Mirrors and Curved Screens

[Ferrino]

Hi - I'm looking to use a single 1080p DLP projector with a rectangular curved screen (not settled on desired curvature yet, but hoping for something along the lines of 135 degrees: only curved left-to-right). Have the space for rear or front projection, although I am looking to test with front projection only initially (due to cheaper screen material).

I am trying to establish if a curved mirror (matching the curvature of the screen) offers any benefits other than doubling the effective throw of the projector for small spaces?

Does it also improve the focus at the sides (compared to projecting directly onto the curved surface) and can it also remove the need for image-warping software when set up correctly to match the screen?

Thankyou!

[riche543]

Hi ferrino

I have not yet got to my visual setup, But im in a very small space with no height for a projector, so i hunted around & found this regarding corved mirror http://www.flightdecksolutions.com/forum/viewtopic.php?t=1251&postdays=0&postorder=asc&start=0 I have only run breif test,s with mine . But have the projector low down in front of the sim pointing back to a curved mirror fixed roughly behind MIP . I fixed each corner of the mirror so they can be ajusted to approx the same curve as screen which is made from sintra. I found though if you curve the mirror to much the edge,s of picture stay focused but the centre goes blurry as you are streching the mirror so to speak. Im guessing you would get better results wiyth short throw projector as you wouldnt have to bend the mirror as much. then use sol7 or similar software to correct image
Good luck with it, like i said im really no expert on this as im not at that stage with mine & havnt experimented enough. Keep us posted , im interested in the results you get.
Cheers
Richard..

[Neil Hewitt]

I doubt it would be practical to use the sorts of mirrors within the range of the average cockpit builder to optically correct for all the distortion introduced by a curved screen. It's certainly physically possible, but you'd have to make a mirror precisely shaped and ground to correct for the focus and image distortion. This is what they do in commercial simulators, which use a collimated mirror arrangement that both precisely corrects for the distortion of the image and also re-focuses it at infinity to avoid parallax issues (where two viewers see an object at different positions due to their angle relative to the screen, which is a problem for two-pilot simulator operation). Such mirrors are custom-made and approximate a section of a sphere, which you couldn't do with ordinary acrylic mirror sheet.

What you certainly can do is use a curved piece of acrylic mirror to reduce the distortion when reflecting onto a curved screen and then do the final corrections using NTHUSIM. Optical manipulation of the image is clearly better than digital manipulation.

I'll be using a curved mirror arrangement in my own project and am planning to use NTHUSIM as well. Make sure you buy 'first surface' acrylic mirror - this has the mirrored surface on the front, rather than on the back as with a traditional glass mirror. Otherwise you get imperfections and distortions on the image as it has to pass through the acrylic on its way back to you.

Alternatively you can use reflective mylar sheeting and glue this down to a backing material of your choice. You're likely to get lots of little imperfections this way, of course, so I figure acrylic mirror sheet is better. It's pretty cheap too.

[Matt Olieman]

Neilh is 100% on the point. This been a topic ever since the beginning of Cockpit building. This is the one item of cockpit building that has not been conquered by amateur builders. Believe me, I've tried. :)

Matt Olieman

[Ferrino]

Thanks for the replies. OK, so I guess what I was getting at was:

Given a rectangular projection screen which is moderately curved, is it there a significant improvement in picture quality (focus on the side views, distortion) when reflecting off a similarly curved first-surface mirror (+ finishing touches with NTHUSIM) compared to projecting directly onto the surface? Assuming a single viewer and a single projector.

The main thing that is putting me off directly projecting onto a curved surface with one projector is the blurry/out-of-focus side views, which I guess vary with severity depending on how aggressive the curve is.

Thankyou!

[Neil Hewitt]

A well-aligned and suitably curved mirror will certainly help a little. How much it helps is implementation-specific. The shape of your curve should never exceed that of a simple circle, certainly. Some people have tried to stretch the picture from a single projector right round the sides, but invariably it doesn't work well.

Acrylic mirror is cheap and easy to obtain. I'd say buy some, and give it a try. You can do it on a small scale to test it, and then scale up to the real thing when you're ready.

