I’m trying to understand touchdesigner’s role in projection mapping. When I first thought about diving into projection mapping, I assumed it would be a matter of adjusting geometry, texture projection, and camera settings within touch to ‘map’ it to a real-world scenario.
Now that I’ve started experimenting, and doing some research, it seems as though TD is perhaps not fully suited for calibrating a projector to project on 3D surfaces, and that people are using 3rd party software in tandem to get results like Vsquared labs, digital obscura, etc. Like the ISAM project, when they adjusted the projector using the ipad-- was that not all done in TD?
To be clear, I’m talking about using 3D models to achieve proper warping, not using something like kantan mapper. I’ve seen a couple posts regarding an update to the camera comp in 088-- is this the first time touch can do 3d projector calibration natively?
Hopefully you guys can clear it up a bit for me. I guess I don’t want to start creating some crazy rats nest of TD ideas to try to warp my image properly if there is already a solution that people are using.
The long and short of it is that there is no turnkey solution at this current time for 3d projection mapping. My belief is that once there is, it wont be a thing any more
That being said all of us are using different methods that are typically all oriented at the same goal of using 3d models to map instead of distorted flat shapes. The trickiest parts of that whole procedure is getting an accurate location for your projector in the virtual scene and matching the virtual projection surface to the real world surface.
These two problems occupy the majority of effort being made amongst several companies and individuals to tackle 3d mapping. Everyone would like a turnkey tool, and there are some making steps towards that (see Map a Mok type stuff for generating a transform matrix for the virtual projector), but really every project poses its own issues and inevitably there will be lots of work put into adjusting for the specific nature of each installation.
One thing that I have found has helped me greatly in the adjusting geometry part of this is the lattice SOP which allows you deform your already textured geometry to match the scene once you’ve gotten as close as you can with the camera/projector location.
In the end, you’ll be hard pressed to find anything out of the box (I think pandoras box has some of this, but, well… I’ll hold my tongue). Most of us doing it here find all the tools we need within TD, and maybe a bit external C++ (utilizing the C++ CHOP/TOP). I know I wouldn’t want to be using any other program to do 3D projection mapping!
Hey archo-p, thanks for the reply! I agree, turnkey would kinda take away much of the impetus. It’s good to know that people are using the tools in touchdesigner to figure it out. I’m slowly learning to be confident in the raw potential of TD, and not to hesitate for hope of an existing solution.
It always comes down to learning a programming language I’m slowly gettin there.
All projects are different and I attack the situation differently depending. If I know the exact measurements of what I’m dealing with in real life (and the measurements are accurate and consistent), then I tend to make a model in TouchDesigner, map everything onto it, then do minor adjustments/warps in person if they’re needed after lining up the projector.
But on a side note, I end up spending far more time working on the content than I do the modelling and mapping, which I didn’t expect when I started my first projection mapping project. They don’t mention that in the manuals!
It is a fact that Installing 3D Projection mapping has become an inevitable part in business marketing as they can produce stunning 3D images of an objects it is easy to convince a customers …
The best approach I’ve found is to use a mapper system to get close, then distort the model. Sounds like Peter (archo-p) uses a lattice sop, which is a nice trick. I just use group SOPs and translate the vertices that don’t line up.
The last part is very time consuming, but the first part is so quick, it’s moot.
The other trick I’ve used is to make a distilled model from the key points of your real model, and use that for alignment. Since the mappers tend to make a camera projection matrix, it can apply to the simplified model and the real model just as well.
But yes - we use TouchDesigner almost exclusively. There’s a place for high end 3D systems in making the content, but that doesn’t seem to be what the OP is concerned about.
as they can produce stunning 3D images of an objects it is easy to convince a customers … 