PDA

View Full Version : Brainstorming for a collimated mirror display



s4sha
12-17-2005, 03:35 AM
Ok guys, admit it - - we are always looking for some way to improve our "pilot's perspective" when simming after a long nine-to-five. I personally use a DLP projector with both MSFS and X-Plane. It is definitely an improvement over a small CRT or LCD panel, but of course does not provide the true depth cues needed for a "real-as-it-gets" flight sim. I've been studying the collimated display design for years, and have all but given up on that level of display for the average simmer. "All but" meaning that I have an idea that may be worth investigating. I am planning on trying this myself after the holidays, but I thought you all might find an interest and expand upon it...

Has anyone ever tried something related to the following:

Build a plywood/plexi enclosure to accomodate a large section of mirror sheeting with both vertical and horizontal curves to provide needed "radius of curvature" for a flight display. The "box" would need to be sealed with silicone or a similar material, then again on the concave edges that would contact the mirror sheeting. Vaccuum would be provided by a vacuum pump, with possibly a relief-valve if needed. Amount of curvature of course would be determined by the location of the pilot in reference to the mirror.

The actual projection screen is another interesting idea, as it would need to be suspended/supported above and behind the pilot, with adequate curvature to reduce distortion. I thought maybe a thin bedsheet draped within a convex frame may do well, or maybe as far as soaking a sheet in a clear polyurethane-type coating that would help with rigidity. I have experimented with my DLP projector and there is quite a bit of forgiveness when projecting on a curved screen. Not near as much blurring on the edges as I expected!

Well, there is a challenge for my fellow simulator addicts - build a true collimated display!

Like I said, I will try this and may end up sadly disappointed but maybe you guys may be able to pull it off! I just couldn't help but share this idea.

Here's some links to get you started:

www.mirrorsheeting.com (the 7ml 10ft should work well)
http://www.harborfreight.com/cpi/ctaf/Displayitem.taf?itemnumber=3952 (would multiples be necessary?)

Keep on simming,

s4sha

Trevor Hale
12-17-2005, 10:51 AM
S4sha,


What I have done is similar to your second idea, and I find it gives excellent depth. I have placed my Projector aprox 10 feet in the front of the screen. So it shines directly at the pilot sitting in the box. I took a sheet that was purchased at wallmart (aprox 200 Thread count). and placed a few staples at the top and then at the bottom dead in the middle of the framethen pulled the screen tight and on about a 30 degree angle (where the sheet meets the wall on each side)I placed another series of staples all the way down keeping the sheet tight. then along the top and the bottom I used a couple staples to get rid of all the wrinkles. See the ascii pic below for a better representation.


http://www.fsflightservice.com/sheet.gif



This is not a rounded curve, however from this top down view I am sure you get the idea. What happens is as the image moves across the angles sheet it gives you the feeling that there is motion past you, thus increasing the feeling of motion.

NOTE: #1 The sheet must extend past your normal view out the cockpit windows, otherwise the setup is useless.

***************** WARNING********** "Ask your wife before you staple the crumbs out of one of her sheets" ****************************


Best,

Trev

Matt Olieman
12-17-2005, 11:32 AM
Collimated mirror display. You hit the mother load of all challenges of cockpit building.

Is it possible and do we have information on how it can be done? The answer is yes. HERE COMES THE BUT :)
We don't have the technology and the curvature of the mirror information.

The technology to be able to be able to build a collimating mirror is absolutely crucial, and the precise curvature is also crucial, otherwise the mirror would easily be flawed.

I've spent 100's of hours in research and testing to duplicate this process. I've seen this brought up in several forums and with such enthusiasm, only for the subject slowly dissipate.

Don't even think of buying one of these...... $$$$$$$$$$$$$$$$

Collimating mirror is the way to go, ultimately. If someone finds a way... I want to know :)

s4sha
12-21-2005, 02:05 AM
Trev,

Thanks for sharing a great idea to enhance motion cues!

