The dark corners are from the Diy Projector Company fresnels. If you look at the left image you will see a yellow splodge in the centre, it’s either a yellow centre or blue corners. It if a fault of this particular fresnel and it is the best I can get from them. So for lens shifting I wouldn’t recommend them. They are barely able to work in a straight line let alone keystone or lens shift. My avatar is actually a pic of these fresnels collimating light, it should be square.
Dark corners may appear when you try lens shift but you should be able to adjust the lamp or fresnel position to compensate. They are a result of the light being blocked by the triplet aperture.

DJ

Could someone tell me why is it 15 degrees the max keystone? (FF tilting method) I would say the only problem is the field fresnell being too far from lcd so distorting the image cuality. But if we take care of schleimpflug in all the adjusting procces, i don´t see why couldn´t be keystone done in more degrees.

This is the idea on how to adjust Real lcd and field fresnell (we need to know the virtual lcd position to do this)
Click to view attachment

Cool cool stuff!

From what I can see the focus distortion is very slim and virtually unoticable, especially if you sit back a bit.

Rox you've stumbled upon the same theory I did, I emailed it to frugaltech. I've tested the idea and it does work to some degree. However not quite as well as I'd hoped. Lens shift still seems to look better. What you do is simply project at an angle and tilt the collector to correct the shape, then tilt the triplet very slightly to correct the focus. This explains why so many claimed tilting the triplet helps. But the tilt is very slight.

Gadgetsmith I think a 330 mm fresnel will not work with out standard triplet. I was trying to find a larger set of fresnels to see if I could focus a 17 inch image with the standard triplet. I found some pretty big fresnels for sale with 330 mm focal length. I asked brain if it would work but he said no. We need something closer to 320mm. Now whether or not it works with the fresnel BEHIND the lcd I wouldn't know. I think its not just a question of distance, I think it has something to do with the angle at which the light enters the triplet.

Tommorow I should be finishing my projector. Just have to do the wire and switch work. Then maybe tuesday I'll make an adjustable triplet holder so I can lens shift on the fly.

(yawn) was bored so I drew this up. Its a lensshifting mechanism idea I came up with that utilzies two parallel supports with pivots on the lamp mount, fresnels, lcd and lens mount. This would allow very quick, very accurate shift adjustment and automaticly reposition everything properly. This might be more effort then its worth but I thought I'd post it for the heck of it. The box around it would be somewhat tricky. But of course you could make the parelell supports closer together and keep it contained without really increasing the size of your box.

This is the kind of thing only joechevvy would make. Lol.

Gadgetsmith.

Regarding the pincushioning, I've been playing around a bit. Try reversing the triplet. If the chromatic aberrations are tolerable there might be some improvement. Actually I am seeing slight barralling, less disturbing than the pincushioning. I think the FOV might be wider in reverse.

do you really think 330mm fresnell wouldn´t do the same work as 320mm field fresnell mikau?

330 fresnell has been used where 550mm fresnell was needed, i cant believe 330 mm isn´t good enough where 320mm is ideal, (anyway, could someone post why is it 320mm ideal?)

I wouldn't know, rox. Makes little sense to me. Thats just what brain said.

I was referring to a 330mm fresnel used behind the LCD. This was the original configuration prior to a newly designed fresnel. The new fresnels are not only larger but have a focal length of 317mm to facilitate being able to use them in front of the LCD, which allowed use of keystoning. If we forgo keystoning and use only lens shift to adjust the position of the image, there is no reason not to put the fresnel back behind the LCD (as long as this doesn't effect the quality of the lens shift itself). I like this method better as there is nothing between the LCD and triplet, so I should get the best possible image... of course if the lighting is even, and we're not outside the triplet FOV, etc, etc... Really it was just worth an experiment to see if lens shift would be as effective with the fresnel behind the LCD as it is in front. (I don't see why it wouldn't, but I never know...) As I said, I may try the 317mm fresnel behind the LCD, even though i'm thinking i'll get light loss, but it should give me an idea of how this arrangement will work.

gs

Nice. If thats true then we'll be able to use a 17 inch lcd with the standard triplet. And we can use lens shift instead of keystoning. Of course if we use a 17 inch the shift will have to be even less.

Yes, but the problem still remains is that the LL fresnels are not large enough for a 17". (well, unless you already have a pro lens set) I fear the fresnels from other sites may be lacking in quality, based on DJ's avatar !

hang on a sec I need to grab a link.

