darin's contrast ratio threads

Quote:
Any device can get no light. Just put the cap on. It is basically the next step up that it can differentiate from no light that is what is used for those f-stop calculations.


If the camera doesn't "register", do you think it should be played back with some level of light? I could see that if the person doing the telecine saw that it should have been something other than no light and wants to encode it above video 16.

I can see where some people may want to raise the absolute black level because the step from the lowest to the next lowest is too large, but you give up the ability to reproduce situations where we perceive no light, as can happen in real life. And as far as whether you want to raise the lowest light level because of the step to the next, have you ever seen any standard that called for anything higher than no light for the lowest value?


--Darin

Quote:
Let's say they point the camera at something where basically no light is getting registered. Then the person doing the telecine transfers it. What should they do, encode it to display some light? Maybe if they can see that the scene called for it for some reason, but why should a "scene" with no light get played back with light (or more light)? The only reason I can see is the reason above with the large step between the bottom 2 values, but dithering can be used for much of that. As far as noise, dealing with that is one of the tasks of the person doing the telecine or some version from the master.


If a camera captures nothing then the person doing the telecine can decide to raise the level above video 16 by dithering or whatever if the scene looks like it should be done that way, but if they think the scene should be a blackout then having the ability to encode it and display it that way makes complete sense to me. Why would we want to limit this ability? And if you don't think that a frame of all video 16s or lower should be displayed with no light, what light level do you think should be used in the end?


Let's just take "Cast Away" as an example. Tom Hank's character blacks out. Should the full frames for the blackout portion be displayed with some light, and if so, why?


--Darin

Quote:
That is why you want somebody intelligent doing the telecine. You could just have a machine do it if all you want is 1:1 mappings.

Quote:
If the camera somehow got .7 lux, I was doing the telecine and saw that it shouldn't have and the scene would be more correct if it mapped to 0 lux, then I would definitely want to map it to 0. Also, I don't know if the camera indicates that it got .7 lux if it gets no light or if it is a less than .7 lux case, but either way the issue is really the same.

Quote:
If you choose a floor higher than zero light for the lowest value then you have tied the hands of the person doing the telecine and then the display, so why would you want to do that? Values in between can be chosen with dithering, as I mentioned?

Quote:
I don't think that would be the on/off CR. That would be the video 235 to video 17 CR unless you want to map video 17 higher and make the lowest output value be .7 lux . You aren't suggesting that video 16 should map to .7 are you? If so, why? Maybe you want it to map to half that or 0.35, which would be closer to 6k:1, but that is arbitrary.

Quote:
You seem to be making an assumption that if the camera gives its lowest reading then it must have been looking at .7 lux, but that isn't the case. It will indicate that for no light also as I said. And CR for a capture device where the CR is calculated by using a +1 on the lowest end kind of measurement is different than CR for a display device where the lowest output value is used. The reason for the +1 kind of measurement for capture devices is because of that unknown range at the bottom and that since all of them can be put into situations with no light the method of using the actual lowest they can receive would be pretty much useless. But, that doesn't apply to displays.


Put another way, if you know the scenes filmed had more than 3k:1 CR (or an intelligent person can tell that they should have), why would you limit the display of them to 3k:1? Why should a blackout that somebody tried to capture on film be displayed with light? Setting the standard for lowest value to higher than zero limits what can be shown. And that is one reason no standard calls for the lowest value to be higher than zero. At least I can't find any. Rec. 709 sure doesn't since the lowest position calls for no light.


It seems that what you are proposing is that when the camera doesn't know how much light it is getting (it is outside its accuracy range on the low end) you should assume it was the brightest possible to get that (or .7 lux in this case) when playing it back. While I can see how somebody might jump to that originally, I don't see how that position is very supportable when looking at what making this assumption and limiting the range of things that can be displayed means.