I've been thinking (that's a scary thought) -- Here's another interesting idea that you could try:

Since the projection surface is relatively close to you, you probably lack the vertical and horizontal parallax cues that you would experience if the observed view was at an infinite distance (your view out-the-window does not change relative to your head position, left-right or up-down). For example, when piloting an aircraft or even driving a car and you see a cell tower in the distance, if you move your head from left to right while fixating on the tower you will notice that the tower seems to move right and left in relation to a point on your glareshield/dashboard. Now if you do that with your current setup i would imagine the relative difference between the two points(the tower and panel) is limited.

You may be familiar with NaturalPoint's TrackIR device for 6 DOF (degrees of freedom) viewability. I was thinking that one could have a large single-projector view out front, then utilize the TrackIR device to simulate the horizontal and vertical parallax.

The TrackIR does allow you to also "roll and yaw" and zoom (z-axis) to "look around", which is great-- but if it could be used to help that cell tower in the above example to appear to move left in relation to the panel when a left head movement occurs, one could experience the same sense of "infinite depth", albeit artificially.

I hope you understand my concept, and if anyone thinks I have too much time to daydream about flight simulation displays, please tell me to shut up and have a beer...

S4sha

Trevor Hale
12-23-2005, 09:16 AM
Actually,

I have seen some people use that. My only problem with the track IR is I am sure it is used in Virtual cockpit. That being the case you need to create an mdl file that has no VC. Secondly, if you talking with a copilot in your sim and you turn your heas to the right to talk to him the entire view will rotate right. :)

Not sure if that works, but I would think it would be a bit confusing.

Any thoughts?

Trev

Matt Olieman
12-23-2005, 10:06 AM
s4sha,
Interesting concept. I can see that working.... Then I can understand the confusing bit Trev mentioned if you were the F/O or a passenger.

The challenge continues:)

s4sha
12-24-2005, 12:51 AM
Yes, I did forget to mention that this experiment would only apply to a single-pilot configuration. Matt, I'm glad you understand my idea. Trev, you are absolutely right, the VC would have to be omitted, and I was thinking of utilizing only the vertical and horizontal head movements that are parallel to the screen, and not the head turning capability. I would imagine that this would induce an artificial parallax perception that may be worth the small investment to enhance the experience. I only have the TrackIR 1st generation, but hope to research the TrackIR 4 to see if you can indeed isolate certain axies. Anyway - again, just a thought...

PhilM
01-06-2006, 03:16 PM
Evening guys...I'm new here so Hello :)

A little about me, working as part of a team building a BAe 146 STA simulator -- its a bit generic since the 146 STA was only a concept, so the cockpit is a modern EFIS/ECAM system, with 737/A320 style panels and systems that are true to the 146. I work as an aircraft engineer on B737 and A320s also.

Anyyyway...collimated visuals! On a par with Motion really, hard work, and awesome!

Ignoring for now the rear projection collimated visuals, as I don't know enough about those (bar you need a HUGE spherical mirror!).

If we look at the mirror based collimated displays such as this:

http://img.photobucket.com/albums/0703/philm/Sim/colmirror.jpg
Image courtesy of 737sim.com

We have a few elements here as you can see.....a CRT montitor ona frame, a partially reflective mirror (beam splitter) and a spherical mirror.

Now, the CRT part is easy and cheap! People can't give them away these days!.

The beam splitter looks a rather simple affair, it's a sheet of glass with a very fine layer of metal on it, that reflects the light back onto the mirror. I havent been able to find these yet, but I don't expect they are too "bad" price-wise, of course new, from an optics supplier they will be! I am sure they can be found elsewhere!

Spherical Mirror, ok, this is the hard part. The mirror surface itself is easy, materials such as mylar and others (as linked at the top of the page) are availible and make life very easy, the difficult bit is how to get the spherical/parabolic mirror shape. Well, I had a thought....

http://www.satellitesuperstore.com/images/td54.gif
Satellite dishes are actually parabolic mirrors! And they are like CRTs, availbile everywhere, ten-a-penny :) Of course, to our eyes they are not very mirror like, but they do a fine job of reflecting electromagnetic waves! Of course, for them to reflect visible lght, we will have to take some mirror film......seal it round the edges of the dish, drill a small hole in the centre of the dish, and apply a small vacuum - tada...parabolic mirror :)

Alternativly, we could take a metal dish to be chrome plated or something similar, there are even Chrome paints availble....Audi did a Chrome A6 I believe for a Middle Eastern customer.

