Some people are still being told that 200:1 is all we can see....

I was specifically told that very thing by a dealer who told me he has attended "ISF courses" and that is where he learned this. I was talking to him about the Anthem Statement D2, when he asked me what PJ I was using. When I told him I was going to upgrade to something with better contrast (the RS1) that's when he threw out the 200:1 number.

Contrast discussions always seem to provoke heated debate here and rightfully so. The measurement of projector contrast and its impact on the projected image is a subject of great mystery and interest to most of us. The use of phrases like "image depth" show that we don't currently have (or agree upon) the technical terminology to properly describe all the projector characteristics needed to produce a great image. Contrast measurement standards do appear to be the most glaring lack in this area.

There have been several recent post on this subject that have highlighted the short falls of our ability to accurately assess and describe the interaction of the various ANSI and ON/OFF measurements on the projected image. I have also heard the 200:1 contrast ratio number thrown around in the past. As a stand alone number, it is just as meaningless as some of the other measurements. The spatial separation, absolute brightness and relative contrast of different points in the image are going to affect our visual perception of different objects in an image. Compounding the complexity of this issue is the fact that no two people have the same visual acuity. That is clearly shown by the fact that two people can look at the same projector image and walk away with two very different impressions. I read someplace that even the color of your eyes affects your ability to differentiate high contrast object, not to mention the age of some of our eyes!

I think the ongoing discussions on this subject will lead to new and improved test suites that will be able to more accurately describe picture qualities that result in the description, "an image with great depth" being definable.

This is unfortunate to hear.

It might be productive to contact the folks who ran the ISF seminar. There are a number of folks around AVS I believe who have taught or do teach ISF seminars. Most are interested enthusiasts like us, and know a huge amount and have a great deal of experience, but no one knows everything! There have been some discussions in the past with a member or two who taught ISF seminars who were confusing some contrast issues, and I can also understand how it'd be quite easy to believe errant contrast numbers because it's not a very intuitive thing for a lot of people. If you were to contact the ISF and/or the instructors of that course nicely, they should be interested in exploring the issue and I'm confident they would correct themselves so as not to spread disinformation, which is the last thing they want to do!

I'll make a linky thread over in the calibration forum because there's a lot of ISF folks over there who might be interested in this, or helping out, etc.

(it would also be nice if tbrunet didn't enter this thread)

Well, technically, I believe that the eye can only see a very limited (possibly 200:1) contrast ratio. However, our eye works like a DI iris. It adjusts our range to fit the material we're looking at. That said, our eye is EXTREMELY fast in adjusting it's iris. Therefore, the dynamic range of an image is still important, and we have no problems seeing the difference between a high contrast projector and a moderate contrast projector. While technically we may only be able to see a 200:1 (again, haven't verified number) contrast ratio, much higher contrast ratios (15000:1) still make a very noticable difference in PQ. When your eye scans a movie image to take in the entire screen, it's rapidly adjusting it's iris to view the entire contrast ratio of the movie, even for a single frame. The eye may only see 200:1, but the brain sees a much higher number.

EDIT: I was wrong.

EDIT: err.. I was right? Sorry, I'm doing my best from a vague memory of someone describing the above to me.

That is not correct. Our iris adjusts relatively slowly, at at any one iris position we can see about 100,000:1 across a scene. This is, obviously, significantly greater than the 100:1 or 200:1 numbers so often erroneously quoted. If humans could only see 100:1 or 200:1 instantaneously, we wouldn't hardly be able to drive around at night! With iris adjustment over time, it's more like 10^14:1.

Contrast ratios in a display of 15,000:1 are a LOT lower than you think. CRs many many times that are easily achievable on CRT displays, that aren't even going fully to black.

I have been around the visual display industry for years and am fairly familiar with the issues of visual acuity as it relates to apparent resolution and how contrast affects that.

As maddogmc points out, there are a very wide range of factors involved when it comes to how humans perceive contrast in an image and what that perceptible range is for any given person. I won't add to his comments, but published scientific data that I've seen puts the upper end of the range under best conditions to be a little under 300:1 and you go down from there as you factor in age, image intensity and spatial separation of the light/dark elements.

Bottom line is that there is no one test that will accurately model perception for a wide variety of people.

Cheers,

Chris,

You have it exactly backwards...

Can you please cite that source?

