New theories for keystone correction, merged topic Options

[Mikau]

Well techniclally this isn't keystone correction, more like idea's to project at an angle to get the projector out of your viewing area.

Heres what I was talking about when I talked about optically shrinking the size of the lcd, then using that to project at an angle.

This would require a "special triplet" that would project a 7 inch image. This would be tricky because you'd have to use a lot of fresnels to keep the light where you want it, as you can see, this uses two lens kits.

Attached File(s)

 keystone_theory_1.JPG ( 24.89K ) Number of downloads: 7

 

[Mikau]

Idea 2!

I'm sure this idea doesn't work, as I'm sure someones tried it before. But what the heck.

In this you place the triplet off centered from the collectors focal point, so all the light is pointed above the center of the lens. This might result in an angled projection.

Smaller projection size would probaly allow you to off set the lens even more.

This might also require you to tilt the triplet back or forward a bit (its kind of hard to explain why I think you might)

So anyone who has a lumenlab projector could try this idea, (don't have mine yet)

Just off center the lens a little bit and see what happens. Try doing it with a smaller projection size, about 6 to 8 feet.

Attached File(s)

 keystone_theory_2.JPG ( 12.85K ) Number of downloads: 7

 

[Shrivel]

The ONLY way to adjust keystoning with the standard design is by tilting the fresnel. Moving the triplet lens simply dims the picture. Tilting the screen, LCD and/or the triplet will focus the picture (or defocus - depending on your perspective) in the direction of the tilt.

[Mikau]

Theory 3 is sort of a mix of the first 2.

A triplet that focuses an image very very close and reduces the size by half, before it goes to the projection triplet which is offset.

Attached File(s)

 keystone_theory_3.JPG ( 12.66K ) Number of downloads: 7

 

[Mikau]

hmmm maybe so. Just idea's. But remember this isn't keystone correction, its just theories to allow you to off set the projector, the lens is still perpendicular to the screen.

[Mikau]

Besides this isn't the standard design I'm using at all.

Anyway, I'm sure there the people at lumenlab who know 100 times more about optics could figure out a better way, but would they want to? Maybe research isn't being done for better keystone correction/off set projection anymore. Maybe they are satisfied with the quality of 15 degree keystoning. Or are occupied with something else right now. Maybe they know of an expensive solution but figures no one would bother to use it.

Its a matter of preferance, me I'd be willing to spend a lot more money to get the projector out of the line of viewing. I'd also be willing to make my projector a lot larger.

So maybe some of these idea's work but they haven't been brought to light because they aren't very practical.

[Mikau]

Heres another theory.

I've discovered that when you project a straight image onto an angled surface, and place a lens in the path of the projection parrallell to the screen, it corrects the projection. As shown in the top diagram below. However my experiments did not use high quality lenses and there may have been distortions I didn't notice, but lets just assume there weren't. This is actually similar to tilting the collector fresnel if you think about it only you are using a lens instead. This seems like a better idea to me because fresnels are not great for changing the size of the image, they are only ok for bending light. The reason keystone correction works is because the fresnel makes the top of the image larger and the bottom smaller, which will be reversed after it passes through the triplet. So maybe tilting one or more of the lenses in the triplet will work the same was as tilting the collector. NOTE! You could only tilt the midle and back lens of the triplet, the front lens would have to be parralell to and centered on the lcd according to my theory.

Diagram 2 shows the triplet with lens b and c tilted back slightly, however there are many combinations you could try. Tilting only lens b, tilting only lens c, tilting lens b and then tilting lens c a little more then lens b, there are lots of combinations.

I don't know how hard it is to dissasemble a triplet, but I know one guy here who got a faulty triplet and dissambled it to try rearranging the lenses, so I guess it must be doable.

If titling the lenses inside the triplet doesn't work, then maybe some type of lens placed in front or behind the triplet and titled will do it.

Again, I think a smaller projection will allow greater angled projection then a larger one.

Maybe I should post this stuff in the mad scientist forum.

Attached File(s)

 lens_tilting_idea.JPG ( 25.49K ) Number of downloads: 7

 

[Squalish]

What about using flexible reflectors curved in 1 dimension (cylinder arcs)? I'd imagine that if you could get reflectors that didn't totally screw your image, you could get effective keystone adjustment from that.

Image: Projector is 3 feet from a wall, projecting upward @ a tilt onto a very slightly concave cylindrical mirror (vertical). The beam then goes back over the projector at the opposite wall. Since the bottom edge of the beam hits a shorter angular curvature, it isn't compressed very much. The upper edge of the beam hits a larger angular curvature and is compressed more. For this to be conveniant, you'd need a relatively low-angle beam, of course.