[Mike.Powell]

I would not expect the curved mirror to help with focus.

Focus depends on the optical path length and the depth of field of the projection lens. Most projectors are designed to focus on a flat screen. When the screen sides curve toward the projector, those portions which are moved closer will lose focus. The smaller the lens depth of field, the more defocusing you get.

Lens depth-of-field depends on the exit aperture of lens and the optical path length between the lens and screen. Think of a cone of light with the cone base at the lens exit aperture and the cone point at the screen. If the screen is exactly at the point, the image is in focus. As the screen moves closer to the lens, the screen intercepts a larger cross section of the cone, the light spot size grows, and the overall image appears less focused.

Depth of field or "being in focus" is simply the range of very nearly focused that is acceptable to the viewer. There actually is a loose camera industry definition of depth of field based on spot size, but it comes down to looking at the image and saying "That's okay",... or Not!

Most consumer LCD and DLP projectors have small exit apertures and provide much greater depth of field than did the older CRT projector with their massive lenses. You can boost the depth of field by choosing a long throw distance. That will give you a longer, skinnier light cone.

A nice thing about mirrors is that the optical performace is simple. The angle of incidence equals the angle of reflection. That means you can easily and accurately map out the light paths for different mirror/screen geometries on paper. You can get an idea of what will happen without building a complex test setup.

[Neil Hewitt]

Hi Mike! Didn't realise you hung around these parts (of course, had I bothered to look around a bit I'd have spotted Mike's Tips). I've been using your site among several others for inspiration over the past few months. Your stuff on collimated displays in particular is fantastic.

[Mike.Powell]

Hey, what can I say? MyCockpit is just a great site to hang around!


I find collimated displays very interesting. While the FAA/CAA/etc. require them for level C/D sims because they present the same perspective view throughout the sim cabin, I think the hobby community would benefit because collimated displays can provide an expansive view within a restricted space. The optical design is straightforward, but as I'm sure many know, the physical implementation is a real handful. I've been researching the topic for a possible new book.

[mikesblack]

Hi neilh,

I'd be intersted to know more about the curved mirrored you are using and more generally about your approach. I am considering all options related to creating true stereoscopic vision. I plan on building a semi spherical surface that will surround my fov to 200 degrees.

My thoughts are using Wideview, rather than T2go to accomidate a true circulair FOV, Nthusim for image distortion and 3 or 5 projectors. I've been considering ways other than collimated mirrors as a of the considerations mentioned here(cost and dificulty of precision fabrication), but have come up short.

If you can add more detail, would be greatly appreciated.

Mike

[castle]

Hey, what can I say? MyCockpit is just a great site to hang around!


I find collimated displays very interesting. While the FAA/CAA/etc. require them for level C/D sims because they present the same perspective view throughout the sim cabin, I think the hobby community would benefit because collimated displays can provide an expansive view within a restricted space. The optical design is straightforward, but as I'm sure many know, the physical implementation is a real handful. I've been researching the topic for a possible new book.

Not an optics guy, so could be all wrong on this....

I was under the impression that the visual systems in high end sims use curved mirrors to reflect a virtual image back to the viewer. That is not the same as viewing an image on a curved screen; i.e. bouncing an image off a curved mirror to a curved screen is not the same as viewing an image of a curved screen on a curved mirror. Or is it?

And that is what makes the problem so difficult for the hobbyist not to mention the cost and precision required.

As you've noted, the rays of light from the viewing surface hitting the viewer's eye must be parallel or collimated which is easy to understand from a math and geometric point of view ( ooo, -2 for bad pun) but hard to build.

JW

[Neil Hewitt]

Hi neilh,

I'd be intersted to know more about the curved mirrored you are using and more generally about your approach. I am considering all options related to creating true stereoscopic vision. I plan on building a semi spherical surface that will surround my fov to 200 degrees.

My thoughts are using Wideview, rather than T2go to accomidate a true circulair FOV, Nthusim for image distortion and 3 or 5 projectors. I've been considering ways other than collimated mirrors as a of the considerations mentioned here(cost and dificulty of precision fabrication), but have come up short.

If you can add more detail, would be greatly appreciated.

Mike

Hi Mike.