Providing we can then find suitbale beam splitters, which to me seems the easy part, we have collimated visuals el cheapo. Repeat 3 times and have a decent field of view too :)

Pochflyboy
01-06-2006, 04:03 PM
Sounds good, but much harder than one would think. In my first days of researching sims, I decided I was going to do something similar to what you are thinking. The only problems are a) the mirrors have to be EXACTLY the same (not really possible to do at home. b) the beam splitter mirrors are VERY $$$$$. Anywere I found them, I would have to contact the company and could not just buy them. Usually they were in the multi-thousands of dollars. I never found one low enough in price that it would make the difference. So thats what I have found. Hope it helps.

Trevor Hale
01-06-2006, 05:52 PM
I was at an Electronics shop, and a guy showed me a can of reflective spray. $18 CDN per can and you spray it directly on a sheet of glass, it makes the glass reflective. Makes an excellent "Transparent Mirror"

Maybe that would be a route to go.

Trev

Pochflyboy
01-06-2006, 06:44 PM
The only problem, is you need a very special kind of film, that will Reflect light coming form 45 degrees up, and lets light through that is coming from 45 degrees down. This does not come cheap!

AndyT
11-20-2006, 11:32 PM
If you can wait a few months, this company has created a holographic projector the size of a matchbox that will solve the problem. They are still working the bugs out but when they finish it... Look out!

http://www.lightblueoptics.com/features.htm

Matt Olieman
11-21-2006, 02:50 PM
I've heard of this technology a few years ago. First time I've seen this.. It's my undestanding this is definately the way in the near future of projection.

But... it will not replace collimated mirror display... still a challange for us hobiest:(

Jackpilot
01-20-2007, 06:48 PM
Hi,
My first post on this wonderful site , even if I *"know" most names here, of FDS fame...

What about those stain glass mirrors used everywhere to see and not be seen...would they do the job or are they too dark?..

Matt Olieman
01-20-2007, 07:05 PM
Jack, I think you're right, they may be to dark.

BTW, welcome :)

Bob Reed
01-21-2007, 12:27 AM
Has anyone looked at PM's collimated boxes?

Trevor Hale
01-22-2007, 09:15 AM
Bob,

I did. The non Pro Version of Collimated Display is Much more expensive then you would think,. and worst off.... They require a minimum purchase of 2. So... If you have just shy of 4,000 Euro's your good to go...

:) Needless to say too rich for my blood. Maybe a retiree like Matt can afford this setup......:mrgreen:

LOL

Matt Olieman
01-22-2007, 09:38 AM
Maybe a retiree like Matt can afford this setup......:mrgreen:

LMAO :mrgreen:

If I had the money, I would go for the 160 degree collimating Mirror display. I heard, around $150K USD.

I seriously considered the Collimating Display form PM. They are made by a Professional company, with two grade options.

What I did not like about that type of display is, CPT can't see out of F/O display and CPT can't see out of CPT side.

The positive side of this, is the Collimating Mirror effect of the visual being centered to where you move your head.

BTW, I know at United Flight Training Center (the largest in the world) still have sims with this type of display in use.

Even with a projected 160 degree screen, you give up some of the features the PM Collimating Display has to offer... Decisions, Decisions and Decisions :roll:

Trevor Hale
01-22-2007, 10:19 AM
Matt,

I have been playing with your idea of the curved mirror. What I am having troubles with is by the time the image gets from my projector to the mirror the image is to big, and when I make the image small enough to fit on the mirror it still isn't big enough to fill the screen.

How are you dealing with this issue?

Mike.Powell
02-11-2007, 04:01 PM
I've been doing some research on this topic as well. Here are a few sources you may not have seen.

"Collimated Displays for Flight Simulation" by A. Michael Spooner was published in the May-June 1976 issue of Optical Engineering. It mostly describes Redifon Flight Simulation's decision process for developing their "Duoview" wide field of view display system. It nonetheless give basic descriptions of the small spherical mirror and beamsplitter display and the larger off-axis panoramic collimated display.

