|
|
For the mapping gurus-does this technology exist? With all the amazing thing going on with Wii's, and augmented reality, etc, I'm curious if this type of technology exists: I'm wanting to have unusually shaped pieces of fabric on the stage (all two-dimensional) and I'm curious if I could attach sensors of some sort to the edges of those and have a way to automatically map those areas so a projector would only project within the boarders of those sensors. Basically looking for a fast way to projector map unusual shapes so you wouldn't have to manually map every time you move the projector. Any thoughts? |
OK-I see a couple of videos here that seem to be able to do this: http://www.youtube.com/watch?v=ap46efdp58I http://youtu.be/XgrGjJUBF_I
So now the question is how do I do this? Where do I start? Do I need a kinect? Can anyone point me to kinect mapping tutorials?
How quickly will the shapes be moving?
Johhny lee's other video would also be of interest to you
http://www.youtube.com/watch?v=nhSR_6-Y5Kg&feature=relmfu
You could use and IR camera to find the corners of your fabric and then just set the corners of a quad to those points
Yeah, I saw that video, too Very cool indeed! Actually, the shapes will not be moving other than possible stage vibration, accidental hitting, and accidental projector shifts, so I think scanning every second or so would be more than adequate. The important key for me is being able to set up and map quickly and the ability to set up even in daylight-IE: even if we are setting up outside in the day at a festival and the projection won't be happening that night.
if the shapes won't be moving IMO there's no point in tracking. Setting up a projection during daytime is ... not recommended.
Is there no technology where I can use sensors around the edges of the material that will give the coordinates of the boundaries of the shapes? It seems like this should exist, and should be able to work regardless of ambient light situation.
We play in situations all the time where we have to set up in the day and then play at night. it's not always possible to set the projection at night right before the gig (festival style schedules, etc). There must be a way...
but what does your "complex" shape look like ? Any single convex/concave polygon is mapped in seconds (5-10 minutes during the change-over is generally enough for a, say, 30+ points shape)
plus you have the problem of the sensing device (camera/kinect) being in a different position from the projector which makes this a non trivial task to automate with a computer
Ok. I understand there are difficult points to this, and it sounds like the only practical way of mapping polygons involves using the light from the projector itself as part of the calibration-is that correct?
We'd like to have multiple polygons on stage-each ideally represented in a layer in something like VDMX. 5-10 minutes to map one complex polygon seems like a very long time when the above videos are showing technology where he's doing 4 and 6 points in a matter of a couple seconds....
What are the options for mapping? The projector will be 40-60 feet away from the stage usually....attached is a pic from last weeks show where the projector is going on to the whole band.
bha, the show seems nice however I don't see any complex shape there. It just looks like a plain picture projected on stage.
FYI, this project takes about 7-10 min to map: http://www.1024architecture.net/en/2009/11/boom-box/
and this one less than 5: http://www.1024architecture.net/en/2009/11/square-cube/
that one, however, took a night: http://www.1024architecture.net/en/2010/09/3d-bridge/
with the latest tech (aka camera matching), this one, although quite complex due to perspective deformations, took me only 5 min (but I had the 3d model): http://www.1024architecture.net/en/2010/11/perspective-lyrique/
In MadMApper, we also have a Space Scanner module that allows to turn the projector into a ... scanning device, which is quite handy in some situations. The technique is called Structured Lighting.
That's because the shapes aren't there yet-no reason to have them there if I don't know how to do the mapping yet... ;-)
I was just wanting to show you the scope of the size we're working with.
Thank you for the links-I will take a look at them when I have finished giving a music lesson here in an hour...
o i c , well, if the shape is constant over each show, having it in a perspective quad may help reduce the amount of control points to setup.
for instance, a n-gon can easily be contained in a quad, so instead of having n points to warp, you just have 4 ... depending on the shape, there are always a few nice tricks to reduce the setup time.
the ultimate trick being the mapping process as a part of the show itself |:)
Those videos are beautiful. That house is amazing!
OK, so a couple points:
We play on different size stages, so even if we try hard, the space between the different polygons will be different from night to night. Also the angle of each will be slightly different from night to night.
If I'm understanding you right, the perspective quad would be like fitting the polygon inside the boundaries or a rectangle and then mapping the points of that rectangle. I guess then I'd have a pre-rendered mask that would fit the actual boarders of the n-gon and have that live inside the rectangle-is that correct?
So what is the technology I need to be using for such a thing? How do I find those control points?
It seems like a kinect would be too short range for such a thing. Can anyone recommend a link so I could get started? Thanks so much-I appreciate your help-as I know I am a real newbie and I appreciate you taking the time to help a new guy...
try _1024_Perspective plugin. it doesn't have a mask input like the built in sprite, so you'll have to mask it yourself using some other built in patches. (this is due to the fact that I never found any explanation on how to implement image masks in the official API --- any info welcome on that matter --- )
I don't understand-the perspective plugin is basically a plugin that is altering the keystone and perspective, right?...I'm trying to figure out what will be the best way to capture the edges of each of the polygons...Something like a tutorial on how to use those sensors in the first video I posted. Similar the the Scanning you guys are doign on the MadMapper site, but I don't want to have to remask in photoshop everytime we move the stage...
So at this point, is the only way for the common man to map is manually or with a DSLR camera? This guy is the only one who has managed to use sensors? :-p
There's been some library made recently that uses visible structured light for calibration, in a way that looked somewhat similar to what was going on in this video, but without needing sensors I believe. I'm pretty sure I saw something about it on the open frameworks add ons page. If I run across it again, I'll post. Or maybe it's easy for you to poke around and find it.
is that in the downloads section? I've been poking around a bit so far with no luck...
http://code.google.com/p/structured-light/wiki/ofxStructuredLight