I'm a long way from the visuals part of my project. Still on the shell build, probably not even halfway done there yet. It's slow going this year.

That said, my tentative plan, because of a lack of space, is to rear-project onto a curved screen positioned relatively close to my windows via an intermediate mirror to give me the throw length. I was planning on creating a simple curved mirror surface to match the curvature of the projection screen. NTHUSIM to adjust it all nicely. I briefly entertained crazy ideas about CAVE-style visuals but I don't see that working. I know that with this setup I'm not going to get anywhere as good an impression of depth as I might if I could put the screen a few meters away, which is why I was interested in collimation as a topic.

Having followed this thread and others, I am wondering what could be done with mirrors, even non-spherical, non-collimated mirrors. It seems like a way to get a large display in a small space. But I'm a long way from truly understanding what would need to be done. More study time called for, I think.

[Neil Hewitt]

Not an optics guy, so could be all wrong on this....

I was under the impression that the visual systems in high end sims use curved mirrors to reflect a virtual image back to the viewer. That is not the same as viewing an image on a curved screen; i.e. bouncing an image off a curved mirror to a curved screen is not the same as viewing an image of a curved screen on a curved mirror. Or is it?

And that is what makes the problem so difficult for the hobbyist not to mention the cost and precision required.

As you've noted, the rays of light from the viewing surface hitting the viewer's eye must be parallel or collimated which is easy to understand from a math and geometric point of view ( ooo, -2 for bad pun) but hard to build.

JW

Yes, absolutely. Mike Powell's site has a great section on exactly this. Commercial sims put the mirror in front of the pilot's seating position as a display. The use of curved mirrors to reflect a projected image onto a projection screen is something different that some builders have used to get over not having a long enough throw length or enough room in their sim room. I'm in precisely that boat. The use of the curved screen itself is (as I understand it) to lessen the horizon-distorting effects that you can get with the traditional 'three screens at 45 degrees' approach. You can't (as far as I know) do a collimated display that way. I suppose you could put a collimating fresenel lens in-between the projector screen and your viewpoint, but making it work would be, erm, hard.

I should shut up now. Plenty of people here know more than me ;)

[mikesblack]

I wonder what the most dificult part someone without "profesional equipment" would encounter in the attempt at building a diy parabolic mirror. Matt I believed mentioned he had been working with this and ahd made an attempt. I'm currious what difficulties he had encountered and what things he thought made this project improbable.

Also, I am quite new to optics as well, and so I may be somewhat nieve with my ideas I'm expressing, but I wonder about Wideview. Given enough seperate windowd views, could we theoretically converege on a collimated type view.

Say for example we had 10 networked PCs or 20 or some crazy number, let's say for example 180 projectors each beaming at 1 degree change along the circumferance screen would we in effect be doing the same thing a curved mirror would.

[Mike.Powell]

A collimated display creates an image which appears to be at a great distance from the viewpoint. I am unaware of any way to create a collimated image directly from a nearby projection screen. A distant projection screen will approach collimation. Lacking that, light bending optics are required. Given the required size of the optics, mirrors are the approach of choice.

If you look at you own reflection in a flat mirror you see an image of yourself which is as far behind the mirror as you are in front of it. If you could slowly curve the mirror into a spherical, concave shape, your image would appear both to get larger and to move away from you. At some point the image appears "infinitely" far away and "infinitely" large. The light ray paths from each image point are parallel. This is collimation.

In this example, the image also becomes quite distorted. A spherical-section mirror can collimate, but also distorts because rather than having a flat focal plane, it has a spherically curved focal surface.

In simulator collimated displays, the real image is created on a spherically curved surface. Generally this is an acrylic rear-projection screen. If the screen has a radius of curvature that is one half the curvature of the mirror, AND if the mirror and screen share the same center, the virtual image (what you actual see) will be both collimated and undistorted.

The reason that the mirror is a spherical section is that spherical mirrors are easier to make, and don't have a required or preferred viewing axis.

Most, if not all, collimation mirrors used in contemporary, cross-cabin collimated display are Mylar film mirrors. Reflective Mylar is fastened to a carefully shaped framework with a closed back. The air pressure behind the Mylar is reduced and the Mylar is forced back into a curved shape. If the Mylar is stretched just right, the curved shape is spherical.

