Text-only prompting habits don't transfer cleanly to image and voice inputs. Here's a Q&A breakdown of what changes when you're working across modalities, and why.