(edit)

Ok here's the one I found. Its listed on this page:


http://www.3dlens.com/largefresnellens.htm

Its got a focal length of 330mm and its 395mm by 395mm (about 15.5 by 15.5 inches.) This is definitly big enough.

If this works we may have at least a temporary solution for the dozens of people dying to make a 17 inch projector. Using lens shift we won't need to put the fresnels in front and you can still ceiling or floor mount your projector. Though as I said the shift would probably be limited. I'd love to know how much shift the pro lens could tolerate!

I should be a little more specific here. The fresnels I have are both the 330mm FL versions. The 220/330 FL combination may very well work correctly. I’ve only just discovered (last night) some information about fresnel design that maybe an alternative explanation to the crappy results I got.

Here is the most efficient way to set up lens shift.

DJ

The collimator fresnel does not need to be (and in fact should not be) offset, nor does the light for your proposed setup to work. Only the field lens and triplet need be moved for such a setup to work.

The problem comes in that the available fresnels are not large enough to make such a setup practical at the moment.

The collimator fresnel should be positioned like this, if we were using plano lenses then we could just lower the lamp. The lamp is at its optimal position, at the FL and positioned at the principal axis of the collimator. For a collimator fresnel to operate correctly the lamp should stay close to this position. Lowering the lamp away from this position can be done but at the cost of efficiency and you get to a point where the light becomes completely unusable.

3dlens.com fresnels should work for us.

DJ
(edit) I just read this post and it appears some has already patented the idea.
http://www.lumenlab.com/forums/index.php?s...pic=4591&st=200

dazzlas fresnell idea sounds nice. (i guess he doesn´t mean to place them as shoun on the image, i guess he means to cut it down when it is propperly placed, but the meain idea is that the centers of the fresnells are not aligned and this way, the lamp would be at the axial center of the 1st lens, and tje triplet at the axial center of the 2nd lens.

Anyway, the amount of shift on the 1st lens is lees than the shift of the 2nd fresnell.

I mean; triplet shift=2nd fresnell shift (both elements need to move same amount)
but 1st lens need to move 220/330=0.666 times the other shift and the bulb as well.

Yes. This is exactly what I was talking about in the first experiment that I ran with the 50mm shift, except, as others are also now pointing out, there is no need to move the collimating fresnel. With everything aligned as in the standard design I had shifted only the triplet, and then the Field Fresnel (in front of the LCD) and that gave very good results... except the FF was not large enough so the top of the image gets cut off. With a larger FF this may be a simple and very effective method. Putting the FF behind the LCD will only increase clarity (I believe, i've never had the FF behind the LCD, so I wanted to experiment) By my calculations I should be able to get an 80mm shift with a 15.5" fresnel height with and appropriate FL. (317mm FL if in front of the LCD and 330mm if behind the LCD ?)

gs

moving 1st lens down, would increase the light on the bottom side of the lcd a bit, this resumes into a more even image on the projection i guess. I mean that the light "evenness" is optimized on the two fresnells.

a fresnell lens will do the best work when the light is at it's focal, I believe dazzla's idea is fine, please if someone has an easy way to test it, give a try, and post result.

Hi Rox,
In my first experiment with 50mm lens shift these are the results I got. post 331 You can see with the triplet and FF both shifted up by 50mm the picture is evenly illuminated and in focus. The top of the image is cut off as the FF from LL is not large enough to shift, therefore I cannot verify light eveness at the very top, but by the looks of it, it will be very good. For an 80mm shift, as i'm planning, I would not be surprised to see some light uneveness, and therefore moving the collimating fresnel (or lamp) may be necessary. Just need to get some 330 fresnels.

yes gadgetsmith, I saw your results, they are very good examples.

I though an easy way to check this with no need to wait for getting larger lens, is it posible for you to rotate your fresnells 90 degrees on their own plane? you would losse left and right sides this way but you could check the eveneness on the top and bottom side.

maybe it is non practical to do this test in your setup (because of your attachements...) tell us, is it posible for you to try?

yes i can. that's a great idea for checking, never occurred to me and so simple! With the FF rotated 90° I could also shift from side to side in order to check light eveness in the corners.

thanks.

I'm going to try comparing lens shift to keystone correction. By tilting the triplet and finetuning for a while you can get pretty clear focus. So clear I'm wondering if lens shift is better, especially since we ae having some shape distortion at the top.

I'm going to compare the two and see what works better.

