; often people would claim things like "a 50 lumen 100:1 projector has visually better blacks than a 1000 lumen 1000:1 projector"


Contrast ratio is easier to tell than actual lumens. Depending on environments, your eyes can easily be tricked into thinking something is brighter than it is (my inaccuracy for telling lumens is about twice that of telling contrast ratio). After years of observations I finally discovered one of the popullar tricks is for a company to have a very dark section at a trade show and then have an even darker HT viewing room. With dark adapted eyes, projectors look much brighter.


-Mr. Wigggles


(My color temperature guesses are more impressive, but ironically I can't accurately guess the coordinates of primaries very well.)
 

·
Premium Member
Joined
·
4,708 Posts
"(My color temperature guesses are more impressive, but ironically I can't accurately guess the coordinates of primaries very well.)"// MrWigggles

..................

One of your best, one to remember
 

·
Registered
Joined
·
8,006 Posts
Unless it is an especially dark scene (night, star field, etc.), I don't see ANY improvements over approximately 4000:1. Anything above 2500:1 looks satisfying to me for the vast majority of scenes. Much below that and the image starts to looks flat to my eyes. Below about 1000:1 it appears simply unacceptable. BTW, I'm referring to native CR, not the phony CR you get with dynamic irises.


Other than that, I wouldn't trust my eyes to report much of anything in the way of CR.
 
1 - 5 of 5 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top