Attached File(s)

 setup.gif ( 19.38K ) Number of downloads: 7

 

[Mikau]

That might work . But getting a mirror to the exact curvature wouldn't be a bit easy. A larger mirror would be easier though. Maybe bulding a frame for it.

[Squalish]

What about using the 1/8" acrylic mirrors that are being used for Light Fusion? One could get a small one + experiment with framing/mounting options. How one could consistantly apply the pressure, I know not.

I suspect, though, that you would be forced to use steel throughout, and keep the mirror at more than a slight bend. I think clamping it from both sides is going to be the only way to equalize the pressure so that it won't distort vertically. Then, the clamps could be mounted to vertical tubing and it could be contracted via adjustment of threaded rods at top/bottom.

Anyone up to try this? Or have more ideas about the optical theory that could produce good keystoning like this?

[Squalish]

Also, I know a lot of you have fresnel-adjustable keystone built into your PJs.

What are the downsides of having it enabled vs having it disabled? I would imagine increased blurriness proportional to distance from the LCD, and decreased light output proportional to angle the fresnel is at relative to the LCD.

[Mikau]

Good idea's, it just might not be easy to get the correct angle, experimenting would be good.

I like the idea though.

Someone needs to try my triplet tilting idea.

[Mikau]

Actually, using a collector or collination fresnel and putting it in front of the tripet off cented, might work the same way as the mirror.

[Mikau]

On second thought it probably wouldn't, you need something to bend the light, not just redirect it.

[Mikau]

Maybe something more like this. A fresnel that is just a bunch triangles, instead of bending the light, it just shoots it out sideways. But I think how well this works would depend on how small the triangles are, and they may need to be super small.

Hmmmm..it would probably need to be one triangle per pixel, so you'd need 768 triangles in a 9 inch tall fresnel. It could be easier if the fresnel was bigger.

(edit) Wait, that probably wouldn't work, because the fresnel is still centered on the middle of the screen. PHOEY! :angry:

Maybe you could try using one of those convex mirrors you see in supermarkets, those are easy to get.

Attached File(s)

 fresnel_theory.JPG ( 11.62K ) Number of downloads: 7

 

[Squalish]

Fresnel fabrication... not gonna happen

Concave mirrors... After examining one when I saw your thread, they're very very rough - you have regions that compress and regions that don't. Anything over a few inches away looks terrible because of nonuniform magnification.

[Mikau]

I just found out that some of the newer overhead projectors have optical keystone correction. I'm sure we can get a hold of one or at least observe how it is done and duplicate it. The one I read about had keystone correction up to 13 degree's, which isn't really much, but there might be better.

[drkilldrmz]

Can you use a laptop lcd screen in this type of configuration?

[Dergrin]

I wish I had more optics knowledge. It seems that this is the area that all the vast improvements will be made in DIY projectors. I am lucky enough to go to one of the four engineering schools in the US that has an opical engineering. And I have one friend that is about to graduate in optics. I have talked to him about some of these lenses and such, and the only problem is the application we are using these optics for are a little different from his textbook optics but I will show him your ideas and see what he says.

[Mikau]

Thanks man!

Heres a website I found called info opctics or something like that. Heres a quote:

"Keystone distortion that occurs in overhead projectors when the projecting lens head is tilted upward to a high screen is commonly observed. Here we suggest a modification of the typical overhead projector to eliminate this distortion of the image. © 2004 Optical Society of America"

It says "Download pfd document" below, but you need to be a subscriber to download it.

(edit) Heres the link:

http://www.opticsinfobase.org/abstract.cfm?id=80577

[eudaimonia]

It seems like much of what you describe is "lens shift" which is common in most higher end projectors and is now making its way to entry level units. You might want to take a look around at projectorcentral.com or follow these links..

http://www.hometheaterpeople.com/tutorials/lens-shift.asp

http://www.projectorcentral.com/anyplace_projection.htm

[Mikau]

Right! Lens shift! Looks like a lot of projectors now have lens shift capability. Time for brain to develop a triplet with lens shift capability.

BRAIN!!!

[Agent707]

I'm pretty content with just tilting my fresnal. Maybe you should give it a try. It works well if you also align your triplet with it.

You forget that when you run your image through all those optics, your just dimming the image more and more.

[Mikau]

Didn't forget. I know it will.

Tilting the triplet to align it with the fresnel?! Thats the same as just tilting the lcd! Thought that didn't work. Obviously you can't simply tilt the lcd or the back lighting will be messed up, but still thats pretty much the same thing.

[Agent707]

It works indeed. Though my system is only doing 7° Keystone adjustment. I have my screen angled at around 7° also to make up for the difference.