I was very intrigued to find out where that Newcastle xG came off, just to find out that over half of that came from this cross of Trippier. Total: 1.3xG (0.7 0.6) The cross, as you see, was too far and too high for both players to control. What a disgrace! #avfc #utv

My guess is that MidJourney has been doing a massive-scale reinforcement learning from human feedback ("RLHF") - possibly the largest ever for text-to-image. When human users choose to upscale an image, it's because they prefer it over the alternatives. It'd be a huge waste not to use this as a reward signal - cheap to collect, and *exactly* aligned with what your user base wants. The more users you have, the better RLHF you can do. And then the more users you gain.