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.
(I like the software idea, as aging is something that architects would appear to overlook in any buildings built in my lifetime).
Not exactly an example of aging well, but nonetheless deserving of a honorable mention: all the old Flagler Hotels that now form Flagler College in St Augustine. They were pretty much making up the modern use of poured concrete for houses as they went and it does show in many places. Most of the concrete surfaces are looking rather rotten by themselves (despite what i suspect to be quite intensive maintenance), but the crazy playfulness of all the decorations that surround them do a surprisingly good job at distracting from that. You notice, but you don't care.
If I had a few man-decades of developer time to burn, without any chance of ever seeing adequate return, my dream software project would be a tool that analyzes architectural model data, simulates aging and creates an "aging forecast" version ready for rendering. This could help architects and clients to avoid badly aging designs if they choose to factor that into their decisions. Or at least it could generally promote the idea that sustainable is not just about day to day energy use but also about long term attractivity .