My suspicion is that, like many of the other very low numbers quoted, is that it is a misinterpretation of the study. The PhDs at brightside who study this stuff in HDR environments state numbers that are significantly higher. The much lower numbers often come from confusion about CSF, or from measurements right AT a high contrast boundary where localized flare reduces the CR through the eye and obscures detail. A very common experience we all have frequently in real life (except in Seatte ) is being indoors and looking outside on a sunny day. The CR of a scene like that can easily be 100,000:1, and yet we have no difficulty seeing details in the darker interior room, and outside simultaneously with little or no iris adjustment. According to many people who are loosely repeating numbers that float around, this would be impossible! Yet billions of people across the world have no difficulty seeing when encountering these kinds of very normal situations. Some people with degenerative eye diseases or other problems can end up with very serious reductions in the ability to see contrast in an image and end up with numbers as low as you describe, but this seriously impacts their ability to function in the world. For instance they would be prohibited form doing things like driving or operating machinery...

What do I have backwards? Please elaborate. I'm fairly confident that what I've explained is quite correct.

I understand, but that is exactly the problem. It's just commonly held belief that we can only see a hundred to one, or a couple hundred to one or so, and this gets spread around and pretty soon reputable people start saying it, and by that time of course it becomes FACT and then it ends up getting taught as fact. A few minutes of simple testing and measuring can very quickly disprove this. In fact, I'm quite surprised that Tom is stating these things, because he should be able to measure and test his own vision quite quickly and easily on a checkerboard pattern and see how much he really can see. An easy way to do this is to throw up a checkerboard pattern (he can measure the ANSI, or a lay person can go by the rough ANSI measure of their projector), and then put his hand in a black square and see if he can see a shadow of his hand. If the projector's ANSI is up towards 300:1 (or whatever number you think is the maximum we can see) then presumably the projector is exceeding what we can see and so there is no way that we could see a shadow in the black square. Impossible(the claim is). Try it and see what happens.

The dynamic range of the human visual system is up in the range you mention WITH the use of the iris...in a static scene, the range is <300:1, typically far less depending on the factors maddog and I mentioned. I'll post some references for your review in a later post...have to dig them up.

I will agree that the eye can accommodate scenes of a greater contrast ratio than that, but the ability to resolve any significant detail at both ends of the range, simultaneously, is limited and thus irrelevant.

My recollection from ISF training is that the 200:1 figure referred to intrascene (ANSI) performance.

One should also keep in mind that while the on/off performance of CRTs can be very high, this requires a fairly long waiting period between measurements. While watching a CRT in a dark room, try turning off the display and note how long it takes before the dislpay goes visibly dark.

I don't mean to be argumentative with you, but I would like to see a source for this claim of "<300:1" in a static scene without iris adjustment. No source I've seen has ever stated a number like that except via the grape-vine.

I also strongly urge you to do a quick test with shadow puppets. It'll only take you a minute, and I'm sure you have enough projectors laying around . The RS1 looks like greg measured it at somewhere around 250+:1 ballpark. This should mean that shadow puppets would be pretty difficult or close to impossible to see in a dark square of a checkerboard pattern. If you have a DLP laying around(i know I know!) with higher ANSI, try that one too. That, presumably, would be impossible to see shadows with in a dark square.

That is precisely the assertion that is being challenged. It is quite objectively mistaken. We can see a LOT more than 200:1 instantaneously and it's not hard to prove it. In other words, it is unreasonable, given the very significant capbilities of the human vision system.

Again, if you could only see 100:1, 200:1 or even 300:1, you would have very significant visual impairments and at the very least you shouldn't ever be driving a vehicle at night because it would be nearly impossible for you to see anything. Yet that's what is being claimed, unfortunately, by people who should be able to explore this with relative ease and easily "see" that this claim is quite silly.

Chris,
Could you cite some of your sources pertaining to both human iris response speeds as well as static contrast ratio? Please don't take this as me challenging you - I have absolutely zero scientific information on this subject, but I'd definitely be interested in seeing some. I find this kind of stuff fascinating, and I really think it's often overlooked when discussing high fidelity audio and video equipment. Thanks,

HP

P.S. - I loved your HP/retro-reflective review in the other forum, you've made me seriously consider switching to that type of setup in the near future!

I'm with Chris on this one. It looks like people started extrapolating from theories until they got to a point where it no longer coincides with demonstrable reality.

On the checkerboard question, should I be able tell that my black velvet boarder is darker than the black checkerboard, if my instantaneous is demonstrated to be over 200:1? Because I'll go check it when I get home if you predict it won't... but I'm fairly sure of the outcome.

The more direct source is from conversations Darin and I had with the folks at Brightside. At the time we were fighting the same misconceptions here on the forum about how little we could see instantaneously. At that time the popular number was 100:1, so at the very least it's nice to see the claims have creeped up to 200:1 or 300:1 or so, which we presented to the guys at Brightside who laughed at that idea. They quoted 100,000:1 instantaneous (5 orders of magnitude), and roughly 10^14:1 over time with full iris adaptation.

The closest text-source they have which mentions this is found here:

http://www.brightsidetech.com/tech/p...ggraph2004.pdf

See section 3.

