DIY anamorphic lens

[MikeP]

Originally posted by roar


I thought there was one picture here of someone who had epoxyed something to the bottom of his lenses? A post over on avs detailed a great case that 'sandwhiched' the lenses together, using a thin 'foam-like' fabric to help keep them together. Check it out here...

http://www.avsforum.com/avs-vb/showthread.php?t=739315

These are the different lenses though that the aussies are using, not the ones we have here in N.A.

Here is one of the pictures from that AVS thread. It's a nice looking enclosure, although I'm personally more comfortable working with wood rather than metal.


But my real question is: Hey Mark, do you still have any of those CAVX plates available? They sure give it that nice professional touch

My only problem with wedging the lens between the foam is that you'll end up needed to tilt the lens to fix the geometry of the picture. For ceiling mounted projectors, that means tilting the lens down, and I'd be worried about the lens' falling out or getting out of adjustment. I'm probably going to do something more like Steve did (I think it was Steve) and epoxy some plates with some threaded rods so that I can adjust the angle of the prisms when they are mounted. And then they can't fall out.

[Mark Techer]

Like the CAVX badge? I had a heap made for my speakers and still have a bag full, so decided to badge the lens as well.

Michael decided to ditch the MDF case and make his own alauminium case which looks really cool...

Originally posted by dracul
Also,

Have you noticed any increase in brightness on the screen? If so how much?

See my BLOG and check out the photos from SW4...

Originally posted by MikeP


Excellent! Now that's a picture everyone can related to. After all, it's probably a scene that most people have seen many times. It shows the black bars really nicely, and also shows the stretch with R2. Very nice!

The color shift is also very interesting. I'm glad you found a scene that would show this difference. I was wondering about it, but hadn't played with it yet. I'd be interested in seeing what calibration values changed and by how much. Like whether one specific color is being effected more than another, and if everyone sees the same effect (since we have different prisms).

I watch a lot of regular HD material, so there are going to be many times that I will have the lens move out. So it looks like I'll need to do 2 different calibrations and store them in the projector presets (lens-in, lens-out).

Yeah, well I didn't notice it that much until I took a look at the photos. All calibration is done with the lens in place, even the 16:9 mode, becuase to watch 4 x 3, I can actually use the 16:9 mode which stretches the 4 x 3 image to full panel, then turn the lens around and Horizontally Compress the image to restore the geometry...And it is really bright!!!

Basically I shot the "scope" shot first, then removed the lens, took the "scaled" shot, then switched to the 16:9 mode to capture the letterboxed image. On screen to my eye, there was not that much difference.

The camera is a Fujifilm FinePix S5600, was mounted on a tri pod and set to auto with the flash off...

Mark

[Mark Techer]

Hey guys, there is an interesting link to an artical at AVS in the 235 section about "lens flare" so it appears that even the pro's suffer from this too.

HERE is the link...

Mark

[Bud Bray]

Kewl, now I don't feel so bad. Was watching a variety of movies and deceided the ghosting really isn't that bad. On dark movies like DaVinci Code, you see it a bit and the Cars credits you see it but for the most part, you really don't. Still gonna give the 2 prism setup another try, will probably stick with the 4. Nothing's perfect.
Bud

[Steve Scherrer]

Originally posted by Mark Techer
Hey guys, there is an interesting link to an artical at AVS in the 235 section about "lens flare" so it appears that even the pro's suffer from this too.

HERE is the link...

Mark

Good find--I had reviewed Wikipedia's "Anamorphic Lens" entry, but failed to follow up the Anamorphic Widescreen entry. Good reading...

One thing I wanted to ask, especially from those of you that have access to professional lenses, is what characteristics of professional lenses are we not reproducing in these DIY lenses?

I can think of a couple--any others?
1. No anti-reflective coating on the surfaces of the prisms (Mark--any updates on that cryptic entry you made a few pages ago?)

2. Failure to nail the placement of the lenses (although I would say this is a weak difference, considering I would bet that Mark has probably hit the sweet spot with his lens (and I am working on a way to easily dial in the proper placement, angles, tilt, etc. in my housing).

