Ok charlie, heres what I was talking about in that pm.

You can see how the rays from the triplet are compressed by the mirror thus shortening the focal distance. The rays intersect before it is able to reach the second mirror, to correct the abberation the first mirror produced. Like I said, to prevent this from happening we need either a less extreme curve or if we're lucky, the rays from the triplet move in at a less extreme angle.

Remember, all we need to do is find a way to turn the projected image, into a trapazoid without effecting the focus. But this needs to be done after the light leaves the triplet.

I'm going to try some drawings using a single elipse to see how it effects the image. I want to shy away from anything thats not a regular curve but I'll just see what I can do.

I don't think there is a single pixel ray when it leaves the triplet... instead a bunch of rays, shaped roughly like a cone, with the widest part at the triplet and the narrowest part at the screen.

Only if we had a tiny point source of light would the pixel create a line-shaped ray. In that case we wouldn't need a projection lens, since there would be no need to focus any light beams.

Thinking in terms of cones (beams?) instead of rays has been a real struggle for me... I hope what I'm saying isn't totally off-base... but it seems that the projection lens is all about ratios, 1/f = 1/d + 1/D... f is fixed, d is affected by the field fresnel, and D is affected by... curved mirrors in the light path?

I'm thinking of getting a copy of Feynman's lectures on optics, based on a recommendation from a diyaudio member.

Yeah, I'd expect their are an infinite number of cones coming from the triplet. I wonder if they are as wide as the entire triplet...

At any rate, unless you can recreate an accurate simulation of how the triplet functions, I'll probably just have to get an acrylic mirror and test it (physically) to see what happens. If that doesn't work, I'll see if I can do anything with the animorphic lens design.

Well charlie, I was happy to find that I HAVE an acrylic mirror in my house, never knew it was acrylic.. anyway I tried a simple projection by bending it with my hands and it didn't go to well. Of course bending it by hand is hardly accurate, but i've also been reading up on mirror reflections. It seems a perflectly round mirror causes distortion. No matter what. However I did read that parabolic mirrors work (exponential curve )

Anyway, I'm going to try doing the concentric mirror test again, this time first going through the convex mirror (instead of going through the concaved first.

Charlie, I'm begining to wonder if that fresnel bending idea would work again. Originally I ruled it out because I thought the angle of the fresnel riderected the light path differantly. But since you can tilt the fresnel without moving the triplet, maybe this isn't true.

So maybe bending the fresnel will work. It may work by itself, or together with the schleimpflug adjustment.

Heres what I'm talking about, charlie. I need the total length of line abc, line dec and line fgc.

A lot of cunfusion, charlie. I'm trying to figure out where the distances from the triplet line up, or don't line up. I'm trying to figure out if the distance is a straight line or a curve. But heres the tricky part. When you measure the light backwards, the middle beams are shorter then the outside beams, but once it reaches the collector the light is redirected and they both head for the lcd with even length distances. Shown on the top left of the pictures below. If you straighten out these rays it ends up producing a curve. Shown on the picture next to it. But since this doesn't cause distortion like this, I believe that when the light is moving straight forward, the rays can be even to not throw off the focus. When they are spreading out, the outside rays must be longer then the inside rays to stay in focus.

If tilting the collector pulls one side of the lcd away as if you had simply tilted the lcd, we can work with that. But if tilting the collector produces distortion as if we had used a curved lcd, then thats trouble.

Its hard to explain where I'm coming from. Do you get the picture?

The other two drawings of projections are possibilies for how using the schliempflug adjustment, the top by using a page magnifer in front, with the collector and collimator behind the lcd, and the other is simply tilting the collector.

I think the second ? drawing is the better representation. And based on what you've written, I'm thinking the tilted collector indeed does produce distortion as if we had a curved LCD. I'm guessing the reason why it still works, is that the distortion is small enough that it's unnoticeable on the screen once the projector is well tuned. Perhaps the radius of distortion is smaller than the projected area of one pixel?
Taking this guess a little further, I'd say the larger the lamp arc, the less tilt can be used. Before the focal distortion becomes noticeable. Why? Because the larger lamp arc creates a wider cone for each pixel, allowing less margin for distortion.
I wonder whether projecting onto a screen which is slightly curved, matching the "curved" LCD, would cancel the distortion? Hmm

Sure it would. But I doubt it would be a regular curve.

Which one is "the second drawing"?

I'm really curious about if a page mangifier will do the trick. I did test it on a tv screen and it worked down to the pixel, but the light path on a lumenlab projector is a lot differant.

The weird thing is just because we see something one way, it doesn't mean the triplet will. Sometimes something looks one way but you throw a lens in front and try to project what you see and it doesn't work.

Well my idea has been tested in a projection at least.

When you put a page magnifer on a piece of paper flat on top of it, there is no noticable magnifaction. But when you pull it away it begins to magnify. I'm hoping it will work the same way in front of an LCD.