Another small trial. I moved the lamp and collimating fresnel back to the standard design position. I used the FF turned 90 degree's, which only allowed about 43mm of lens shift due to the size of the FF (only 12.5"). The area of the LCD inside the dimensions of the FF looked very good. Sharp focus top to bottom. Even lighting. I come to the same conclusion as my first test, if we had a large FF (something like 18-20" in heigth) with a FL of 317 or 330 we would have something. In my experiments there was too much light loss and the introduction of a bow at the top of the image associated with a lamp and collimating fresnel shift. IMHO, until we can get larger fresnels, lens shift will just not be a viable alternative for a 120" screen with PJ located on the floor. Anyone know where to get larger fresnel lenses, please let me know.

I am now back to keystoning, about 14°, with good focus from the top to near the bottom. The lower portion of the screen where the clock and start menu's are, are almost unreadable... this of course is due to the FF being too far from the LCD (due to required keystoning), and is effecting focus. I've had no success tilting the triplet to regain clear focus across the entire screen.

I'll keep my eye peeled for larger fresnels.

cheers,
gs

sorry i didn´t understand what you did with the lamp and the first fresnell. And what do you mean by standar design posiion?
Did you move the same amount on the second fresnell and the triplet?
Why do you loos light?
please clarify this dounts i have. Thanks

By standard design position, the center of the lamp, collimating fresnel and LCD are all on the same horizontal axis. (in other words no lens shift or keystone adustments) everything is like is should be based on the standard design outlined in the guide.

For the test I moved only the FF and the Triplet (both the same amount). I could only shift them up by 43mm. If I were to go higher the fresnel would no longer cover the bottom of the LCD (top of projected image).

I only lost light when I shifted the lamp and collimating fresnel poisition. Once these were returned to the "standard design position" the light was brighter and more uniform. Somehow the bow in the top of my image during my 2nd test (87.5mm shift) is caused by this change in lamp and collimating fresnel position. I spent some time moving the lamp and collimating fresnel to improve the brightness and eveness of lighting, but nothing worked for me.

hope this clarifies things.

ok I see now.

What was the amount moved on the lamp and the first fresnell? (i guess you moved down direction) did you move the same amount on the first fresnell and the bulb?

anyway, the 90 degrees turn on the fresnell, would give you 3" of margin at all (9" high 12" wide lcd) this 3" would give you teorical 76mm shift, why do you say it was 43mm the max you could have done?

(Aclaration FF=field fresnell? RF=rear fresnell?, if yes, we can refer this way to be more understabdable)

The height adjustments for the 87,5mm shift are as follows:
( '-' denotes lowering, '+' denotes raising )
( 'RF' = rear fresnel (collimating fresnel) ; 'FF' = field fresnel (front fresnel) )

Lamp => -58,3mm
RF => -3,9mm
LCD => 0mm
FF => +5,7mm
Triplet => +87,5mm

The real measurement of the FF is 12.5inches. (317,5mm)
Once the FF is turned 90° and Centered there is only half of the difference available for shift. 12.5 - 9 = 3.5in / 2 = 1.75in (44,5mm) Having the edge of the fresnel match the edge of the viewable portion of the LCD is not practical, so you can really only get about 43mm of shift (assuming a 1,5mm of overlap... which may not be enough)

Sorry, you are right, it is half of the 3" margin i though.

Lamp => -58,3mm
RF => -3,9mm
LCD => 0mm
FF => +5,7mm
Triplet => +87,5mm


Let's see, RF and FF, you mean 39mm and 57mm? (the coma is ok?)
So you mean, this setup was the one with uneven light, and then

Lamp => 0mm
RF => 0mm
LCD => 0mm
FF => +57?mm
Triplet => +87,5mm

this one improved the eveness on the image brigtenns?

What about turning the RF as well? Or leave them horizontal and do the lens shift side ways.

DJ

The comma's are ok. The RF and FF do not move much as they are close to my "zero" point, which is the LCD. (15mm and 20mm respectively)

No, the FF and Triplet both move the same amount with the Lamp/RF and LCD at 0mm.

See the following pics:

My setup for 87,5mm shift:
Click to view attachment

My setup for 43mm shift using FF rotated 90°:
Click to view attachment

This would be my ideal setup using a new larger FF. The question is would this work ? I don't know how much shift we can get out of the FF without losing brightness or introducing some distortion/loss of focus. I can't even really test this without the larger fresnel. Problem is, there isn't a fresnel out there like this one. Another option is to put the FF behind the LCD with the RF. In that case we would need a 330FL (I believe)... And i'm not sure this would cause other problem associated with the angle of light going thru the LCD.

