I’ve been on family vacation for awhile and I am just getting caught back up enough with work to start messing with my table again.
Sorry about the delays in responding to email/posts. We were in Montana where the internet does not seem to reach yet
@tetsu81vn: Thanks. I may post a “blueprint” so to speak after I get some time. For now it recognizes fiducials and fingers only.
@haythem: I am tracking fiducials with reacTIVision and passing that to my application. We are writing the client app in C++ at the moment.
@JasonB: Here is a link to the IR arrays. I am using 4 at the moment. They take awhile to arrive. http://www.goodluckbuy.com/ir-infrared-48-led-illuminator-light-cctv-night-vision.html
I have recently switched to a Rosco Gray material from the vellum because I couldn’t stand replacing the vellum after a few uses. It really is easy to scratch/rip/dirty the vellum and I also has issues with it staying flat on the surface. The Rosco Gray provides a great image and it a much more durable surface though it has decreased the sensitivity and fidelity of the touch surface. I am hoping that by increasing the amount of IR or better diffusing the light I can get my sensitivity back up. I have purchased some Mylar material to attempt to line the inside of the table to get some better diffusion/illumination. I will let you all know how it goes.
Edit: A quick note about the IR arrays. They have a photocell build in that switches them from low intensity to high in darkness. Put a little electrical tape over the sensor for better illumination.
SpynalTom, your work is amazing. I am building a similar DI table for my thesis in multimedia. What are you using for your surface? Acrylic, glass? And how thick it is? I don’t know if one is better than other and I’m trying to buy the right one. Here in Argentina every part is more expensive so I can do much experimentation. Did you put the vellum (and know the rosco on top of the surface material and work directly on it? What happen if you put a very thin layer of acrylic to “sandwiched” the rosco just to protect it? I´m planning to try that once I build mine.
I´ve already bought the ps3 cam, same projector as you and built the table structure. The more difficult part for me is getting the right leds. The illuminator you all use here cost almost $50 each! LOL. I’m trying to build my owns just with leds. But the leds I got didn´t give me good result. I´m currently using as a filter for my ps3 a piece of floppy disc. Could it be too much? What would you recommend instead of the floppy?
Just a quick update to all. Thanks for all the positive comments and PMs. It is really motivating. I recently lined the inside of my table with Mylar sheets to help diffuse some of the light around. To make a long story short I just need more IR arrays, simple as that. Though I believe the Mylar helped diffuse some of the light a bit it also created some reflection of the visible light and ended up making little tiny random shimmering star bursts on the projection table. The more experimenting I did the more I became aware that what I need is just more IR light. I currently have 4 IR arrays with 48 LEDs in each. I have ordered 4 more. With a table this size I think the more I have the better, but we will see when they arrive. When I was experimenting, the more light I added to an area gave much better tracking results in reacTIVision. For all you folks out there experimenting with reacTIVision here is a tip:
Red “+” = Barely tracking. Lots of jumping and shaking. (BAD)
Blue/Turquoise “+” = Tracking but still not the best. (GOOD)
Green “+” = Solid motion, speed, and rotation tracking. (GREAT)
Seriously until I started messing around with more IR light I had no idea there was even a Geen “+”. LOL I though I was doing well with the Blue ones.
Some folks have asked for a copy of the client app I use in the videos. It is based on the TUIO C++ demo program. Please do not give me a bad time about the ugliness of the program. I threw it together for testing and was not worried about form or style. I also provide no support with this code. I am sure you will need to recompile to use. Use at your own risk.
Also if anyone is interested in reviewing or helping out with the real application here is a link to the GIT repository. This is still in the very early stages so please don’t expect much. If you don’t have a table it works very well with the simulator.
@diegom22: I am using a glass panel with Rosco Grey on the top. the glass is 1/4 inch thick. The problem with putting anything on top of the vellum or Rosco Grey is that the fiducial will become blurry and difficult for the camera to track if it is not making direct contact with the material. It may be OK for finger tracking but for fiducials I don’t recommend placing any other acrylic or glass on top of the vellum/rosco. Also instead of a floppy disk try a bit of negative film from a developed roll. Find the darkest part (usually at the beginning of the roll). Be warned that once you remove the IR filter from the camera and replace it with an IR pass material you will have trouble focusing the camera due to the stock lens housing not allowing for manual focus. I cannot stress enough the need to buy a cheap web cam with a similar housing and replacing the stock one. This will save you a lot of headache when trying to focus the image after altering the lens. Check the forums for a cheap web cam with the same lens housing.