--Darin

Quote:
The dimmest that the camera could ever get was 0-.7 lux. It doesn't know what it was at that point, so it seems that you are back to assuming that everything here was at the top end of that, or .7 lux. Do you agree that if the camera says .7 lux then you don't know if it was getting .7 lux or something lower? If it can do .7 lux accurately then the same thing applies to .6 lux or lower.
Quote:
Somebody doing a telecine sure better know what they are doing.
Quote:
If .7 lux is your lowest then you will lose shadow detail because now you have to raise the next level enough to get a just-noticable-difference, even in mixed scenes. How are you proposing that you get shadow detail by mapping "black" to .7 lux? You are claiming I will distort it by using 0 lux, but you are proposing distorting it by mapping everything in the 0-.7 lux case to .7 lux from what I can see.
Quote:
How do you figure that? 200k:1 on/off CR doesn't tell you what the gap is between two adjacent values (not even counting dithering). Do you believe that the gap between .0001 and .003 ftL is an incredible gap? Why? If you are just looking at the 200k:1 to 3k:1 looking like a big gap, then you could look at the fact that every standard that calls for zero light for the lowest level (as every one I've seen does) calls for basically an infinite CR to some smaller CR between the lowest level and the second lowest level. But in absolute light terms the gaps aren't big like that.
Quote:
We have been talking about video 16 the whole time, so you now seem to be proposing basically what I've said (other than me saying that the person doing the telecine should have some control, as they do). On/off CR is video 235 to video 16, not video 235 to video 17. That is a different matter and if 235 to 17 is 3k:1 it does not mean that 235 to 16 could not be 200k:1.


And how black should the projector be able to muster for a blackout in order to not have any room for improvement in this area? If you say more than 0 ftL, then based on what?


As an example, imagine a scene with a light on in a room and your 0.7 lux camera shooting this scene. The end of the scene consists of the light being dimmed slowly until there is no light. Basically, a fade-to-black. Then that holds for a few seconds. If a person was doing the telecine or transfer, how do you think they should encode this as it fades out? And how should a projector display it? Should light be shining on the screen during those few seconds where the blackout is held and should the viewers be able to see the screen? If so, how many ftL (or CR to 100 IRE) do you propose would be ideal for the blackout?


--Darin

Quote:
We are in some agreement, but I consider blackouts to be reality and raising the light level to purposely make blackouts impossible to be cutting off part of reality.
Quote:
Yep. I had a mistake in mine. I used 10 ftL for white and should have had .00005 instead of .0001 for the 200k:1 case.
Quote:
You keep saying that this will crush your blacks. Why? If you are talking about having a gamma that comes out very slow then of course you can crush your blacks, but this is a different discussion than where the black point should be. REC.709 calls for an absolute black level that is darker than your 200k:1 level (since it calls for no light) and video 17 to be about 1k:1 from video 235. Do you think matching that would crush your blacks? As I pretty much said before, raising your black level is going to hurt your shadow detail if your other points are kept constant because of the just-noticable-difference factor.
Quote:
The original discussion I was in wasn't about how many levels to give between black and where the camera was sensitive enough to pick up detail, but about where the ideal black level should be given that your original statement was:
Quote:
At this point what do you think the goal for projector absolute black level should be? I think no light (or at least the perception of no light) and haven't seen anything that disputes that.


--Darin

Quote:
I am talking about the ideal display, not limitations of CRTs. There is no reason that having the absolute black level at zero means that the next level up will be too dim to be seen other than with CRTs because of the way they normally come out of black when you make them blackout. We've discussed this quite a bit in the past and tse and others are working on circuits that will allow the absolute black level to be no light, while still having the next levels up be reasonable for shadow detail. Even the standards with linear tails above black have the lowest point at no light.


And the ideal may be to have a very slightly elevated black level in mixed scenes, but we already have that with ANSI CR limitations and with the ability to dither between video 16 and video 17, which I addressed earlier. But in my view ideal will never be something that can't reproduce what I see (or don't see) when I walk into a room, closet, or bathroom in my house and have no visible light. If a display can't reproduce that then there is room for improvement.


--Darin

Quote:
It is true that you would need no other light lighting up the screen to where you could see the screen, since the darkest your screen can go is how it looks with the projector turned off. However, we are on the CRT forum here and one of the main advantages of CRTs is the lower absolute blacks. If you have much light in your room making your screen visible this advantage just doesn't exist against good digitals because many of them can be used on gray screens in situations like that for better blacks and CR.
Quote:
Your room can be white. It is just a matter of not having other lights on. Those who want lights on should consider that the on/off CR advantage of CRTs pretty much disappears, as I mentioned.
Quote:
It is true that these things vary by the person, but everybody can view blackout conditions in real life. They just might not need the level of light to be as low to perceive no light as someone else. As we get older our ability to see darker stuff does go down (this results in lower Contrast Sensitivity Function scores) and on the surface it probably seems that this means we want everything to be less dark. This is true for the brighter parts of images, but the just-noticeable-difference comes in here and doesn't mean that we want the anchor point to be raised up. If the absolute black is anchored at no light and the next step up is kept at some brighter level, then in any scene (including mixed scenes) it should be easier to differentiate the shadow detail than if that absolute black level was raised up and the next step up was kept constant. In other words, I believe as we get older we want just as much (or more) on/off CR, but we just want things to come out of the absolute black level faster and probably want brighter images overall. In short, we may want video 17, 18, ... to be brighter as we get older, but I don't believe we want video 16 to get brighter (or whatever the black level is at the current APL with this moving).


