The old version was a blind box. You’d type a prompt and pray. The new update lets you upload a Reference Image alongside your text prompt.
The biggest complaint flooding Reddit and X (Twitter) right now is the aggressive safety filtering. Because the Beta tests new models, Adobe has clamped down hard on C2PA content credentials. If you try to expand an image of a human hand, the filter might trigger "Unsafe content" because it mistakes a knuckle for skin violation. It is frustrating and currently over-tuned. adobe photoshop beta ai upd
AI-powered contextual scene construction & smart asset blending The old version was a blind box