Gray Scale & The Dynamic Iris?  

Congratulations, that's the rub! :-) You've just found one of the big potential problems with a dynamic iris. Seems to me that if the design is not pulled off well, you'll end up having to calibrate at every point in the dynamic iris' range. For greyscale AND shading, actually.

Widescreen Review's review of the Sony HS-51 hit on this problem, I believe.

Heck, it may even be a problem if it IS well done. I wonder if the Ruby provides any sort of means to calibrate through the range of DI openings.

I can't see the iris, in and of itself, being the source of grey-scale errors. As long as the lamp is not being modulated as well, the iris should act as though it were a variable ND filter. If the lamp is being modulated then there would be some issues with the variance of colour temperature, but I wonder if this would be great enough to have a major impact?

I think the main culprit is the dynamic gamma that is coupled to this mechanism to make the dynamic iris truly act as a useful means of varying the CR window.

The AE700 and I believe the HS51 incorporates lamp modulation as part of the iris/gamma loop. The depth of lamp modulation and the ability of the variable gamma to track accurately would determine the quality of the greyscale.

As to whether a DI can be made to work without lamp modulation - I would like that answered. With broadcast cameras we simply adjust gamma when we wish to stretch detail in low APL scenes, but because of its electronic nature we can maintain a white point. Since I'm trying to understand projectors, I would think that lamp modulation is used to maintain that white point.

Then again I might be out to lunch altogether.

ted

Cat eye irises do cause changes in greyscale. There is a TI white paper that discusses this. You need a degree in optics to understand why though - I don't have one so not sure my explain would do justice to the topic.

I have measured this in the HT1100 - it is indeed real.

Even lamp power modes can change greyscale - the H31 I tested adds more red spectrum at the higher wattage

tvted - I think consumer electronics manufacturers care less about keeping an electroptical white balance than broadcast cameras! Professional broadcast actually cares about video engineering standards - consumer electronics this is rare.

It is not possible to get a good grayscale with the HS51 and the auto IRIS enabled.

EDIT: The use of "good" might be a bit strong. It is not possible to get accurate gray scale tracking when the auto IRIS is enabled. When you set the IRIS to one position, then you can get a much more accurate grayscale tracking from bottom to top.

But is this because of the iris or poor gamma tracking when the dynamic gamma is engaged? If the dynamic gamma part of the implementation needs work - surely this is addressable.

ted

Yes, I think it can be addressed, but it is not clear that the H51 allows the necessary adjustments. It basically takes a separate gamma LUT for each position in the iris.

Bingo! This thread is discussing the very problem I have noticed with variable iris. Much as the DLP haters can see rainbows, I can see the iris changing and it bothers me more than RBE. How can you calibrate a good gray scale knowing that it will NOT be the same calibration when the top ire signal is 35% of the scale? Does this not skew the calibration if setting for full ire and then rechecking at 35% of full ire? I can see the changes in contrast and brightness but it is hard to describe much like wearing fast reacting photogray lenses. If I was to own a ruby I would have to set the iris on some fixed setting and probably leave it at that.

I'm planning on leaving mine on fully-open when I get one.

A cats eye would only cause a % change in the gray scale value but not it's color.
Adjusting power to the bulb wouls cause ashift in color temp.

MichaelZ
I do not remember if it was you or someone else saw the DI hang in the demo. I guess we all expect this not to happen with real production units. Also this DI is supposed to be much faster than previous DI. I understand the calibration issue. If it is close enough to perfect D65 few will notice and possible prefer the incorrect, horrible thought.

This is why we are in a long wait for my longed for lasers to replace bulbs. The lasers would be perfectly "DI"-capable. Also as pointed out Ruby is supposed to have dlp level contrast without DI.

I do not hate DLP and I observe RBE. Observing RBE prevents me from considering DLP. I don't hate DI-equipped projectors (regardless of the digital technology) and I have observed DI operating behaviors from products using different approaches in DI design.

There is something you need to realize and that is when some observed goes from being bothersome to downright annoying. Bothersome is an inconvenience I can tolerate to a certain degree. Annoyance is a outright distraction from what I am trying to watch. RBE annoys me and DI operating behavior(s) bothers me.

I don't fully blame observed RBE on the technology, becaue it is an interaction of technology AND my brain. Its like me not hating a bullet, but only the interaction of it and my body. :eek:

So please don't try to lump people together. Some of us have our reasons for not considering DLP up to this point. Also, the aspect of RBE is annoying to me to a sufficient degree that I haven't even had time to consider other potentially negative aspects of DLP, like dithering. I am sure the day when RBE no longer annoys me I'll start to focus on other aspects in a critical nature--all without hate.

It was probably me who posted that I saw the Ruby "hang". I didn't know a good way to describe what I saw those 2 times, but I believe Michael's: Is a good analogy to what I was trying to describe that I saw. The scene was dark, it cut to a (supposed to be) brighter scene, but it was dim. I was just about to open my mouth and say hey, that dosen't look right, but before I could move my lips, bzzzzzp, the sceen brightness came up to where it was supposed to be. So instead, I said "hey did you see that?" to some stranger. The scene I remember best, was someone standing next to a car in a daylight scene (Nicolas Cage?, forget, I was paying attention to the picture, not the movie).

GetGray
If it changed to looking good I think it was just a glitch. Some hardware or software issue that is to be fixed.