--Darin

Quote:
This depends on screen size and more lumens doesn't have to mean brighter images. It can mean a darker screen for better CR (rejection of reflections off walls for better ANSI CR) or more of both ANSI CR and on/off CR with any lights on or coming in.
Quote:
The Panasonic AE900 has no SDE at 1.4x to anyone I know. There may be other issues, but SDE even at 720p isn't one of them.


--Darin

Quote:
I'm not sure people would consider it cheap (at under $4k), but the Optoma H78DC3 will do 48Hz and that is basically as good as 72Hz on a digital (since they don't flicker at 48Hz likes CRTs). As far as 72Hz and the main subject of this thread, I don't believe the G90 can do 1080/72p, but it looks like the Ruby will be able to play 1080/48p (which again won't flicker like it does on G90s). if it is fed 1080/24psf.
Quote:
Nothing really particular, but mostly just that there are other PQ differences other than SDE. I saw an AE900 the other day and it still didn't have the ANSI CR of DLPs or the on/off CR of CRTs, but it wasn't really calibrated and I will be getting one and checking to see how it does after proper setup.


And I don't consider the stuck pixel issue to be big if you just buy from a place with a guarantee in the first few hours to check for stuck pixels and things like that. My impression is that pixels that become stuck later on aren't any more common than CRT tubes going bad prematurely and probably less common than that. The only stuck pixel issues I've seen were there at hour zero.


--Darin

Quote:
I just noticed this. For one thing, $10k is the MSRP and I doubt most people who frequent a place like this will be MSRP. And as far as a scaler, some people might want one (especially for mosquito noise reduction or the like), but the jury is still out on how necessary one will be as the Ruby of course does scaling internally. Especially if BluRay puts out 1080/24psf for films and 1080/60p for video and a person is mostly going to be concerned with those after BluRay is available. In those cases I don't see an external scaler buying people much other than more connnections or features for their other sources. But a scaler that takes DVD film or 1080i film to 1080/24psf would likely be very nice with a Ruby for D-Theater and other material in the meantime for those who want to spend more money for those options. They just don't look as necessary as scaling with a CRT to me given that the Ruby comes with one internally.


--Darin

Quote:
If your friend with the AE700 or AE900 doesn't have a color filter then they may want to consider getting one in the meantime. I just ordered one that people use on the AE700 and I'm guessing will also be a reasonable one for the AE900 (given that I think they have the same bulbs). I'll probably get it next Wednesday and then have an idea about good calibration setup with it later. If he wants one, he could search for item 65073766 at www.bhphotovideo.com and see the one for $53 plus shipping.


If the Optoma is the H79 then if the owner of that wants to try a color filter to optomize it they could search for item EK1496728 at www.bhphotovideo.com and find the CC20R for $24 plus shipping. Not all of the same projector model are the same, but I think they tend to be close enough that I could probably give him or you some calibration numbers that would be likely to work pretty well with a CC20R filter on one of these, if the sources are standard video levels (not PC levels).


Have fun at Art's.

Quote:
I may wait until that filter shows up to really try to calibrate one of these AE900s and I haven't looked at that scene from LOTR yet. I didn't see washout like that with other things I watched though.


--Darin

Quote:
Doesn't surprise me at all. We are talking about a G90 here. I would take a Ruby to a fight with a G90, not an H78. And even with the C3X I would have used the lumens from it to use a dark screen and help the ANSI CR retention, but I didn't hear what the room was like. I wouldn't aim for a bunch more ft-lamberts from a C3X than the CRT because it just makes it easier to see SDE and brings out artifacts from the sources. I was surprised that somebody said they could see screendoor from 2x with the H78 and would have seen it from further. I had a CRT owner with good eyes in my theater and he was having trouble seeing SDE from a position less than 1.5x with the way I had an H79 (the version I have) setup. But, I also don't focus sharply and don't make the images a lot brighter than a comparable CRT would be. Part of that was modifying the H79 for more on/off contrast ratio in this case.