Any others? It's just shocking to me that with these prisms, we can duplicate the effect of effect of spending upwards over $1000 on a piece of professional equipment, all for less than $100. And without really suffering a real decline in quality (except for a couple of stray reflections, that still might be solvable...) It seems too good to be true...

[Mark Techer]

Originally posted by Bud Bray
Kewl, now I don't feel so bad. Was watching a variety of movies and deceided the ghosting really isn't that bad. On dark movies like DaVinci Code, you see it a bit and the Cars credits you see it but for the most part, you really don't. Still gonna give the 2 prism setup another try, will probably stick with the 4. Nothing's perfect.
Bud

Yeah I agree. My GF doesn't even see them when they are on screen. I put it down to the same as "rainbow" with a DLP. Some people don't see it becuase they don't know what it is that they are seeing...

Originally posted by Steve Scherrer
Good find--I had reviewed Wikipedia's "Anamorphic Lens" entry, but failed to follow up the Anamorphic Widescreen entry. Good reading...

One thing I wanted to ask, especially from those of you that have access to professional lenses, is what characteristics of professional lenses are we not reproducing in these DIY lenses?

The key differences between our DIY lenses and a pro lens starts with the type of lens where pro lenses are mostly Cylindrical not prismatic. There is also the matter of stretch, where we are working with video, therefore should be obtaining a 33% stretch where a typical film lens is 2x stretch. We have found that by tilting the prisms, that we can correct edge distortion. I am likening this (in some sense) to the anstigmatism (not sure if that is the right word) correction which allows the pro lenses even uniform focus from corner to corner. I am refering to the ISCO III here btw...

Originally posted by Steve Scherrer
I can think of a couple--any others?
1. No anti-reflective coating on the surfaces of the prisms (Mark--any updates on that cryptic entry you made a few pages ago?)

No I have not yet recieved responses to my email enquiries...

Yes the lack of anti-reflection coating is what is causing grief with "green streaks" or is that "lens flare", but as we can all agree, it is not too bad...

Originally posted by Steve Scherrer
2. Failure to nail the placement of the lenses (although I would say this is a weak difference, considering I would bet that Mark has probably hit the sweet spot with his lens (and I am working on a way to easily dial in the proper placement, angles, tilt, etc. in my housing).

Angles of the prisms are best or easiest to get right from simply having a screen with the correct AR - that being 2.37:1 when using a 16:9 projector. If your using a 4 x 3, then a screen of 1.78:1 is required...

Originally posted by Steve Scherrer
Any others? It's just shocking to me that with these prisms, we can duplicate the effect of effect of spending upwards over $1000 on a piece of professional equipment, all for less than $100. And without really suffering a real decline in quality (except for a couple of stray reflections, that still might be solvable...) It seems too good to be true...

It has been a fantastic find. I am loving my lens. It is money well spent. No way would I give it up now...

Mark

[Steve Scherrer]

Originally posted by Mark Techer



The key differences between our DIY lenses and a pro lens starts with the type of lens where pro lenses are mostly Cylindrical not prismatic. There is also the matter of stretch, where we are working with video, therefore should be obtaining a 33% stretch where a typical film lens is 2x stretch. We have found that by tilting the prisms, that we can correct edge distortion. I am likening this (in some sense) to the anstigmatism (not sure if that is the right word) correction which allows the pro lenses even uniform focus from corner to corner. I am refering to the ISCO III here btw...

I was thinking more along the lines of Prismasonice and Panamorph, which use prisms, than cylindrical lenses like ISCO III.

[Mark Techer]

Well the only thing I can identify is the multicoating for the prisms.

I've never actually seen a Panamorph lens but have played with a few Prismasonics.

Panamorph claim to have corrected for anstigmatism, but how does one correct for this in a prism based lens apart from the tilt factor we have discovered? The is also no loss of focus if you space the prisms or have them touching unlike the elements in a cylidrical lens which require a certain space to focus.

There was a poster back about page 39 (where I first came here) who basically ranted that this was not possible, that we were all waisting our time using prisms. He seemed fairly knowledgeable about optics and went with a custom set of cylidrical optics made in china that cost him around $100.

