Our Mixed Reality Table is built in DI technique and supports Full-HD resolution.
We solved a basic DI problem by realising a homogeneous IR-illumination for a large surface with 51” (130cm) in diagonal.
A technical report about the details of the illumination is in review process currently.
Already now, there are lots of images and plans on our website.
that’s Huge & absolutly insane !!!!!!
a NASA project ? how many smart guys were involved in that project ? how long does it take to build and test that “ mr (e.)T “ ?
Can’t wait to watch your 3D Modeling apps and Motion Capturing in action !
Are you doin’ home delivery ? Sure; it will fit in my workshop !
how many smart guys were involved in that project ? how long does it take to build and test that “ mr (e.)T “ ?
We spent eight weekends for the construction realisation. During that time, up to three group members were working in parallel. Six members were involved altogether.
We also spent some weeks on construction pre- and postprocessing.
Our goal was to use a surface out of glass. In our test settings during project start, we found, that it is very hard to realize a homogeneous illumination without distortions in the camera image caused by IR-LEDs and their (direct) reflections under the surface. With classical illumination techniques you either need a broad frame around the surface or use multiple cameras (as in MS-Surface). We wished to have a narrow frame for ergonomic reasons.
Or solution is a crossed illumination with shadowing (see side-view plan and shadowing images).
Details on the dependencies of the arrangement of surface, IR-LEDS, camera and projector will follow in our report.
We looked for a Full-HD projector with a small projection distance for a diagonal of 130cm. Unfortunately, we didn’t found a device with a projection distance much less than 150cm. So we had to install three mirrors to establish this length virtually.
We didtn’t want to use an auxiliary wide angle objective because of deformations, color fringes and other image distortions. Additionally, their price range is exorbitant.
If anybody knows a Full-HD projector with a higher ratio of diagonal divided by projection lenght, please let me know and I’ll drop the mirrors for the next design.
but there is no substitute for resolution. I have two mirrors in my setup and have also eliminated the need for any keystone by tweeking the angles just so.
I would like one of those pany’s for my home cinema and the 3000 has lens shift memory oh yeah!!
I agree - especially if you like to build a real multiuser device, with users having their own working space.
We also eliminated the need for any keystone.
Yes i had originally intended to use two 1024 x 768 projectors side by side to increase the res but i got so frustrated with the fiddly nature of two that i just used one projector instead! I would like a 1080p projector but for me it is not that nessacary.
One other issue when butting together projectors is the shade variance and colour difference between them even two identical projectors can have slightly different colours, bulb gas mixtures etc. I did a gig for Cold Cut (name drop!) They like to use two 4:3 projectors on a 16:9 screen. We had two 6k sanyos side by side with identicle setup and the difference was shocking!
Ultimatley having 4 1080p projectors on a simalar sized screen to the one you have shown, would give almost infinate fidelity! If there is such a thing.
I would like to see a surface that is as high res as paper!
Ultimatley having 4 1080p projectors on a simalar sized screen to the one you have shown, would give almost infinate fidelity!
If you compose a rear projection with multiple projectors, you have another problem in addition to the shade and color problems you mentioned.
Our experience is, that the diffusion never works perfectly. There is slightly stronger illumination power in the direction of the beam originating from the projector. If you walk along the projection surface, you may get the impression, that a part of the bightness is walking with you. This effect produce contrasts at the boundaries of the images of projectors with different beam origins.
I would recommend to wait for projectors with higher resolution ...
I’d like to know what kind of projection screen did you use for your setup. Was it glass or acrylic? What thickness? What is the projection surface? Is it on the projection side or on the user side?
I am getting together the stuff for making a 48” vertical rear DI wall and wasn’t sure about what to use for the projection for a large MT wall like that.
Thanks a lot,
Greg
The projection screen is built as a two-layered glass ‘sandwich’, with the diffusion foil in between. The distance travelled by the light between the diffuser and and object on the surface should be minimal. Thus the top glass layer of the sandwich is 2 mm thick, while the bottom layer carries the weight with 6 mm thick.
The diffuser needs to fulfil several requirements: It needs to diffuse enough (visible) projector light, so that the projected image is visible. Too much diffusion makes a too dark image, while too little results in a ‘directed’ image, meaning that the light intensity of the projected image varies depending on where the user stands. The diffuser needs to be as permeable for infrared light as possible, it should allow infrared light to pass through and be reflected back again. To determine the optimal trade-off between these contradicting requirements we performed test trials with rear projection screens and colour filters. Rear projection screens proved great for displaying the visible image but were not very permeable for infrared. The LEE diffusion filter ‘Neutral Density Frost 225’ offered the best compromise for us.
Thanks for the explications mr.T.
I had the feeling that the sandwich layered glass was a good configuration for these large screens, but others told me that rear DI wouldn’t work with this two-layered method. But its good to know that actually it does work, and probably the diffusion filter is one of the key elements to make it work.
I saw that you use custom IR led arrays for lighting up your box. Have you tried those round IR illuminators, that can be bought on eBay, or you went directly with your custom system? Do you use the Soft Silver Reflector 273 Lee filter to bounce IR around in the box?
The problem with DI and the two layered method is that direct reflections of the IR-spotlights appear in the camera image.
Yes we have custom IR-Spotlights. Of the shelf arrays would not work in our setting. The reason is described in our technical report which is online now.
A lot of details of our construction are given in this document - especially our solution for a homogeneous IR-illumination.
Have a look.
The choice of the silver reflector material for the narrow sides is not critical. Just take any good mirroring material.
Wow, your technical document is awesome! Can learn a lot from it!
Now I am planning to build something like what you see on the image. So it’d be quite different configuration from yours. I planned to buy IR reflectors and bounce the light off of the walls covered by Lee 273 reflector.
After reading your document I could think of problems of unwanted IR reflections from the elements inside of the box. If this occurs, I will probably also have to walk the directed IR light and the shadowing method.
Thanks again for sharing your knowledge, it is really helpful.
Hi Jens,
The problem I have is that the specs of the spots are not listed on eBay, so I don’t know the projection angle. And the question I have is that if turn them in the opposite direction from the projection surface to bounce them off of the walls, how can I calculate the cone if the lee filter will disperse it in all directions?
There are many approches, that use an indirect diffuse illumination. Some of them have the diffusion layer at the surface back side. In this case you have no problem with reflections. But then the front glass itself is the load bearing part. For a larger surface, this requires some mechanical thickness, which implies blurring of the fingertip image. Other approaches use software to correct distortions. E.g. the reactivision software analyzes local gradients to overcome inhomogeneous illumination and external distortions. But this is error prone, because small changes regarding to the calibration setting result in detection information. It also reduces the camera image dynamics.
In the mr.T setting, we have a quite homogeneous IR-illumination with no distortions. For the detection we are able to use constant thresholds.
In addition the illumination is powerfull. This yields a good signal to noise ratio, concerning external distortions. The camera image dynamics allows for a robust detection of hovering hands.
