Together with Peter Bomark, Johannes Hirche and Tim Roth we have been working on a Diffused Illumination setup. We use 4 illuminators, frostened lee filter on the inside of the box, a 3m 700 DMS projector, lee filter as a diffuser and an unibrain cam.
Other then that I think my personal preference goes out to FTIR, because it’s more robust especially in different light environments, even though you can use fidicuals with DI, it’s less robust, because of inputs that you might generate with for example your wrist.
With our FTIR setup I can generate inputs by blowing hard from very close. So I hardly have to touch the surface with my fingers, very smooth. Other then that it’s easier to use for example brushes and pen alike materials.
Maybe hybrids are worth testing out? I mean, i’m working with FTIR, not really sure how come it is so difficult for some people to get working and like you said, it is more robust. Although, I would like to recognize patterns too so I wonder doing a duel setup is too far off the path for me.
Nice setup by the way, the design is simple yet effective and looks professional. You guys should be proud.
Thanks ethical, yes a combo might be interesting to test out. With DI you can use fidicuals, but FTIR is very robust and I think pen support is better as well.
Other then that I think my personal preference goes out to FTIR, because it’s more robust especially in different light environments, even though you can use fidicuals with DI, it’s less robust, because of inputs that you might generate with for example your wrist.
I’m a bit confused about your statement that FTIR is better for DI when it comes to different lighting conditions since DI has been used in the Citywall installation which worked outside throughout the day and I found my FTIR table was really sensitive to light. Do you have any advice on how to decrease it’s sensitivity to light (eg a IR blocking film)?
Could you provide some details about the Lee filters and the exact surface settings ? Details on DI setups are quite scarce… Thanks a lot !
And by the way, I understand that it is desirable to have a very diffuse rear IR lighting, but I somewhat expected the insides to be black painted for the sake of the image projection (ghost lights, contrast, etc.). I think I’m totally missing a point here, could you “enlighten” me ?
Could you provide some details about the Lee filters and the exact surface settings ? Details on DI setups are quite scarce… Thanks a lot !
And by the way, I understand that it is desirable to have a very diffuse rear IR lighting, but I somewhat expected the insides to be black painted for the sake of the image projection (ghost lights, contrast, etc.). I think I’m totally missing a point here, could you “enlighten” me ?
Check out Tim Roth’s blog for more info. He helped on the design. http://iad.projects.zhdk.ch/multitouch/
The reason for the inside being light that is so the light is spread around evenly inside the box. This shouldn’t affect the projection at all. The only light really being reflected is the IR coming from the illuminators. Only minor light would be reflected back from the projection.
Other then that I think my personal preference goes out to FTIR, because it’s more robust especially in different light environments, even though you can use fidicuals with DI, it’s less robust, because of inputs that you might generate with for example your wrist.
I’m a bit confused about your statement that FTIR is better for DI when it comes to different lighting conditions since DI has been used in the Citywall installation which worked outside throughout the day and I found my FTIR table was really sensitive to light. Do you have any advice on how to decrease it’s sensitivity to light (eg a IR blocking film)?
true, CityWall window was outdoors but does anyone know if they had lighting issues or not? I think FTIR is better with environment issues beccause the blobs are really bright and vivid.... in my experience my DI set-up was more sensitive to lighting conditions. But again, my DI setup was for fiducial tracking so the patterns had to be crisp and sharp, and not blurry or blobby. I also found that different projection surfaces play a big role too. Mylar is great for diffusing any light whereas tracing paper lets a lot of light through. I’m surprised to hear that other ppl are having lighting issues with FTIR; what projection surface are you using?
i think they best way to avoid exterior light pollution is to use glass that has been casted frosted.... not a coating or a projection surface. this is the projection material that smartboards use (albeit, their tracking is not multi-touch and the cameras are on the frame)
There has to be way to track fiducials using FTIR, we just havent found out how yet. I would want to try etching the patterns on the bottom of an object and see if that will work because in theory, only the positive-space of the pattern should reflect light back..... but this is a hunch so far.
at joobs, u can get IR blocking films from ifoha, for example http://www.ifoha.de, they block some IR 60-87% or so i believe, they have 3 different ones
I’m curious how you got your FTIR compliant surface to be so responsive. Are you simply using silicone + diffuser on top? What shore hardness gives you such great results? I have 40a with just a thin sheet of tracing paper on top and it works quit well but not as well as you describe yours! I still have to apply a little pressure.
Yes silicone rubber with a diffuser on top. I think its shore 40 as well. I have to apply pressure as well, the pressure of blowing really hard. But we use a protective film now, so I’m not sure if we can blow input anymore hehe
at barrante, just a regular lens from unibrain, dunno what size mm though
Ok. The reason I ask is, I just purchased a 4.7mm 42.50º Lens, which a few people have used here. Through further reading, I have also come across examples of people using the wide-angle lenses. But with that, you have to rely on the software to correct the distortion - and I also couldn’t find a wider angle that didn’t have the IR filter on it. It’s nice that the Unibrain camera lenses are fairly affordable and I can be easily swapped out - especially if you have the board camera.
would this setup, projector/camera wise, also be advised for a reacTIVision project? so we are tracking fiducials (pattern recognition), but also using infrared so I guess the spec’s are about the same?
sorry for my ignorance, but I am having trouble finding info on fidicuals-- by context I’m guessing: ‘detecting hand gestures’ ?
I think with programming, error proofing DI is not too far away. Giving it a dynamic threshold to work with, such as the citywall did with daylight/nightlight conditions, is a great start. Even ignoring incorrect detections like wrist/sleeves, or random light sources is very well possible-- it will just take time, and trial and error...just my ramblings in the morn
