Depending on how much full-on contact the armor will be taking, one possibility (purely a concept I've had floating around in my head, nothing I've actually tested yet) would be to build the armor out of EVA foam,as many here do, paying extra attention to reinforcing the seams on every piece. Then, coat the whole piece, inside and out, with white or light colored flex-seal, plastidip, or any other flexible rubber coating to seal it completely, further reinforcing the piece. Add any color, lines, or other such "painted on" details using more plastidip (which comes in a variety of colors and I imagine those could also be mixed to create different colors and/or shades). Seal over the whole thing again with a coat or two of clear rubber coating (I believe both Plastidip and Rustoleum brands offer clear rubber coatings in semi-transparent, clear mat, and clear gloss). This final layer should help protect the underlying color from minor nicks or cuts, and using rubber coat/plastidip as the base coat, top coat, and clear coat theoretically will not only offer flexible reinforcing (and no cracking, chipping, or flaking as some regular paints are known to do), but will also be more resistant to light scratches and scuffs than typical paint. The downside is 3+ layers (at the very least) of rubber coating will add a bit of bulk to the piece, so it might be necessary to make the parts slightly smaller from the start to compensate for the added mass of rubber. That also means small details can get tricky to retain, short of trying to build them up later by framing a template on the piece and using that to "mold" the detail on with the plastidip (obviously this will be easier to do with the non-aerosol variety). The clear coat, if applied in thin coats and slowly building up coverage rather than laying it on thick, should still allow the details to show through without obscuring them too much. Very, very light buffing might improve transparency, but that's definitely something to test first on a scrap piece.
As I said, this entire process is all theoretical concept. I haven't tested any of it, and I haven't seen anyone else mention using such a method. I don't know if that's a matter of this being something no one else thought of or if it's already been attempted and dismissed as not even worth mentioning. If someone does try this, I would love feedback on how it turned out, any obstacles that needed to be worked around or methods that needed to be tweaked to get the best results. If someone else saves me hours of trial and error to get it done right, I'd say that's well worth the trade off of sharing the "trade secrets" of a potentially original and unheard of concept (although I'm sure someone else has thought along these lines at least once or twice before me).