The paradigm that madmapper uses to get visuals from the media bin to the real world is currently single visual > input surface/s for sampling > output surface for warping to a projection surface. But there's no upstream 2D compositing stage to arrange multiple media spatially into a 2d composition before getting onto a surface. For example:
A mapped sphere playing a visual, if you needed to place an additional image onto a portion of that sphere it'd require creating a new quad, assigning and input mapping the visual to it, but then having to essentially remap that output surface to the portion of the sphere it needs to go.
Unlike other paradigms that warp an entire output canvas, we get the granular control of each surface on the output which is great, but without an upstream composition stage we're essentially stuck with a single visual for each output surface. I think MM's origins are the reason for this because it was conceived as a final mapping stage for incoming, pre-composed media either syphoned from another source or finished files. As it's gotten much more powerful it's making less sense to be missing a stage where we could composite multiple media into a composition that can be sampled by a single surface.
One workaround is to have a dummy video ouput that goes nowhere, multiple media placed onto it and then brought back into the media bin via the video output source type, then assigned to other surfaces for mapping. This seems like very much a workaround.
Montage tracks are partially helpful but they're sequential rather than spatial. Are there any thoughts on adding the ability to compose 2D media compositions within madmapper in an initial stage?