Patents have proven to be excellent sources of information. The patent search capability of Google makes fast browsing easy. When collimated displays were first being developed they were referred to as "infinity displays", so you'll want to include that in your list of search terms.

A particularly interesting patent is #3,549,803 - "Virtual Image System for Training Simulator" as it includes a description of making a low cost spherical mirror using reflective Mylar film, urethane foam and fiberglass.

Researchers at the University of Strathclyde in Scotland have worked with very large concave mylar mirrors as part of 3D CAD systems. The construction details of those mirrors are disclosed in US patents #4,822,155 and #5,109,300.

Matt Olieman
02-11-2007, 06:07 PM
Hi Mike, Thanks for that wonderful information. As always your the man with all the knowledge, and always appreciated :)

I've got three rolls of 6' X 10' mylar, each having a different thickness. I've experimented and experimented and ...... well, you know how it goes :rolleyes:

The collimated displays / infinity displays thing..... although that's the best, without the specialized tools and knowledge... it's almost impossible. But, I guess I/we can keep our hopes up, and someone will find a method for our madness ;)

Keep feeding us Mike :)

spitfire9
07-19-2007, 02:55 AM
Has anyone tried or would anyone have any info on the collimated displays they have for sale at Project Magenta??

I sent an email a couple of days ago asking about the cost but I haven't recieved a reply yet.
I have a feeling they will be quite expensive but would still be very interested to know if the visual display is that much better than some lcds in front of the windows????

Twizzstyle
08-22-2007, 02:58 PM
Bringing back a bit of an old thread...

I've been really wanting to do this for a long time. I work in actual simulators from time to time, and the depth is just amazing, really adds a real feel to it (even in the non-motion sims). Currently my screen is about 6 feet from my eyes, and although its plenty big, it looks like its 6 feet in front of me... not off at infinity.

So my idea is to build a frame out of plywood, and then sheet it for the right contour. Then lay some kind of mylar material (if I can get it to stretch enough and form to the shape). I wrote a MatLab code to calculate the mirror shape...

http://photos-d.ak.facebook.com/photos-ak-sf2p/v74/225/12/10708561/n10708561_34062135_9390.jpg

http://photos-561.ll.facebook.com/photos-ll-sctm/v72/225/12/10708561/n10708561_34073966_8425.jpg

Matt Olieman
08-22-2007, 09:12 PM
I'm all for it!!!! If you can figure out how to make this, you'll be the flight-sim builders hero forever :)

James Twomey
08-23-2007, 01:42 AM
That would be something if it could be figured out. Twizzstyle if I may ask, where in Washington are you located? Im located in Port Orchard.

Twizzstyle
08-23-2007, 02:18 PM
I live in Kirkland, and work in Flight Test at Boeing (at Boeing Field) :)

I've confused myself a little on this though... the matlab code I wrote to calculate that shape just looks at geometry. When you look at the bottom of the mirror, it should reflect to the bottom of the screen (that is above my head), if you look at the top it should reflect to the top of the screen, etc. Just simple incident/reflected angles and geometry, and it creates that curve... But where I'm confused is on simple optics...

So aparently, in order for the image to appear behind the mirror (what we want here obviously) the object (the screen in this case) needs to be within the focal point of the mirror... mine is obviously way outside the focal point (the focal point is approx. half of the center of curvature)...

http://www.glenbrook.k12.il.us/gbssci/phys/Class/refln/u13l3d.html

So by simple geometry, my image should fill the whole mirror, but is it going to appear like it is right in front of my face? :(

Matt Olieman
08-23-2007, 02:38 PM
The image or screen above the cockpit should also be curved to match the mirror (although smaller).

If I recall correctly, the bottom of the mirror actually looks at the top of screen rather then the bottom. So the image is projected upside down on the screen.

Please correct me if I'm wrong :roll:

Twizzstyle
08-26-2007, 10:52 PM
The image or screen above the cockpit should also be curved to match the mirror (although smaller).

If I recall correctly, the bottom of the mirror actually looks at the top of screen rather then the bottom. So the image is projected upside down on the screen.