Click to view attachment

The 43mm shift (similar to the 50mm shift I had done earlier) have very good brightness. The 87,5mm shift had light loss. I believe this was due to the lamp and RF moving.

then you havent tested what I was saying to test.

Click to view attachment

this is the design dazzla said and I would like you to test as well (basically your 1 pic, but the introduced shifts on the RF and FF need to be the same shift done on bulb and RF / FF and triplet.

the idea is that the bulb is at the axiz of the RF and the triplet is at the axiz of the FF. But they are shifted + X and -Y from the lcd axiz.

could you try it please?

I'm not sure what this would prove ? Larger fresnels are needed if we are to leave the lamp and RF in the standard position and do lens shift with only the FF and triplet... no matter what direction we shift.

Ah, I see what you are saying.

No I didn't try this exactly layout, but my layout for the 87,5mm shift is basically the same thing. I don't see what rotating the fresnels 90° would accomplish ? The fresnels appeared to be large enough for this layout.

Unfortunately my setup does not allow me to try your layout without having to cut my fresnels, which i'm really not willing to do at this time. Sorry.

Rox: your idea of moving the rear fresnel's center point won't make any difference at all in the amount of projection angle, all it will do is result in the bottom corners being a bit darker than the top corners.

The thing to remember here is that there are really 2 separate and distinct sets of optics in these projectors. The fresnels are one set, and their purpose is to get light to the projction lens. The projection lens is it's own set. The only thing that matters there is the relationship between it and the LCD.

Okay, there's a "sort of" tertiary set of optics as well, with the field fresnel doing double duty as a keystone lens, but I think we all agree that this is sub-par, and should be minimalised as much as possible.

Let's break it down a bit: What's the entire purpose of the collimator fresnel? It's to bend the light from the bulb so that it all goes through the LCD in parallel rays. Given that information, As far as the light going through the LCD is concerned, it does not matter where the center of the fresnel is, only that the bulb be at it's focal point. It is in the interests of making the fresnel as small as possible, and getting the light as even as possible that we center the collimator fresnel and LCD together. This still applies, even with lens shifting of both the field fresnel and projection lens. This is fundamental to the way that the projection works.

We work with the lamp at the focal point of the fresnel, because this gives us the most intuitively obvious result, but it is really only important that the lamp be in the focal PLANE of the fresnel. Anywhere in that focal plane, the collimator will still do the same job, but it's important to know what will happen. The light will still come out in parallel rays, but they'll be parallel to the line from the center of the lamp to the center of the lens. Here's the tricky bit... The spot that gets the most light will still be the spot closest to the lens.

well, that is exacly the opposite thing i am trying to fix.


It is not my idea, it was originally posted by dazzla, but i think He is Righ.

let me tell you. If we had lamp-RF-LCD all centered, the light peak would be at the center of the lcd. But since we shift the FF, the light peak would shift somewhere up between FF center and LCD center. If we shift the RF to the opposite direction, The light peak would be at LCD center as well (where i think should be). I think that if the light peak was not at lcd center, one edge would be dimer than the opposite edge. So this is what i don´t want to happen.

The following represents one of the things to look out for in using an off-center light. The light is still most intense around a center point, but where that center point is may not be where you think that it is. Light doesn't conveniently center around what we dictate the center light to be, it centers around the same thing that it always did. Not the center of the fresnel, but the point closest to the lamp. Ideally this should still be as close to the center of the LCD as possible. (I know that you can't test this without cutting your fresnels, and I'm not asking you to. But this is how it is.) So be aware that moving the position of the bulb is also moving the brightest spot in your image.

Can you get around it? Yes, by tilting the collimator fresnel and moving the bulb back to the focal point of the lens.

Click to view attachment

if 1 of each 20 of you understand what i will try next to say witth this strange image, i would feel happy .

this is the RF (blue,) and the FF (brown) drown. those would be the circles of each fresnell. Note they are shifted X distance.

The numbers indicate the brightness on each circle, the number1 is the brighest on the blue fresenell (RF), then goes to 2 to 3... decreasing brightness.

then we have the shifted FF, it will collect the light form RF and focus to the triplet.
this FF lens, has its own efficiency as well, the number 1 would be the most efficiency zone, then goes to 2 .. 3... (if it was even light, the brigthest point would be the number 1, but it is not even, the light distribution is the one RF has done).