Excerpt:


Their HDR display has enormous on/off CR, and significantly greater dynamic range than any conventional display. In addition, they achieve something like 25,000:1 ANSI contrast (yes that's ANSI, not on/off).

If you google for "disability glare" you can find a variety of sources to journal articles that discuss the reduced CR range that we can see right at a high contrast boundary. This is important in real life, it's why stop lights have those big black borders around them, and why the entrance of a tunnel is lit more brightly than deeper into the tunnel for example, things which you'll find discussed in these studies.

Also, you can google for 'contrast sensitivity function' which is another source of confusion about the specific 100:1 number. CSF does not have to do with contrast range between black and white, instead it has to do with the Just Noticeable Difference, essentially bit-depth requirements to avoid visible banding. But some apparently have confused that and the 1.01:1 or 1:100 expression (1% JND) into a maximum visible CR range of 100:1 which of course is a completely erroneous interpretation.

I agree. The < 300:1 per visual frame is certainly the CW in the HVS community. I am on the BOD of a artificial vision startup, and that is the number the Chief Scientist uses. The joke at our board meetings is; 300:1 is the number we will use until my video projector "pals" come up with 50 years of experimental data to prove otherwise.

Note that the < 300:1 is not the same as an ANSI CR measurement. It is more a measure of the range of light levels that the HVS can discern in a single visual frame. levels outside that range are clipped into white and crushed into black.

That is also another simple example to test. According to the low numbers being suggested here and many other places, masking a display that has good ANSI CR performance would not at all be necessary. You would be unable to distinguish the mask from a white screen surface and would provide no benefit. Or maybe there's just a lot of people out there with super-human vision abilities...

Any idea where I can pick up my own HDR display? I'm sure it's reasonably priced for the average person

Thanks for the material - that should keep me occupied for a while. If I actually had a real job that involved real work, this whole home theater hobby would be quite a bit more difficult to find time for!

You will have to clarify what you mean by the 300:1 simultaneous figure and how that is distinguished from a simultaneous contrast measure such as ANSI. Simultaneous contrast ratio in one case(apparently), simultaneous contrast ratio in the second case, yet the two are different? Seems like a desparate logical leap to me. (Of course, I suspect you are confusing a max percievable CR figure *at* a boundary with the max CR across a scene. This is the same claim you made last time this discussion arose...)

Are you capable of driving a vehicle at night? Your 300:1 simultaneous CR range limit seems to suggest such a task would be impossible.

The folks at Brightside who designed an entire display that hinges on the localized flare characteristics of the HVS clearly disagree with you, as they quote 100,000:1 as the instantaneous capability without iris adjustment. A simple shadow puppet test, or masking material test will easily illustrate that at the very least the number is quite significantly greater than 300:1, even in a relatively localized view.

It seems like I am misinterpreting HoustonHoyas point, but he was being fairly unclear IMO.

In the other thread he says:

Which is what I am trying to say. *at* a boundary, the CR range is drastically reduced due to flare/spill. But this is at very small angles, very localized. At larger angles such as across our field of view, the max CR range we can see is quite enormous. So if you have a single-pixel black-white transition, on either side of that transition for very small angles there is a severe reduction in our ability to see the full CR range and detail may be obscured due to glare/spill. But if you move slightly away from that boundary beyond the localized spill region, we recover the enormous instantaneous CR range that we can see.

This is precisely why we have to be careful to be clear about what we are talking about. If one person says "we can only see 150:1 (as the Vos paper cited above claims)" that claim may not be incorrect in the context of what that claim actually says. Unfortunately, people are using those claims to say that "150:1 ANSI CR" is all that is needed because that's all we can see. Of course that is complete nonsense and is completely unsopported by what they are mistakenly citing.

I don't want this reply to get lost in the thread. I know there are a lot of bright and knowledgeable people that are defending the 300:1 number, and I would like to give them an opportunity to explain this.

Let me restate the challenge directly:

Q: Can we see that the black velvet boarder is darker than the black checkerboard pattern on a display that has over 300:1 simultaneous contrast? If you admit that we can, then doesn't that prove that we would benefit from more than 300:1 simultaneous contrast in our displays?

It would seem to me that it does.

That makes a lot of sense.

God, I love this place! The level of debate and knowledge on a wide range of subjects is amazing.

The quotes and descriptive phrases used so far highlight my position on this subject. ANSI CR, intrascene CR, simultaneous CR, high contrast boundaries, etc. All these terms attempt to describe something regarding the visual image/impression/information received from a display.

Just to highlight a problem with one of these, I will pick on the newest phrase to me, "high contrast boundaries". I am not picking on you Chris, I highly respect your knowledge and passion for this subject.