Managing the distribution of tension across the Mylar surface is key to controlling the mirror shape. It gets more difficult as the vertical arc of the mirror gets larger. Mirror size limited the vertical field of view to about 45 degrees for some time. Eventually SEOS figured out a way to pre-tension the Mylar, and got the vertical field of view up to the 57~60 degrees needed for helicopter and aerial refueling sims.

[mikesblack]

Hi Mike,

Thanks for that. Very interesting indeed.

Mike

[lineman55]

Here is what I did when I had my 737 pit running.I no longer have this setup but have switched to the Dodo Bell 206.It worked well for me although not perfect.And still affordable.
http://biagettis.com/737simproject/737simproject_008.htm

[mikesblack]

linemann55,

I had seen your set up before. Very interesting.

Can you answer a couple of questions? I was wondering how you managed to allow for the FO position to crab free as you mentioned on your website. Could you imagine a way to do this with a curved screen of say 220 degrees, and multiple projectors?

I was wondering about sintra material. I am currently experimenting with EVA or running shoe midsole material. It comes in sheets of various color and density. What I like about it is that it is smooth, light and extreemly flexible. The issues I see are finding a method to blend the seems with glue cleanly. Also, considering it is a porous matt serface, I wonder if I will get good relectivity. I did test with what seems good results, but I don't have much to compare with.

[lineman55]

Well. what I was able to discover is by curving the screen just enough without distorting the image to much it gave the Captains view a more realistic view.As for the FO ..it was better but not as good as the Captains side.I was never lucky enough to have anyone in the area to share my hobby so I wasnt to concerned with what the FO saw or how.
I believe by curving the screen and setting the zoom just right and shifting the projector image over to where my view was more realistic it satisfied my want for a noncrabbing view.
Sintra or cintra ,not sure how its spelled..worked fantastic for this pupose.Sturdy and flexable and strong enough to drill,sand and still light enough to easiliy move about.Plastic suppliers for sign companies usually will stock this.You can even use it for creating a shell surface for your project.
Im not familiar with the material you are interested in using for your screen.It sounds workable by your description.
The curve I implemented in my project was the maximum for use with 1 projector.
Any more then that and the images on the sides tend to run or blur somewhat.
James:D

[Sean Nixon]

As far as I can tell, the FO crabbing effect only applies when sitting close to, say a monitor. The effect just isn't there with a large screen placed way out in front of you.

[Matt Olieman]

It doesn't matter what distance the monitor/screen is to overcome the crab effect. The only way to overcome crab effect is to center the monitor/scree from the POV.

Example: If the pilot is on the centerline of the runway, runway line infinity would be in the center of the screen. No matter what distance or how large the screen may be and the center line starts in the middle of the cockpit, the runway line infinity would be aligned to the center of the cockpit. If traveling to the runway line infinity, you would have to crab the airplane.

Collimating mirrors overcome that problem for what ever position you sit in the cockpit, the centerline of the runway always is the center of your POV.

I hope I made sense.... :) If not, someone please help make some sense of what I'm trying to say. Mike Powell you know what I'm talking about :) :) :)

Matt Olieman

[Sean Nixon]

What you say makes sense, and I remember being a little concerned about the dreaded crab effect before I started building. But I've yet to see a setup where the effect is noticeable to the extent it's off-putting. When sitting in front of a large projector screen about 6ft away from the eyepoint, I would say the view is almost identical (without close scrutiny) from either position, left hand seat or right hand seat.

[Matt Olieman]

What you say makes sense, and I remember being a little concerned about the dreaded crab effect before I started building. But I've yet to see a setup where the effect is noticeable to the extent it's off-putting. When sitting in front of a large projector screen about 6ft away from the eyepoint, I would say the view is almost identical (without close scrutiny) from either position, left hand seat or right hand seat.

Exactly Sean. I ended up moving my projector so that the center of the screen was aligned to what ever seat I was flying. Unfortunately the opposite seat would view an approach as crabbing.

