2 Place Collimated Display

[fsaviator]

So, back in Germany now. I've got a few minutes to post

While in Colorado I was able to visit Jack and his collimated display build for a dual seat 737 ...all I can say is pics and video don't do it justice. This thing is fourteen plus feet across and after mounting the hanging projector screen, it is almost eleven feet tall. Impressive and truly a study in patience. If anyone has looked into the cost of 1 mil mylar, you'd know why.


Warren

[castle]

The next time Warren visits, we should be up and flying.

The total mirror is actually 17 feet in outside diameter. the mirror radius is 8 feet and the frame or collar to hold the mylar and back panes to form the chamber is 6 inches in width. Add the projection screen and supporting frame and it is just over 10 feet in height. You might be able to get by with a 10 foot ceiling if you eliminate the frame and suspend the screen with cables. Not an insignificant engineering task. The sim room is 22x17x11.5 feet. While adequate, a 22x20 floor area would have been nicer, but the add-on was a compromise with the wife's wishes for a deck area above the room. Anyone contemplating a project like this might want to settle for the smaller version Gene designed for a 14 foot mirror.

As to the mylar, it is not totallyl unreasonable in price, the killer is shipping costs. Way too large to meet UPS or FedEx limits due to length, so you have to use freight which requires a large wooden pallet to hold the mylar roll rigid and prevent damage and since you pay by the pound and distance the charges include the pallet weight that exceed the cost of the material which itself can be heavy. So might as well buy enough for a couple of reskins and then some.

Cheers
JW

[castle]

Decided to do some wok on the cabin shell. Don't need to be moving big and long pieces of material around with the mylar in place. Will do all the detail work later.

Once the frame, overhead canopy, and exterior is done will move the platform back and out of the way. And that will provide a five foot buffer to work with for installing the mylar. Still a few weeks away, maybe by the end of March. The first flight date keeps slipping. Starting to feel like Boeing.

Cheers
JW

[castle]

Hi,

For anyone in project management you know schedules should never be cast in concrete but we're getting closer. The cabin shell phase in completed for now and ready for roll back. And the forward projection screen arms have been removed. When it comes time to install the mylar the rear mounts will be raised to clear the collar and camera and mounts removed providing an unobstructed space to layout and align the sheet and avoid the problem we had with the first effort where the arms got in the way.

Still a few more tasks; first need to measure and establish the mount points for FOV of 50 degrees per projector with a 2.5 degree overlap on either side (reduced from initial setup of 55) , then redo the warp mesh. Decided to spend a little more time here with the existing BenQ ST projectors and rewrite the software and make it more interactive and flexible; something that was planned with the new HD units. Besides Gene and Wayne suggested that the glue works best at warmer temps and spring is almost here and while I heat the shop while working it cools overnight. Takes a few days to get a good cure.

Also spent some time with house-cleaning, but want to do one more detailed pass. Amazing the amount of dust and debris one can generate and where it can hide!

Well, that's it for now. Patience is the watch word.

Cheers
JW

[castle]

Finally ready to reskin the mylar.

here are two pics with and without blending. The ST BenQs are okay for now, but just don't have the FOV and pixel count for a decent overlap region for this setup. Main goal was to come up with a good algorithm to handle the luminence of individual pixels. Using shaders and GPUs to run the blending. Note how well the white clouds are handled.


The line down the screen is a piece of string to help reference the center line of the screen

JW

[geneb]

Jack, is that blending done in code or did you have to "stitch" it manually like what is done in NThusim+ and Sol7?

Looks great!

g.

[xplanematt]

Fantastic work!!! This is a huge project, and it looks like you are carrying on just fine with it! Question: about how wide is your reflective overhead assembly, and how wide is the actual Mylar mirror frame? I am planning a similar setup, and I'm starting with a 10ft diameter back-projection screen from a commercial level-D sim (yep, that saves me all kinds of money and work). My cockpit is about 8ft wide (Sabreliner 60). Reason I'm concerned about the width of the Mylar screen I'm going to need is that I am in the planning stages of building a shop for this beast....if I orient the sim the way I want to, it will have to fit in about 20ft of width. If I have to, I can rotate the sim 90* and get close to 30ft, but that will make ingress/egress of the sim potentially more troublesome (it will be on a trailer for easy removal/maintenance, and for taking to shows).

I'm also interested in the software. It looks like you are using your own, which is awesome. Any plans to share the code? I am using my own code for all my interfacing and avionics etc, but the "high end" math required for advanced optical stuff (mostly, anything that involves a lot of calculus) is not my strong area.

Matt

[castle]

Jack, is that blending done in code or did you have to "stitch" it manually like what is done in NThusim+ and Sol7?

Looks great!

g.

It's done in the code and it will get better. Using shaders to run on the GPU; NOT the CPUs. Started with some basic algorithms and working on some improvements based on that paper out of UCI I sent you a while back. Basic approach is to determine the position and color of each pixel in the overlap/blending region, apply the algorithm, and adjust the color and luminance in the frame buffer before it is sent down the graphics pipeline.

JW

[castle]

Jack, is that blending done in code or did you have to "stitch" it manually like what is done in NThusim+ and Sol7?

Looks great!

g.

It's done in the code and it will get better. Using shaders to run on the GPU; NOT the CPUs. Started with some basic algorithms and working on some improvements based on that paper out of UCI I sent you a while back. Basic approach is to determine the position and color of each pixel in the overlap/blending region, apply the algorithm, and adjust the color and luminance in the frame buffer before it is sent down the graphics pipeline.

JW

[castle]

The 2nd hanging went a helluva lot better than the first. Removing the forward arms of the frame and suspending the projection frame provided clear access to the mirror and allowed for easier draping and positioning of the mylar. Couple of spots that might not meet Level D certification, but good enough for now.

Temptation is to attach the vacuum and suck down the mlylar, at least a little bit. But listening to the sage advice from Gene to first attach the clamping mask before the draw down. So that is a still a few days away.

Stay tuned
JW