[PapaSloth]

A bunch of times on this forum, someone has suggested that you could use an LCD as a variable iris to lower light output in dark scenes (or several variations of that idea). Here's my recap of why this helps:


I'll use the term "scene" to refer to any static image at one point in time, "intrascene contrast ratio" to refer to the ratio of the brightest point to the dimmest point with one scene, and "interscene contrast ratio" to refer to the ratio of the brightest point to the dimmest point within two or more scenes (that is, at different times).


I've read that the human eye is only capable of resolving around a 200:1 intrascene contrast ratio, but that the iris of the eye allows it to resolve everything from a clear day at 10^4 cd/m^2 to moonless, overcast night sky at 10^-4 cd/m^2. So, the intrascene contrast ratio needed to "fool" the eye is pretty low, around 200:1, but the interscene contrast ratio needed to fool the eye is pretty high, somewhere between 100,000:1 and 100,000,000:1 depending on what you read.


The previous point is supported by our observations that bright scenes look pretty good with most of our current display devices, but dark scenes look "washed out" or lacking in contrast. That's because in a bright scene, the intrascene contrast ratio approaches what the maximum that the projector is capable of doing (say 400:1 up to 2000:1) while in dim scenes, the intrascene contrast ratio is a correspondingly small portion of that range.


So one notion is that if you use a "single pixel" LCD to dynamically reduce the total light output from the projector, just as the iris of the eye blocks the total light input to the retina, you can then use the full dynamic range of the projector in every scene (so you have the same intrascene contrast ratio in bright scenes as in dim scenes).


No doubt this explanation was far to slow for some and not nearly slow enough for others


Anyways, here's a high-level design based on my guess as to what's easiest/cheapest/most readily available:

1). buy a cheap VGA overhead projector panel from eBay for around $50.

2). hook the panel as a second display to an HTPC and place it in the projection path. Alignment isn't an issue because we're just going to use the panel as an iris, not try to make pixels line up.

3). write a directshow filter (or ffdshow plugin, etc.) that:

A). calculates the point of maximum illumination for a scene

B). scales each point in the scene so that they use the full range of the projector

C). uses the inverse of the scaling factor above to change the level of the VGA panel on the second display. I.E., just change the color of a full-field bitmap being shown on the second display.


As an elaboration, since color panels aren't really that much more expensive than black and white panels, deal with RGB as separate channels in the directshow filter above and scale each one independently.


Here are the problems with this idea (once again, from previous posts on this forum):

LCD panels are not perfectly transmissive so you're losing a lot of light even when the "iris" is "fully open".

Some projectors (like my D-ILA) already polarize the light so there is probably some weirdness having to do with polarization effects in the LCD.


This has come up enough times that I'm surprised no one has done it and reported back yet. I'd say a good first step would just be to buy a panel and measure the full off/full on maximum interscene contrast ratio without trying for the full directshow filter. I'll do this myself if no one can explain to me that I've missed something crucial


Let's really do it this time!

 

[PapaSloth]

OK, I've been looking around some. The big issue seems to be the transmittance of LCD panels. For non-polarized light, the panel will only pass somewhere between 20% to 40% of the light (monochromatic light at the high-end, and broad-spectrum at the low end). That might be enough with a 3000 lumen projector, but I don't have one of those


However, I haven't quite given up yet. If the light is already polarized, LCDs will pass around twice the light (still not very encouraging). Another possibility is a different technology for blocking the light. Here's a datasheet for a Liquid Crystal Polymer Shutter that doesn't use polarizers and has higher than 80% transmittance. The problems with this are 1). the aperture is only an inch, 2). it requires 120-200V at 1kHz (ouch!).

 

[Gary Lightfoot]

I already have a similar device - it's my girlfriend and a pointy stick.


When a dark scene comes on, I poke her with the pointy stick, and she shuts the iris on my HT1000. If it gets brighter, I poke here again and she opens it. After a while she gets fed up with the stick and does it without prompting. Voila!


Gary.

 

[PapaSloth]

Quote:

Originally posted by Gary Lightfoot I already have a similar device - it's my girlfriend and a pointy stick.

I can afford the HT1000, but there's no way in hell I can afford a girlfriend who is willing to be poked with a pointy stick


You give "wife acceptance factor" a whole new definition.

 

[jeffelliott]

Quote:

Originally posted by Gary Lightfoot I already have a similar device - it's my girlfriend and a pointy stick. When a dark scene comes on, I poke her with the pointy stick, and she shuts the iris on my HT1000. If it gets brighter, I poke here again and she opens it. After a while she gets fed up with the stick and does it without prompting. Voila! Gary.

Maybe consider getting a door lock actuator for a car and rig up a lever system. Push door lock switch, iris closes. Press door unlock switch, iris opens. Plus, girlfriend is even more attracted to "highly innovative" man!


Oh, just an idea!


JE

 

[buns]

Im thinking on something like this at the moment...... simply a variable pass filter that is controlled by the apl. I have a few novel ways but they wouldnt be DIY. An LC is the simplest route, probably the easiet way to DIY would be to butcher the panel from a laptop.


ad

 

[mikejz84]

It seems as though it would be more effective you install a high-speed iris inbetween the bulb and the panel or to figure out how to adjust the brightness of the bulb very quickly.

 

[Gary Lightfoot]

Whatever system is tried, it's got to be pretty quick to react to the video being input/projected to get a satisfactory result, and that'll be the real trick I would think. There is even a little lag with my pointy stick I have to say.


Gary.

 

[mikejz84]

I think finding a way to directly modulate the light output of a bulb is the key to increasing apparent contrast.

 

[PapaSloth]

I was thinking about modulating the light as well, but my projector uses an arc lamp, and they don't modulate very well. Your Mileage May Vary.


I don't think the device has to be _very_ fast because you basically

only need it to change on transitions from darker scenes to lighter

scenes, or vice versa. A couple hundred milliseconds should be

fast enough.


Variable iris sounds like an interesting idea. Do you know anything that's

commercially available that might do the trick?


I thought this link was pretty interesting.

It's another liquid crystal polymer link. I couldn't find any other

product info on google though.

 

[buns]

a mechanical iris is a problem mechanically! I guarantee you that if you had such a thing on a projector, the part of the machine most likely to die would instantly change from lamp to auto iris.


The use of an LC allows for fast modulation, the frequency response is not absolutely fantastic, but for video purposes if is many orders good enough.


Any true breakthrough solution has to make use of something other than mechanics, that is an old thing. If you care to research into research type instruments, the thing most favourable to use is silicon. Video is, comparatively, an era behind because we are still relying on prior technology.


ad