So, what i would like to coment with you is the resulted union on these two systems. We have the magenta circles (few are drown) where would be the resulted brightest point at center, and then it goes decreasing....

So the conclusion is the hot spot has shift a bit because of the FF shifting. So i would try to reposition it with a negative shift on the RF, so the hot spot is at lcd´s fisical center again (just as standar setup with no shift).

Anybody understod what i mean?

This is only true for non-fresnel lenses and it is why the RF needs to be shifted as well. The fresnel is not directly compareable to a PCX lens.

DJ

You know I got some pretty satisfying focus by triplet tilting. Maybe its because I'm only projecting at about 7 feet at roughly 10 degree's but I put the projector about 4 inches off the floor and got very acceptable focus. Maybe it was easier because of the smaller projection size, but I don't really see why you would want much bigger since your just forced to sit back farther because of the screen door effect.

When triplet tilting, I think you have to remember not to simply focus, and then tilt, you have to readjust constantly forward and back and tilting more or less to find the sweet spot. I found I was perfectly satisfied with the focus if your patient and take your time to fix it. Also tilting doesn't mean parellell to the screen, its a very mild tilt, almost unnoticable, but it makes a huge differance in the image quality. I'll post some pics as soon as soon as I get some film for my camera.

I'm thinking about going the lens shift route again. I put my projector about 8 inches off the floor and lifted the triplet high enough to get the image above the floor. I noticed about a 1 inch bow in the top of the image, but thats hardly noticable on an 8 foot projection and I think the better focus it affords is well worth it. As soon as I get my new ushio I'm going to try lens shifting instead of keystoning to see which I like better.

And as I mentioned to gadgetsmith, a slight outward bow in the top of the image could be made to correct the bow if your that picky.

Also, I may be mistaken but I think I noticed a slight bow in the image when keystoning. This kind of makes sense since tilting the triplet pushes the field of view just like lens shifting does.

I think we need to do more experimenting with prisms (or glass wedges). I made one the other day out of left over pieces of lexan. Filling them with water and seiling them with cauking. Puting it in front of the projector the image moves up a bit and appeared to keystone a little, which told me it was just redirecting the image upward. It also put a slight bow in the top and bottom of the image. Putting it behind the triplet did roughly the same thing. I put it close to the lcd but the prism was too small to really notice what it was doing. It seemed that tilting the prism would compress or stretch the image, but also produce the bowiing out at the edges like lens shifting did. There is the animorphic lens that was built though i don't know how much that works. My goal was either to shift the image upward, putting the prisms in front of the triplet, or optically shrinking the size of the lcd. If we could do that we could lens shift more and also fit larger lcds in our projectors with the same lenses. It would mean a breakthrough.

Does anyone know how light behaves when passing through a prism? What the angle of refraction is when it passes in and out of glass? (I guess it depends on the material right?)

I just tried using a single 330 fresnel lens and move the light source back to correct for focal length. I can get 15" of keystone correction from 90" away (9.5 degrees) with the 135mm triplet. My experiment is with a 17" transparency acting as an LCD.

I am keeping the triplet, screen, and LCD parrallel. The light scorce, triplet and fresnel are on the same center axis. The fresnel is angled so that it is perpendicular to the axis from the light to the triplet. (ie light low, triplet high).

I will be rebuilding my 15" projector as time allows, but I thought I would share this information with the group.

Can someone with a Pro lens try this?

Any pics?

DJ

"I just tried using a single 330 fresnel lens and move the light source back to correct for focal length."

(slaps his forhead) now why didn't I think of that?

How did the focus look?

Btw, I noticed that when keystoning to my desired degree, a slight bow appears at the top of the image. I think this is because we are again pushing the triplets field of view. So I think I will try the lens shift route again since it makes little differance.

Besides, the 1 inch bow is hardly noticable on an 8 foot screen.

Here is the setup. A 250 watt work light (not ideal for this) one of my first screens. Sorry the pictures are of such poor quality. Cheap digital camera ( I have to cover the flash)

Here is the setup. A 250 watt work light (not ideal for this) one of my first screens. Sorry the pictures are of such poor quality. Cheap digital camera ( I have to cover the flash)

Click to view attachmentTry this again...

Click to view attachmentFocus at corner

Click to view attachmentSee triplet offset