As far as the H78 having good absolute black levels, they aren't bad for digitals. The on/off CR isn't as good as the Sharp 11k when that is in the high contrast mode, but the H78 is brighter like that. And both have about 500:1 ANSI CR, but I don't recall whether the room will kill this in this case. If the H78 is just used pretty much as shipped then I would expect it to be brighter than the G70, which also means bringing up the absolute black levels even more. That is one reason I suggested a filter.


--Darin

Quote:
I'm somewhat surprised by this since with a CC20R filter I believe a properly calibrated H79 would still give close to 300 ANSI lumens on low bulb (I believe Greg Rogers reported over 500 ANSI lumens without a filter), but it does depend on bulb age and the calibration. Do you have any idea how many ft-lamberts you get with the G90? And what screen type was it?


One thing that is disappointing to me is that the component input on the H79 does seem pretty substandard. Not that it would have necessarily changed your opinions much, but the DVI on it seems much better to me. Greg Roger's reported on the component input not having colors that matched the standards and not as good as the DVI, and there also seem to be some issues with how the component input comes out of video black. But there might be a way to fix some of that with a special function while feeding the H79 all video black, but I don't recall what the key sequence is and haven't tried it. As far as component vs DVI, on the Sharp 11k I've found mostly the opposite of the H79, where I feel that the component input generally looks better than the DVI input. So, there is no way to be positive which input will be best without testing.
Quote:
Even between the Ruby and the Qualia I would take the Ruby to a comparison with a G90 before a Qualia. Although the Qualia can go very bright the on/off CR isn't very high. I said a while ago that I didn't think people here would be all that impressed with the Qualia and people ended up more impressed than I thought they would be. I saw that you guys are thinking of doing a comparison with a dealer and I would recommend waiting for the Ruby. Partially because it would be easier to take somewhere, but mostly because it shouldn't have near the on/off CR disadvantage of the Qualia. In a G90 vs Ruby I'm expecting some people to prefer each and for it to partially depend on the screen used (like a Torus could help the G90), but I think the Ruby will end up getting more votes in a setup for a G90 than a Qualia would.
Quote:
If it was just my Comcast box I would agree with you. I went to lengths to be able to store HD a long time ago, so I'm swimming in it. Even people who have come to the game late can find a lot of HD right now though. I can't tell you where (partially because I don't know the specifics and just have friends that do), but it is out there.
Quote:
I'm lucky that there are people in Seattle the same way. We look at things and are honest about the strengths and weaknesses of each. In other words, none of us tries to keep misinformation from being corrected like a few people here (not you of course). Unfortunately, nobody in the group around here has a G90.


--Darin

Quote:
I wasn't really thinking of Tryg partially because I've never been to any comparisons with him except for screens and there I did correct something and gave my differing opinion. And most of the people I was thinking of as reasonable were CRT owners. As far as projectors, Tryg has his opinions and I do tend to take a lot of his posts with a grain of salt, other than knowing that he prefers really bright, rainbows drive him crazy, and he has been on a 1080p kick for a while, but I believe it is pretty much clear that they are his opinions. To me that is much different than factual information. Or statements that really will throw people off. Right now I can't think of anything that Tryg has said that has been false recently (although I did need to clarify some stuff he posted about what I had said from CEDIA where I probably wasn't clear and I did that) and I just expect Tryg to be obviously outrageous about his opinions.


If you think I haven't had some pretty big disagreements on the digital forum then maybe you just haven't been reading there long enough. I'm not sure how many times we went around about on/off CR there with some people claiming that improvements weren't needed and I was accused of siding with CRTs because I said they were better than digitals in that one department and the digitals needed to improve there.


I know this is going to sound completely stuck up, but I think there have been enough times on the digital forums where I was proven right in the end after disagreements that less people question my facts or contradict what I say. At least that is how it feels to me.


As far as this comparison, I'm glad you guys had fun and learned some things. It is too bad we don't all live closer as I could show you a modified H79 (quite a bit more on/off CR, but lower lumens) in a mostly black velvet room on a 92" wide High Power screen. To me the colors look good after this calibration. The red is one that I like. I also just got a used Panasonic AE900 for $1700. I got a color filter and have done a little bit of calibration, but not a complete one. The reds look a little bit orange so far and the blacks aren't as good as my H79. But for the price and still bright on a 10' wide High Power with a long enough throw to put it behind my back wall, I'm pretty impressed. I would also like to see a G70 in a really dark room and if I recall Cliff has his walls darkened pretty well. The G70 I've mostly seen is in a room with some light colored walls, but I think I'll get a chance to see one in a darker room sometime here.


--Darin