Please correct me if I'm wrong :roll:

Aha! Maybe that is the piece I'm missing that would fix my focal point location conundrum...

Should be very easy to just flip the direction that my code runs to do this... I'll see what it comes up with.

Mike.Powell
08-26-2007, 11:22 PM
For a mirror collimated display, the image source is located on the focal surface of the mirror. The mirror is typically a spherical section. The focal point of a spherical mirror is located at half the radius. We're used to thinking of "focal planes" in optics, but this is an approximation that breaks down with the large optical angles needed in a simulator display system. Because of the large fields of view, spherical distortion becomes significant. To correct for this spherical aberation, the image surface should be a spherical section with a radius half that of the mirror. The mirror and the image surface will share the same center of curvature.

Twizzstyle
08-27-2007, 12:22 AM
So here is what my code produced with the image screen flipped...

http://photos-561.ll.facebook.com/photos-ll-sctm/v116/225/12/10708561/n10708561_35284415_6753.jpg

Mike - you say the image should be at the focal surface... now when you say focal surface, is this because there is no focal "point"? (because or the mirror curvature... maybe I should have my code find the focal point...) Then you go on to say the mirror is spherical, but this is not true is it? It is more parabolic?

Matt Olieman
08-27-2007, 08:59 AM
Mike, thanks for that wonderful explanation, as always a true engineer using proper terminology :)

Twizzstyle (using that for lack of know your name), you show the projected image (above the cockpit being flat) as Mike mentioned it should be spherical.

Wouldn't it be wonderful for us to be able to build one of these and have the technology to do this at a reasonable cost :)

Twizzstyle
08-27-2007, 10:49 AM
Mike, thanks for that wonderful explanation, as always a true engineer using proper terminology :)

Twizzstyle (using that for lack of know your name), you show the projected image (above the cockpit being flat) as Mike mentioned it should be spherical.

Wouldn't it be wonderful for us to be able to build one of these and have the technology to do this at a reasonable cost :)

My name is Sean ;) But either name works, haha

And yes, I do plan on having the image screen spherical. It is much easier in my code to have it be flat for the time being until I figure out exactly how I need to do this. Then I will probably write a new code that is a little more precise with the curved image screen. And again, my screen (well... the mirror) is still just going to be a forward view, as my flight deck is built into a closet. But this same principle could be used for a full width mirror as well.

Trevor Hale
08-27-2007, 11:08 AM
My name is Sean ;) But either name works, haha

And yes, I do plan on having the image screen spherical. It is much easier in my code to have it be flat for the time being until I figure out exactly how I need to do this. Then I will probably write a new code that is a little more precise with the curved image screen. And again, my screen (well... the mirror) is still just going to be a forward view, as my flight deck is built into a closet. But this same principle could be used for a full width mirror as well.

Please excuse my ignorance here Sean, but I am trying to understand how you can take the information from the computer and make the screen itself. What i am getting at, and hope you can help me understand is... if the computer program can build the shape of the screen, would you in essence get that program to give you the measurements of the screen itself? The computer program is flawless but won't account for the human error factor, or building material flaws, Do you think these problems can be overcome?

Thanks,

Trev

Twizzstyle
08-27-2007, 12:00 PM
That's why I have reservations about this. Something like this needs to be VERY precise, and thats why the real thing is so rediculously costly. Still a fun engineering excercise to try to do it frugally anyways :) And thats what the whole flight deck is all about!

So yes, you had it right. From the code I'll be able to get profile curves that I can use to make a frame of sorts. Imagine a grid of plywood or MDF sheets that each have the profile contour. Then I will sheet that with some kind of flexible material (I was thinking some thin ABS or plexiglass, something I can heat up and form into the frame) and then the mirror film over that (assuming the mirror film can be shaped enough).


Please excuse my ignorance here Sean, but I am trying to understand how you can take the information from the computer and make the screen itself. What i am getting at, and hope you can help me understand is... if the computer program can build the shape of the screen, would you in essence get that program to give you the measurements of the screen itself? The computer program is flawless but won't account for the human error factor, or building material flaws, Do you think these problems can be overcome?