Charlie I still think we may be able to bend the collector, or better a page magnifier to correct for distortion. Bending the page magnifier I found I could fine tune the shape of the trapazoid without distorting the focus. So maybe we could do that if it produces distortion.

Can't wait to get my projector running.

Hi Guys,

Just a note from a friend.
When confronted with one of these questions, I search the patents on USTPO.GOV and can usually find a decent answer.

Try looking at Keystone correction for overhead projector (it is patent number 04436393) and click on the "Images" button for more info and drawings. Should work beautifully for Lumenlab projectors.

Have fun!

T.

P.S. You might need to download the patent offices document viewer to view the images. It is at TIFF Viewer.

Pictures won't load up on my comp. But from what I'm reading, it seems to be more lens shift then keystone correction. Using the method me and gadgetsmith have been toying with.

It is keystone correction using lens shift.

You need a TIFF viewer to see the patent images. The link to the one the patent office recommends is in my post above.

T.

I downloaded and converted from Tiffs to Jpegs, hope they are of interest.

">

Sorry, didn't know they would end up so big

This is a great find! Very interesting information in there about "color tuning" and moire.
Thanks for posting it!

Though its kind of cryptic. Those pictures don't make a whole lot of sense to me.

I can never understand those patent documents. They talk too wierd.

Look at the ripples in the fresnels.

It's showing exactly what you described a few pages ago - just shifting everything past the LCD(the field fresnel and the triplet) up/down by an arbitrary amount. It goes on to talk about moire effects that can influence the picture from an arrangement like this, and a ratio of fresnel pitch (or whatever you call the size of the ridges) that minimizes moire.

I see. But still, I think we'd need a triplet with a wider field of view.

I really would like to know what the pro lens' field of view is.

why don't you build, or call on a builder who is building a smaller 7" or 8" lilliput and test the results with that, that should be more than enough to test cutting the fresnel, and the standard triplet would work well for that too

[im building a 17" widescreen right now and im going to have trouble getting my whole image through the standard triplet in a straight line]

good research, nice to see you sticking with it!

I have no doubt a smaller panel would work. But then you loose resolution and contrast.

well... maybe if somebody made a working lens ajusted projector it might inspire brain more to look into... it would be very expense for brain to look into a new larger triplet, especially without seeing any results.

there is no reason that every 7" can't have a lens ajustment, but somebody's gotta start. I have a feeling that if somebody does it and it works it could start a trend in the diy community that might carry over into larger models as well.

I don't think brain is going to worry about having larger triplets made just because a couple people are asking him too, why not show him some results first

With the smaller panel you lose resolution for certain, with the Hami being 800X600, and that being one of the best for resolution. For contrast, I'd think that would be dependant on the panel, which may be as good as the 15" panels. Certainly a large offset would be available for an 8" project. I want to build a test rig to see for certain what kind of FOV is available for the standard 15" lenses.

Oh and with the fresnels...

The fresnels do affect focus, but more as a side effect, rather than by design. The side effect is kind of like a person wearing glasses, in that a lens placed close to another lens (Within the focal distance of the second lens) will make a modification to the effective focal length. The primary reason for using the fresnel is to direct light at the triplet. As a secondary effect, it is used for keystone correction, This is a simple magnification effect of the fresnel, which increases as the distance from the LCD. If you curve the fresnels, then you'll get odd curvature in the projection.

who can do this by himself !?

http://www.mir.com.my/rb/photography/compa...mmTS/35mmts.htm

There seems to be lots of camera lenses with all sorts of unique abilities, but not projection lenses.

Wish I new how to ray trace light as it passed through prisms.

I'm wondering if we could place the collector behind the lcd and maybe put a page magnifier fresnel in front of the lcd. (A fresnel with curved cuts instead of flat ones) maybe that would magnify and correct the focal length as well.

Mikau, I think you are describing something I tried a while back. The idea was to create a lens that would only magnify the top of the LCD and not the bottom. I came up with the lens/prism in the pic. Its shape is basically slithers of different lens curvatures stacked on top of each other to create a gradual increase in magnification, bottom=no magnification and top=max magnification. I made it with two sheets of polycarb filled with water. It was to be placed in front of the un-split fresnels and LCD so that keystone could be corrected without the fresnel rings degrading the image. It worked beautifully BUT it distorted the projected arc image so badly that it would not go through the triplet. I also tried it close to the triplet, this only introduced horrible aberrations into the triplet so I scrapped the idea.

DJ

Here is a good site for learning optics.
http://www.physicsclassroom.com/Class/refrn/refrntoc.html

DJ

Thanks. But the fact that differant colors refract at differant angles complicates things.

Yeah I wanted to make something like that with mirrors. To stretch the top and leave the bottom, but slight curves produced enormous distortion.

We really need professional help.