I'm sure that it would take me very long, not being an expert in any of the required skill areas and all that. Seen from afar i identify several key components of a high quality architectural aging simulation: water/geometry interaction (most stains seem to be related to differences in water exposure), some rough thermal simulation (different drying speed, often you see "shadows" of internal structure on the outside, because the internals influence heat dissipation and thus drying speed), gathering and refining actual aging examples for all kinds of surface materials and finally extracting not only geometry but also surface material information from the input formats you would want to support. Putting it all together to generate output models or images would be the smallest part.

A less sophisticated effort might just try to use image based rendering techniques to normalize before/after photos into flat textures ready to be used as a learning set for some ML magic. Unfortunately, this would completely fail at respecting different aging properties of surfaces that look similar when new. Cleverly alternating between initially similar surfaces could for example be harnessed to get a clean, minimalistic look when new and expose some ornaments that distract from unintended unevenness later.

This isn't a completely theoretical idea by the way: a ca 2005 subway station here that used to get heavily criticised for its bland white walls is slowly starting to expose fascinating watermark-like images embedded in the texture of the walls, consisting of cleverly modulated ribs that seemed perfectly uniform when new.