While it is OK to state reasons why something might not work, it was really annoying to me when he would not part any info on his solution - yet really blue the trumpet hard about how good his cylindrical lens was over a prisms based one...

Mark

[MikeP]

Yeah, my InFocus 7205 arrived! Now I can get serious about experimenting with these prisms.

As expected (and fortunately), the 7205 lens is smaller than my previous PLV-60, so these "small" prisms are just the right choice.

It really is very easy to set the prisms up in front of the projector and adjust them as Mark has mentioned, especially with only 2 prisms.

I started with some test patterns. I found this to be *very* useful. The most useful pattern was just a grid of white lines on a black background (spaced about every 16 pixels). Since I have an HTPC plugged into this, it was easy to create a pixel-perfect grid pattern for testing.

The results of this test were dramatic! I discovered that not every prism is "perfect". In fact, two of my four showed some significant problems. When a pair of prisms is set up to perform the horizontal stretch, two prisms would cause certain areas of the grid to be blurry. I've cleaned all prisms with Windex and have blacked-out the sides using black electrical tape. The blurring areas are not coming from smudges or anything like that.

The right and left edges blur more as you get further towards the edge or corners, but that is the normal, expected CA. With one of the "bad" prisms, the blur occurs just left of the center. With the other bad prism, there is a long area running diagonally from the lower-left, through the center, and towards the upper-right, that is all blurred. The other two "good" prisms do not show any extra blurring at all.

So I highly recommend that everyone gets some sort of test pattern like this to test the optical quality of their prisms. Some of the people who are getting worse results might have some of these "bad" prisms. I think the bad prisms just don't have a perfectly flat surface. I notice when I try to put them face to face that they don't fit perfectly.

I'm glad that I bought four prisms so that I could get two that were really good!

Using this same test grid pattern, I then tried a 4-prism lens, even though I knew two of the prisms were bad. These imperfections seemed worse with the 4-prism setup. There were out-of-focus spots all over the place!

By "out-of-focus", what I really mean is that the white line of the grid is being split into a rainbow, and you can see a red tint on one side of the line, almost like an LCD or CRT that is out of alignment. From a distance, this looks like a blur, and looks similar to it being out of focus. This happens in regions, so that most of the image is focused, but other regions show this CA.

Obviously, this problem is hardly noticeable with normal movie images. So remember that I'm being *really* picky here. Even the worst out-of-focus area is still better than the convergence problems my old LCD was having. But the whole idea of the test pattern was to illustrate any imperfections in the prisms.

I decided to start with just a 2-prism lens using the 2 good prisms that were identified in this test. I then put in some movies.

We've talked about the credits from Cars to show the reflections. Well, I found another good test: the beginning of most any Bond film. I was watching GoldenEye (the best one that Pierce Brosdan did). At the very beginning of the movie, the classic white spot-light circles move from left to right, followed by shooting one to turn it red. Well, as these white circles on the black background reach the right side of the screen, you can see a dim reflection on the far left side of the screen. Just like with the Cars credits.

However, notice that I saw this using the TWO-PRISM lens! So it's not just the 4-prism lens that has the reflection issue. I verified that these are not the reflections that are easily removed by masking the sides, or by building an enclosure. This reflection is coming from the internal surfaces and is inline with the main image path. So until we find a way to treat the surfaces of the prism (without messing up their already excellent surface), we are stuck with this reflection I think.

I didn't find the CA with the 2-prism lens to be any problem at all. In fact, even with the minor reflection during that one scene, I was so impressed with the image quality that I think I'm going to join Mark and stick with the 2-prism lens. The 4-prism lens just seemed to have too many more variables to get it adjusted just right and seems much more sensitive to problems with the prisms.

I'll probably get out my Avia disk tomorrow night and play with some more test patterns and see what else I can learn.

But I just wanted to encourage any new readers that might be intimidated by all of this: go buy 2 prisms! These reflections and CA problems that we've been obsessing over are really minor. The prisms are easy to set up and really provide a spectacular image.

[Mark Techer]

WOW what a good post MikeP

I'm looking forward to reading about how you build a case and seeing some screen shots. Your centainly taking the "scientific" approach here...

Mark