Even the Brightside people seem to agree that "the maximum perceivable
contrast is somewhere around 150:1" for a High Contrast Boundary. The extended question that raises with me is the definition. How wide is a boundary in this context; 1/2 degree, 1 degree of arc? Is this CR perception level constant at all light levels and colors? As the width definition of boundary changes, what happens to my contrast perception? Once I can perceive something, how much information do I actually get and is it meaningful in the context of a given display system and its intended use?

I have similar questions about every other measure of CR with the possible exception of ON/OFF. Even with that measurement, the color temperature of the OFF state may impact my perception of "black".

I still believe the HVS is largely a mystery and there is little understanding of how we "see" and interpret visual information. We have learned a few tricks such as the use of sequential colors that makes single the chip DLP projector possible but we still have a lot to learn!

Let the debate continue!

EDIT: By the time I finished typing this, other people have already covered some of my points... I'm just a slow typist (or thinker)!!!

Here's where I'm coming from: Not at all. Your eye scans the entire field of view, it's iris adjusting along the way to give you a much higher contrast ratio - note that this is different from visual adjustment that occurs when you turn the lights off in a dark room (which you seem to be citing, when you say slow iris response). When you look in one direction and don't move your head, your eye is rapidly scanning the field of view, adjusting it's iris to compensate for the difference levels of brightness. As you hint at, there are different definitions of simultaneous floating around. Your definition of simultaneous is that of a single field of view, when in fact, simultaneous, to me, means a brief blip in time in which the eye and brain is processing only a small portion of the field of view. My definition of simultaneous is only a very small fraction of the time in which you define simultaneous. Either way, we see way more than 200:1 ANSI, since our eye is scanning the entire image and, therefore, is capable of dynamically discerning much more than the 200:1 difference in brightness.

Are you following me? - I may not be explaining it as well as I should be. I wish I could remember where I'm recalling my memories from.

No problem, it's a good question. I don't want to avoid the question entirely, but I do want to point out that our displays currently fail to exceed our visible capabilities in terms of rendering a high-contrast boundary. That is to say, the spill created by a display is more significant than the spill created in our visual system, so our eye's limitations and the specifics of this is not really an issue for normal displays. A crucial exception is the HDR display I note above which relies on our visual limitations to mask its own limitations. That display spills a little bit at a high contrast black/white boundary, but it is below our visible threshold (i.e. our eyes spill more than the display) so it becomes an issue.

This is probably a significant reason why this has never really been discussed that significantly with regards to displays (again with the exception of HDR display) because it's never been an issue. The display limitations are far more significant than our visual limitations(despite claims to the contrary).

So, what we're left with is basically the scientific papers that deal with this more in real life:

http://www.optometrists.asn.au/gui/files/ceo866363.pdf

deals with spill/flare issues in our vision. That has some formulas that you might find interesting, but far be it from me to try to translate that into a simplistic explanation! I haven't read through the whole thing in detail and digested it, but it's a starting point. I'd like to make statements about things from my memory about the size of the object and where it is in your vision, etc, but I don't know if that's just common-sense stuff or if I read it in papers, so I'd rather refrain from that. But common sense does tell us that if you have more are of bright objects, you're going to have more significant glare problems than small point-source types of light sources in the middle of a dark scene. We can see how that's the case when talking about instantaneous contrast performance as the image goes up in APL, similar things are happening in the "lenses" of our eyes too. I believe I remember reading somewhere about the size of the white side of a transition, if it's very large for instance (like say the whole sky) then the area and amount of spill on the dark side of the transition will be more significant than if you had say a narrower stripe of sky. But again, my memory is foggy, so don't hold me to that, it certainly does make sense intuitively however.

Yes, I understand that, but when I say simultaneously, I do mean at one instant in time. If you hold your gaze, with no iris adjustment, you can see across your field of view a luminance range that encompasses roughly 5 orders of magnitude, according to the folks at Brightside. That is at a particular iris setting without dynamic change, in one instant of time. Our iris doesn't move that fast in terms of adjusting in tiny fractions of a second as we dart our eyes around in real life. That's also why dynamic irises in projectors can trick us quite well. Note that our field of view is quite large. The key point is that the 100,000:1 figure cited is NOT for dynamic capabilities over time. It is for a single scene, a still image, without iris adjustment. Within a particular scene, small parts of that scene may cause us to have reduced capabilities because of sharp high-contrast boundaries which I was discussing above, and that's where the number may shrink in very localized areas in very small parts of a scene. But across the scene, that figure is for simultaneous capabilities, and simultaneous means that there is no time for iris adjustment. That is the way I understand the Brightside quote, and I think Darin was quite clear about this when he was talking with them in terms of what "simultaneous" meant. It was quite clear (as I recall) that it meant with no iris adjustment whatsoever. But you're certainly free to question my memory, which is notoriously confused.