How 'black' is the black on most projectors?

[ImagineerTom]

so you'd "only" need a £30 circuit board (but by the time you've boxed it up and added all the cables/sockets/power supply you're getting closer to £100) AND a digital projector costing hundreds or possibly thousands of pounds to give you a fairly crappy profile that's not very durable, difficult to rig and has stupidly expensive lamp replacements costs.... or for £100 you could get a cheep Chinese light that uses existing principles, is brighter, cheeper, more reliable and does EVERYTHING your Pi-Projector gadget can do.

 

Other than as a one-off academic exercise to learn how to program a Pi I'm struggling to grasp what the actual benefit would be?

[gyro_gearloose]

Wow Tom, aren't you just a bright happy ray of sunshine tonight?

 

or for £100 you could get a cheep Chinese light that uses existing principles, is brighter, cheeper, more reliable and does EVERYTHING your Pi-Projector gadget can do.

So for £100 you can get a profile with built in colour mixing, gobos that you can select and rotate from the desk, and framing shutters that you can control from the desk? Please tell us where we can can get these super-cheap lights from. I'm sure we'd all love to know as they would make our jobs so much easier.....

 

Other than as a one-off academic exercise to learn how to program a Pi I'm struggling to grasp what the actual benefit would be?

And I'm struggling to understand how your post is of much benefit to this thread. Come back when you can construct a more helpful and informative post.

[ImagineerTom]

Aside from the framing shutters yes, all those features are fairly common as far as I know; I played with prototypes of units by MAD a good 10 years ago that did all that so it doesn;t strike me as being particularly fantastical?

 

I wasn't making a sarcastic comment; I'm genuinely asking "why" - clearly you've spotted a problem (that I've not spotted) to which you see this as a possible solution; I'm curious / unsure what your thought process is as on the surface of it and based on the experiences I've had with projectors and projector technology this seems to me to go against all conventional wisdom and cost considerably more than the first method I can think of for solving this problem. I'm keen to find out why/how/what this method achieves (retro fitting an interface box to existing projector technology rather than designing from scratch a lantern that exploits certain projector principles) and the ways in which it trumps the existing technologies and ideas.

 

If I thought it was out and out poop I would just say it; I wouldn't beat around the bush posting an over-long question asking WHY you're going down this route and what the benefit is...

[gyro_gearloose]

Well the first iteration of this idea did actually involve fitting a small LCD element into a moving head and replacing the mechanical colour/gobo/shutter mechanisms with just the LCD and a Raspberry Pi. I quickly knocked that idea on its head when I found that no one really sells small LCD screens with a high enough resolution, not to the general public anyway. Mobile phone screens might have been suitable as they are almost small enough and do have quite high resolutions compared to phones of just a few years ago. The problem is some of them use the DSI interface, which although the R-Pi does have the hardware to support DSI LCDs, as yet there is no firmware support. The Raspberry Pi foundation reckon that it will take them about 6 months before DSI is supported.

 

As for why I started thinking about doing this in the first place, it wasn't because I thought I'd come up with 'The Next Big Thing' in lighting. I'm well aware that there are moving lights which do exactly what I was thinking of. I certainly wasn't thinking that I could develop a market-ready moving digital light out in my shed! :)

 

Aside from the framing shutters yes, all those features are fairly common as far as I know; I played with prototypes of units by MAD a good 10 years ago that did all that so it doesn;t strike me as being particularly fantastical?

I'm well aware of the capabilities of moving lights :) However you did say you could get one for £100 that did everything, which was why I was asking where you could get it from.

[kitlane]

and would basically turn a projector into a profile.

Didnt prg or someone similar do this back in the late 90's,seem to recall it wasnt colour just black and white

 

I think you are referring to the Icon M made by LSD (LSD were bought out by PRG)

[Tom Baldwin]

If I understand you want to increase the contrast of digital projection? How important is peak brightness?

 

 

Can you add an LCD panel in the light path, but out of the focal plane? The DLP in the projector is the source of the image, and the source of colour, but this subsidiary LCD exists only to control gamma. While it needs to have a moderate resolution, it doesn't need to be massively high, and the panel doesn't need to be incredibly compact either.

Obviously, each pixel on this panel would be seriously out of focus, and would affect many of the real pixels (some more than others, depending on the spatial relationship).

 

You would need to create some video processing software which worked out how bright each primary and secondary pixel should be to get the necessary contrast. Some cases are simple - a hypothetical digital followspot would have a hard edged white circle on the primary display, and probably a hard edged circle of slightly smaller diameter on the secondary. As the edges of both displays would be black (opaqued), you should get a better quality of darkness.

 

The intent behind defocusing the second panel is to simplify the video processing, and also remove difficult alignment issues.

 

 

On a different tack, if you're up for a far more difficult project, could you use semiconductor lasers as the light source? You would need to create two scanning drum assemblies which delivered horizontal and vertical scan (analogous to CRT displays), and modulate the laser brightness electrically (for on/off) and mechanically (for intermediate intensities). You could certainly achieve a perfect black this way...

 

Final suggestion, which depends on you having far better skills with optics than I do... some modern flat panel TVs use multiple backlight sources for different regions of the screen to allow them to achieve improved contrast in the same way that you want to. Would you be able to illuminate the LCD panel on which you're displaying the image with an array of LEDs, and still combine this back into a single stream of light at the output? If you could, then modulating the brightness of the LEDs on the basis of desired image region brightness would help. Near perfect blacks should be possible?

[gyro_gearloose]

On a different tack, if you're up for a far more difficult project, could you use semiconductor lasers as the light source? You would need to create two scanning drum assemblies which delivered horizontal and vertical scan (analogous to CRT displays), and modulate the laser brightness electrically (for on/off) and mechanically (for intermediate intensities).