Thanks,

Trev

Trevor Hale
08-27-2007, 12:24 PM
Ahhh, so it is indeed a complex project then. Ah well, Either way you will have one heck of a nice screen in the end to paint White and project your image directly on it. LOL

Please keep us posted.

Twizzstyle
08-27-2007, 12:41 PM
Ahhh, so it is indeed a complex project then. Ah well, Either way you will have one heck of a nice screen in the end to paint White and project your image directly on it. LOL

Please keep us posted.

haha! Exactly!

Mike.Powell
08-27-2007, 12:59 PM
Mike - you say the image should be at the focal surface... now when you say focal surface, is this because there is no focal "point"? (because or the mirror curvature... maybe I should have my code find the focal point...) Then you go on to say the mirror is spherical, but this is not true is it? It is more parabolic?

The collimating mirrors used in sim display systems are typically spherical because that shape is easiest to make. To reduce the overall weight, the mirrors are often made with aluminized mylar fastened to a carefully shaped frame such that there is a sealed volume behind the mylar. A small vacuum in the volume causes the mylar to stretch into a concave spherical-section shape. I suppose technically it's a 2-D catenary, but by choosing the proper shape for the frame, the mylar closely approximates a spherical section. There have been systems utilizing "hard mirrors" made of glass or of metal which could have been formed into a parabolic shape, but I've never run across one.

Spherical optical elements (both mirrors and lenses) do not have true focal points. However, for optical paths that are both close to and nearly parallel to the center path through the optics, it's a good approximation to say they do. We can extend this approximation to a flat surface intersecting the optical axis at right angles at the focal point. This is the so called focal plane. As long as the optical paths are close to the optical axis we can form images of reasonable quality on (or from) the focal plane.

Large optical systems like panoramic collimated displays wildly bust this approximation. In these systems it doesn't make sense to refer to a focal plane or focal point because you can't form a reasonable image using a large flat source image. You can, however, form a very nice image if your focal surface is curved rather than planar. For a spherical collimating mirror of radius R, the focal surface is a spherical section of radius R/2 that shares the mirror center of curvature.


Companies like Glass Mountain Optics and SEOS, makers of sim display systems, used to have web sites with more technical information. Now they're simply marketing contact points. Currently, the best source of information seems to be the Patent office. You can search patents for free using the Google patent search. Use terms like "infinity display", "collimating mirror" and "film mirror". The down side of reading patents is that the state of the art has improved a lot. Older patents show designs that are overly complex. Current panoramic display systems have one big mirror and one big screen. Wading through a bunch of old patents may be confusing at first, but it will introduce you to the technology.

While the display system company web sites don't provide much detailed information, some at least do provide product pictures that give a feel for the general optical system layout.

Matt Olieman
08-27-2007, 01:04 PM
Trevor, also keep in mind, the slightest dimple on the mirror can make the image way out of wack.

I believe the spherical mirror backing is formed on a die. The form has thousands of tiny (micro) holes. Glue is applied to the form and the mylar film is vacuumed to the form.

Not much room for error here :(

Mike.Powell
08-27-2007, 01:25 PM
Another reason for using spherical mirrors is that the crew will experience the same mirror curvature regardless of their position. If it were not spherical, different cabin positions would perceive differing image distortions. The symmetry of a spherical mirror leads to better image quality throughout the cabin.

Matt Olieman
08-27-2007, 01:30 PM
I can only drool and fantasize :(

Twizzstyle
08-27-2007, 01:33 PM
Another reason for using spherical mirrors is that the crew will experience the same mirror curvature regardless of their position. If it were not spherical, different cabin positions would perceive differing image distortions. The symmetry of a spherical mirror leads to better image quality throughout the cabin.

My sim is a single seat though, so this isn't as big of an issue.

I've been reading through some of the patents a bit, but having trouble getting any of the images to show up... And yes, it does describe using a spherical mirror with the image source located at the focal plane (half the radius of the mirror) just as Mike described.... That would certainly be easier, wouldn't even have to deal with this code I wrote (even though I'm really pleased with the code... :( )