Here a couple terrible cell phone camera shots of the table in action. I will try to get some better pictures up soon. There is just not enough time in the day. The gaming group really seems to enjoy using the table so that’s rewarding.
Wow ... sometimes Google can show a piece of Nirvana.
@Tom - This is EXACTLY what I want to do for my gaming group! I first saw a demo of some surface computing stuff several months ago and put this project on my “geek to-do list.” Has there been progress since August? I haven’t had a chance to look into the code, yet.
I love your table and it is what brought this community to my attention. I’m new to this community and about to embark on my own MT table project. I’ve been looking at the various techniques and associated pros, cons and costs… DSI is probably the best way for me to go since I want to do a D&D coffee table, but when I started looking at the cost, rear DI is probably more in my budget.
If you don’t mind, I’ve got a couple questions:
1) I saw the 48 LED illuminators you pointed to and was wondering if you have any concerns regarding eye damage from IR over exposure?
2) Are you running the illuminators using the 12v plugs on the PC power supply?
3) Don’t suppose you know of a good and cheaper projector? I have a tiny budget and really can’t afford 600 or more for a projector.
4) Are you still using a vellum or other “lying on top” diffuser? I read somewhere that frosted glass would work equally well?
I expect commercial products available in the next few months if you don’t mind 848x480. I’m thinking of building a variety of display tables (potentially for sale, at some point). The D&D table fascinates me for personal reasons and my first prototype will serve that purpose, as well.
As far as selling tables goes, I’m thinking of a Digium/Asterisk model: Software development would be community based for the main controls. Hardware and specialty software (like D&D content) would have options to be commercial, but there would necessarily be tons of free content, as well.
I’ve sent off for Microvision’s Dev kit, but haven’t heard from them, yet.
I have a quick answer on projectors: Wait. These guys change everything, from mirrors and throw angle to using a flat projection surface and nevery having to focus ...
Hmm.. I’ve seen these things when I started orienting on what I wanted 2-4 months ago. The projectors are quite nice. But the amount of lumen they put out is not very good and their price is still quite hefty in my opinion for what they deliver; around 300.
I think I’d rather fork over another 200 for a projector where I can actually access components when needed. Its more a question of “can I find a good short throw projector for little money?”.
It seems that the main reason you have for adding more IR illuminators is poor finger-tracking illumination.
Perhaps your setup might benefit from a combination DI/FITR setup. DI gives you even lighting for fiducials, however, FTIR gives superior blob performance. Since you are using a compliant surface anyhow (rather than attaching your screen material to the underside surface), perhaps you could illuminate the acrylic edge-on with strip-LED lighting, then place your Rosco grey over a compliant surface (or even silicone coat the bottom of your roscoe).
In this way you could get good finger tracking. It shouldn’t take much IR lighting to light the Fiducials (it seems that it is more important to have them evenly lit than brightly lit), probably a lot fewer LEDs than you already have in your arrays. Considering that you should get some illumination from the edge-on IRs, I bet you could get away with as few as 10 or 20 illuminating LEDs projecting to the bottom of the screen.
It seems that the major advantage of a DI (or DRI) setup is
1) having your fingers on glass rather than a compliant surface, and mounting the projection material on the underside vs acrylic covered with a damagable surface (FTIR). and
2) Fiducial/Symbol recognition.
Considering you are already using a projection material on your top surface this seems to be the ideal situation for a hybrid DRI/FTIR setup.
I am inspired by your DI project and hope to try a hybrid DRI/FTIR setup myself now....
Thanks for all the info about your project, it looks great. Just wondering did you have any trouble with the reflections caused by your acrylic that can be seen in your screen grabs. We have just got our table up and running and are worried about them or does the background removal in reactivision take care of them?
Hi it’s very interesting, i have one question please can somebody tell me which are all the materials for all the surface i mean acrilyc, vellum, plexiglass, did you use rear projection foil?, silicon rubber? infrared blocking foil thank you.
@13alejandro90 : He’s using Rear Diffusive(or Direct) illumination. Different setup than FTIR. As for your other question(about surfaces), you need a surface material( glass or acrylic for example) and a shhet of diffusive material under it(like vellum or roscoe or 7d513 etc)