I want to do a projection mapping on face mask as this one. It is build of cardboard and is 2 meters high.
I have made the abstract Animations that I fitted to the front view of the mask. Now I wanted to start mapping with two projectors, one of it slightly positioned to the left and one right of the mask.The Animations have texture as well as animated white outlines that correspont to the borders of each single shape.
I started to use Kantanmapper. But I have some problems with the white outlines. I discibred the problem in my previous post. https://forum.derivative.ca/t/kantanmapper-problems-modifying-shape/237193
Now I ask myself if Kananmapper is not the right solution. How would you approach this task? Or do I need some other tool like Camshnapper for this? I never used it and dont have great experience with 3D. Yet
In this case you wont have much fun with CamSchnapr either as it basicly I used to reporject the model onto the real model. This means you can do the whole mapping inside of TD or even an external software using UV-Mapping. Also, you will have some nice previz. Any chance for you to get a good model of the face in any way?
I suggest you create a 3D mesh of any object you want to map, especially if it is dimensional (non-flat) like a face.
Blender is a free application you can use to model it and then import into TouchDesigner as an FBX file. Many professional 3D modelers start with JPG image references like the ones you have, in order to “trace” the object in 3D and thus re-create it. While 3D is a steep learning curve, it’s well worth the exploration, especially as an investment in your knowledge for the future. (Life is 3D; ergo, all software is slowly moving towards 3D. Think ‘metaverse’ and beyond).
Regarding manual alignment: it is quite difficult to accurately render the correct lens distortion, field of view and perspective from each of your two projector positions manually; the lines will almost certainly never line up around the nose and eyes and brows, since they are protrusions that appear significantly different from various viewing angles. It’s very very tricky. Honestly far better (and easier) to import the 3D model and add two cameraCOMPs to your network and manually “fit” them into place before going the kantanMapper route, unless you’re only using one projector. But two or more means a lot of additional work.
Always remember that a projector lens is like a camera lens: they both view the world the same way, one emits photons (projector) while the other absorbs them (camera). Hence why camSchnapper is such a neat tool - it calculates your “camera” (projector) for you. Again, well worth some exploration.
Hi Dylan, thank you for your thoughts. In the meantime I looked at Camschnapper and it really looks like the tool to choose. And I would definetely choose it in Future and dive in this “new world”.
The point for now is, that I will need the Mapping quite soon and the animations are done already. So, to build the mask and redo the animations on the “unfolded” UV Shapes of the 3D-model would be a quite tight task.
I understand that Kantanmapper is not the perfect tool. But since the result must not be 100% (its a student-project) i would still hope to avoid the 3D way. So if there is a chance to to it with Kantanmapper I would love to find a way for this time. In general it works doing the mapping for each Shape in Kantanmapper, but my problem are these disappearing white outlines. I posted this already in the forum, but until now didnt have any reply. Do you have an idea to that? https://forum.derivative.ca/t/kantanmapper-problems-modifying-shape/237193
