Hello all, Firstly let me start by saying I’m a 3D and cg/vfx enthusiast rather than a professional. I’ve only been experimenting with blender for a couple of months when I can but used other products over the years going back to POV on a 386.
For the actual question
I’ve trying to experiment for projection mapping 3d animation onto a physical 3d surface. My own upcomming wedding cake. Searching and guides seem to be around using VPT, madmap to warp and project the content but not about how the create the content. I’m not looking for advice on modeling animation etc here but how would I go about rendering the output for the projector but from the viewers point of view.
If I had the projector in line with the viewer I could model the scene and render and see this working. However I require to project from an offset to the viewer but with the perspective relative to the ideal viewing location. In truth I’m having difficulty seeing what I need to do and how I would achieve it with blender.
My line of thinking is I model the scene with 3d animation and morphing.
Render to a camera in a position as the target viewer.
Then use camera projection to project that output to a fixed model.
Render to camera at the position the projector will be located or multiple cameras face on to each face. (this might make it easier in VPT etc and could be re used)
I’m sorry to report I didn’t find any advise on this so would share here what I did so it may help others and perhaps receive improvement suggestions especially around blender workflow.
To generate the rendered output with a perspective correct relative to the viewer position my initial idea was more or less correct.
Firstly I created a basic block out model of the physical target I would be projecting onto. Placed a Camera at the average relative position of the audience adding these to a Collection referred to as Source. I then duplicate linked this collection and renamed it to Target.
Withing the Target collection I then renamed the camera to ‘Camera Projector’. As I was planning to use a projector from each side of the target I then added two additional cameras (Out Left and Out Right). These were Placed square on to the side I will be mapping its output too. I did consider placing these cameras where I would be placing the physical projectors however soon discounted this as camera lens settings and alignment would be difficult.
Generic scene lighting was added to its own collection.
The source collection was then duplicated again, its camera removed and named ‘Scene’. The ‘Scene’ was then deformed and additional elements added to build the required animation. This scene was then rendered to the original active camera (Selecting it and then CTRL-0).
The ‘Scene’ output was then used as input image sequence for texture mapped to ‘Camera Projector’ and the geometry within the ‘Target collection’. All other objects were hidden (using its collection icon). The Target collection was then rendered setting the active camera to each ‘Output’ camera in turn as movies.
The output left and right movies were then used as source for the projection mapping software. This gave the flexibitly to tweak the alignment within the physical enviroment. Unfortuntaly VPT can only be used for one projector. It appears you could used two PC’s and run a videoa server as source for both however I decided to get MadMapper. TBH MadMapper’s UI is vastly better than VPT.