And while I'm at it I'll develop a completely white laser beam :D This system will need one if its to be of any use in theatre.

 

Thinking about it a bit more though, a scanning-laser light source will need some very clever optics if you wanted the beam angle to be adjustable. Otherwise you'd just end up with an array of spots if you zoomed out too far, or the pixels would overlap if you zoomed in too far. In a fixed angle laser projector, the laser beam would need to diverge ever so slightly so that there were no gaps between the pixels. Beyond the lens to do that, the projector wouldn't need any other lenses and it would always be in focus no matter the distance to the projection surface.

 

Of course I could be completely wrong about all this. These are just some thoughts I've come up with over lunch :)

[dbuckley]

... will need some very clever optics if you wanted the beam angle to be adjustable. Otherwise you'd just end up with an array of spots if you zoomed out too far, or the pixels would overlap if you zoomed in too far.

 

Is that not something that can be fixed by the ratio of movement of the scanning mirrors? As they are under electronic control it would be adjustable with a setting.

 

A scanning laser beam to draw an image isn't that revolutionary an idea; that's the way laser printers work, albeit in one dimension only. This is the whine that the laser printer makes as it bursts into life, the motor that rotates the mirror.

 

Some crazy people have done some work on this already, see here and here.

[gyro_gearloose]

Is that not something that can be fixed by the ratio of movement of the scanning mirrors?

Yes it would, but you'd need to be able to adjust the size of the laser beam as well. You'd need some clever optics on the laser itself to accomplish this. The beam angle of the laser must be the total beam angle of the projector divided by the projection resolution so that the pixels don't overlap or have gaps between them.

[dbuckley]

Either I'm very wrong, or I'm failing to explain myself, or probably both :)

 

The size of the laser beam dot is (to the necessary degree) constant, as it's choerent light, but the image size of the dot will vary as its hits the surface at anyhing other than square on. Assuming the suirface is flat, a bit of math will give the size of the dot or elipse formed. As we know that, we can then calculate the number of horizontal scans required to form an image where all the lines touch each other, even though it means some will overlap.

 

The same math will also give us a number of pixels along a scan line, but some pixels will be wider than others, again due to hitting the surface at an angle.

 

I would expect both of the circle to elipse distortions would be miniscule of projecting from a significant distance, but quite serious if you are 2m away from a 4x4 image.

 

Answay, given what has now been calculated, the beam "resolution" can now be mapped to an actual image size. This can then be mapped to the image as produced by the image source. Such mapping may well include showing the same source pixel on several adjacent scans or horizontal pixels.

 

And I'd have though that is it. Given that we have corrected for distorions in the landing dot beam size, why does one need to try and correct that with optics?

[gyro_gearloose]

Either I'm very wrong, or I'm failing to explain myself, or probably both :)

I think I could say exactly the same thing :D

 

Assuming the suirface is flat, a bit of math will give the size of the dot or elipse formed

Hang on a minute, didn't you just say that the size of the laser beam is constant regardless of distance? (I'm ignoring the differences in diameter caused by striking the surface at a slight angle, and assuming that the throw ratio is quite high)

 

The same math will also give us a number of pixels along a scan line,

Yes, but thats approaching the problem from the wrong direction. We already know how many pixels we want on each scan line. What we need to know is how big each pixel needs to be, and we work this out by dividing the horizontal resolution by the horizontal projection angle (which we already know, as it is defined by the movement of the h/v mirrors). Once we know how big each pixel needs to be, we can set its size using optics or a really, really small iris :) Optics would be better. I'll explain why in a minute. Its probably too difficult/not much point in setting the pixel size on a per-pixel basis, unless the throw ratio is quite low.

 

Answay, given what has now been calculated, the beam "resolution" can now be mapped to an actual image size. This can then be mapped to the image as produced by the image source. Such mapping may well include showing the same source pixel on several adjacent scans or horizontal pixels.

Again this is approaching the problem from the wrong direction, or at least not the direction I was thinking about when I wrote my last post.

 

Let me try to explain myself in a bit more detail. Imagine a triangle with the projector at the apex, and the screen at the base. This triangle is divided up into slices depending on the resolution of the projected image (we'll stick with a 2d representation for the moment to keep things simple). Each of these slices is a narrow triangle with its apex at the projector, and its base being part of the base of the larger triangle. With me so far? Good. Now draw some lines across the triangles, parallel with their base. What do you notice about the sections of these lines where they cross the triangle? The closer they are to the projector (apex), the smaller they are. What this means is that pixel size is different depending on the distance from the projector, which in turn means that a parallel laser beam won't work in this situation. While this could be solved by adjusting the laser beams diameter depending on throw distance, a better solution would be to make the laser beam diverge so that the projected image is always in focus, regardless of throw distance.

[dbuckley]

Assuming the suirface is flat, a bit of math will give the size of the dot or elipse formed

Hang on a minute, didn't you just say that the size of the laser beam is constant regardless of distance? (I'm ignoring the differences in diameter caused by striking the surface at a slight angle, and assuming that the throw ratio is quite high)

 

The beam is of constant size, but the size of the dot on the projection surface depends on the relative angles. At 90 degrees (2D!) it is a circle, at any other angle it will be not-a-circle, depending on the angle.

 

I agree everything else you say, I'm just musing whether we can fix the problem is software rather than in optics, along the lines of how keystone "correction" on a projector works. Part of this (in my crazy world) involves scaling so that we have many more projected lines (and horizontal "pixels") than the source image.

[beware]

I agree everything else you say, I'm just musing whether we can fix the problem is software rather than in optics, along the lines of how keystone "correction" on a projector works. Part of this (in my crazy world) involves scaling so that we have many more projected lines (and horizontal "pixels") than the source image.

Or add another scanning mirror so the pixels are made up of sub-scanning squares...