Matt Olieman

[Sean Nixon]

But why would you offset the projector when the view looks acceptable from either seat when it's centered?

[Matt Olieman]

But why would you offset the projector when the view looks acceptable from either seat when it's centered?

Sorry Sean, I misread your reply.... OOPS... :) I thought you meant you hardly noticed the difference from closer screens......

My screen was curved, 7' wide and 10' from the pilot seat. I distinctly noticed the crab.

I got the idea of moving projectors from someone else. He made a moving (turning) projector platform for the same reason as I moved my projector. I can't recall who it was, but he did leave images here of his project (somewhere).

I guess it's all about how tolerable someone is regarding to the crabbing issue..... :)

Matt Olieman

[Neil Hewitt]

This is what is worrying me about my build. Because I only have about half a metre between the front windows and the wall, I can't do a front-projection and get any form of wrap-around or side-view (I only have centimeters spare between the side window and one wall of the sim room). I could mount the projector right on top of the shell at the very rear and front-project an image perhaps 2.5m wide, but it wouldn't be hugely immersive. I don't have room for a 3-screen, 3-projector set up.

My plan is to rear-project onto a screen very close to the windows, using mirrors to fold the beam and gain the necessary throw length to get a 2.5m-wide image from in front, with a curved screen that will provide a degree of wraparound and immersion, and then perhaps TFT panels on the side windows. But with the screen so close to the viewing position, the dreaded parallax problem will rear it's head. Granted, I'll be flying on my own most of the time so posititioning the center line somewhere to the left will probably help, but I'm a perfectionist, dammit :)

I've looked at a mirror-based solution - project onto a screen and reflect the screen in a mirror - since in theory this can be used to make the screen seem further away, but of course the image in the mirror will only be as large as the screen itself unless you can curve the mirror to create a magnifying effect, and you have to be able to curve it in both X and Y dimensions to achieve such an effect correctly. Achieving such a curve with simple acrylic mirror is beyond the capabilities of most amateurs. I guess you could heat-shape acrylic mirror onto an appropriately-shaped mould, but a) you've have to be very precise indeed, and b) I don't know what the application of heat would do to the mirror coating - it would proably stretch and break, I imagine, since it's not actually part of the acrylic itself. A mylar film mirror tensioned with a partial vacuum is what you'd need, but then that would be equivalent to a proper industrial collimating mirror and if you have the ability to produce such a thing, you're already there.

Nowhere that I could find online will sell you a pre-made spherical-section concave mirror large enough that you could use it to cover the display for a single front window in a normally-sized cockpit (say, at least 800mm across) and give you a home-made WAC (wide-angle collimated) unit. If you could get an optics or plastics outfit to make it for you, you'd have a shot, though ordering 3 or 4 isn't likely to be a large enough job for most places to be bothered.

Oh well. I'm going to make do with what I have and see how it turns out.

[mikesblack]

Thanks lineman55-

[N242AM]

Has anyone try one of the T.O.O.B setups yet?


http://www.thinkoutofbox.com/usa/

[Neil Hewitt]

It looks very cool for gaming, but because of the mirror position the seating / viewing position has to be outside the dome itself. I don't see that it's much more immersive than a single big screen in front of you in flight sim terms. What most people would like to approach must be something like Ivar Hestnes' 225-degree wraparound view. I suppose the mirror + dome arrangement goes some way towards correcting the parallax problem, depending on exactly how they've sized and placed the mirror, but the light will still be divergent and apparently focused at a point not so far ahead of you.

I could imagine you could use the same technique - project through a small hole in the screen onto a mirror and then reflect back - with a simpler curved screen and mirror sheet curved in a single axis, set at an angle towards the viewer. That might be worth researching for my own setup, actually. Thanks!

[Ferrino]

Thanks for all the replies!

I was particularly interested to learn that increased throw can alleviate some of the "defocusing" that occurs when you project onto a curved screen (i.e. "blurry side views").

One of the main uses of the curved screen I am implementing is actually driving simulations, where the lateral views are typically seen in your periphery and you don't focus on them directly (except when looking into apexes etc). So I guess the focusing problem is less severe than in flight sims, where you will spend more time looking directly at the sides than in driving sims...