Someone commented the Ruby on display had issues covering several pages to be fixed before the final product.

Someone mentioned something about a special filter or lens that is being used in conjunction with Sony's aperture-style dynamic iris. I presumed this was something like a plate that zipped in front of the optical path much like you would slap an ND filter on.

Also, while it was recognized that the cat's-eye style of iris might affect grayscale, what about an aperture-style?

Yes, I agree. Just something to watch for as they hit the ground. With all the superman eyes that were watching it (mine are "regular mortal" eyes, can't even see rainbows :)), I was surprised no one else, except MichaelZ commented on it.

Grey balance (greyscale) is indeed affected by irises, and I think this is one reason we are only now seeing widespread use of irises: The projector manufacturers have finally gotten greyscale under control enough that they might be able to make it work with irises.

Here's why the iris affects greyscale.

The iris doesn't directly change the color on it's own; it's just a neutral density filter. But what it does do is move the greypoint around on the grayscale curve. A greyscale curve is a plot of the color (usually expressed in kelvin) on the Y axis versus the video level from 0 IRE to 100 IRE. In a perfect world this would be a straight flat line, but in the real world it's usually not quite flat. Some output levels usually have a different color, whether it's 10 IRE, 50 IRE, or 90 IRE.

Now the problem is that the iris will move the scene around on this curve.

Lets say you have a scene with 5 IRE shadow, a 30 IRE grey spaceship, and a 100 IRE glowing white robe. This scene will require the iris to stay wide open. Now lets say the 100 IRE white robe leaves the scene and so the 30 IRE spaceship is the brightest thing in the scene. The iris closes down to about a third of it's original size. In addition, and this is the interesting part, the imaging device must increase it's output so that the 30 IRE spaceship is now close to the "full on" imaging chip level, the same level was used to reproduce the 100 IRE glowing white robe before. In the process the 5 IRE shadow gets scaled up to three times the pre-iris level as well.

Just as an example, lets say the best calibrated greyscale of the system is 6000K at 0 IRE and 7000K at 100 IRE, with a smooth linear transition between. That means that when the robe was in the scene, the 30 IRE spaceship was displayed at 6300K. But when the robe left the scene and the iris closed down, the 30 IRE spaceship shifted to 7000K. I would be worried that the 700K difference would be noticeable, especially if it occurred in 1/60 of a second as some irises are capable of.

So that's how the iris will end up causing color shifts. Whether these shifts are noticeable will depend on how fast they occur, what scenes are viewed, the dynamic range of the scenes, and most importantly on how well the grayscale is calibrated in the first place.

Modulating the lamp output has the potential to cause the same shifts, plus color shifts arising from the fact that color balance often changes based on the lamp output and temperature.

Whether these things end up being problematic remains to be seen... I've always thought that dynamic irises are a great idea. It's the low contrast movie scenes that really demand high projector contrast, and that's where the irises shine.

I would like to see these auto iris systems come with built in calibration systems that would measure the RGB output at different chip and iris output levels. Perhaps with a sensor-containing lense cap that you put on every few hundred projector hours. It really would be quite cheap for the hardware since by tailoring for the projector's color balance you avoid the need for fancy sensors. Just a few bucks for a cheap photocell and A/D converter should be it, plus some software costs.

PerfKnee,

"But what it does do is move the greypoint around on the grayscale curve."

I still don't get it; can't most pj's these days be calibrated for good grayscale tracking?

You're talking about gamma, correct?

You need perfect (not just good) grayscale tracking in order for the iris not to make grays worse. Minor grayscale errors are generally not noticed in non-iris systems because whatever incorrect coloration happens, your eye adjusts and decides that's the color the object was supposed to be. But when the incorrect color changes dynamically in front of your eyes due to another bright subject entering/leaving the scene (causing iris changes), that has the potential to be more noticeable. It's not that the iris is creating greyscale problems per se, just exagerating any greyscale problems that already exist.

BTW, this discussion doesn't relate to gamma, even though there is of course a gamma curve in effect. It's just about what color temp a particular video level or imaging chip transmissivity corresponds to. Actually I suppose there is a tie-in to gamma in that some manufacturers store the grayscale calibrations in what they call a "gamma lookup table" which in reality stores both gamma and grayscale. But that's really just an aside.

"You need perfect (not just good) grayscale tracking in order for the iris not to make grays worse."

Even if a pj had a perfect gray scale tracking, you'd need a nonexistent perfect measuring device to know it, so how close are we talking here?

Thanks

P.S. Apparently the Ruby is close enough.

In my experience the ability to notice the difference between a colour temp of 6000k vs 7000k in an absolute sense is rare. Colour is relative in this sense. The situation where you suggest this would be happening in a 1/60th of a second modulation is what I would call colour cycling and the chance of that happening within a given movie is approaching zero in my view.

As to the linearity of the grey scale - on a calibrated machine how often does a variance of 1k across the 100 IRE range occur?

Dynamic Gamma is an issue as the implementation of an Iris requires it - how else would you maintain low APL detail? I ask this because the overhead that a wide On/Off CR provides is still limited to what the PJ is capable of in an Iris Off situation.

As Michael Grant pointed out the likely need for a Gamma LUT for each position of the Iris to me seems the daunting problem.

ted

And there are times in which one might want to trade-off color accuracy (and grayscale) if it promotes another variable to a much greater degree, like constrat-enhancements by moving from 6000-6500 to 7000K.