Have a look at this: https://www.engadget.com/2018/12/03/nvidia-ai-video-to-video-synthesis/?guccounter=1 This is the future of content generation. It's a city, generated by an ML algorithm using videos of other cities. It's a bit rough, but already impressive. A few glitches are still visible in "the matrix".. ..speaking of which: it's not 20 years ago (1999!) that "The Matrix" was released. Back then it was very distant science fiction - almost unthinkable. Of course people dreamt before of cyber spaces (Neuromancer, Snow Crash, Otherland, just to name a few), but none used the indistinguishability of the artificial environment from reality as the rationale for the story. Almost every book or movie before depicted virtual reality as something innately different from reality (think Tron for example). This is now, 2018. It's not the best, but it's almost there. It is much more realistic than all "realistic" looking games out there (it's also more ugly, but this is reality). If this can be trained with a video, why not train it with movies? That means: why not use the artificially altered settings and lighting used in movies to train it, to produce a nicer, more beautiful version of reality? Think about the possibilities. I am not saying that UnrealEngine is better than StarEngine, or anything. It doesn't matter, they could have used Lumberyard just as well. I also don't expect this technology to appear in Star Citizen anytime soon. But: I do think that this is the perfect match. And it would offer almost unlimited new possibilities: entire planets generated algorithmically with a few perfectly rendered scenes as reference, saving weeks of work and huge amounts of storage space. With the power of AWS this just could be possible.