Futurepedia · AI

Testing a ridiculously complex prompt in GPT Image 2.5

Testing a ridiculously complex prompt: a U.S. map with every national park pinned, plus arrows pointing to retro posters. The model nails the concept, but park locations and arrow alignment? Not so much. GPT Image 2.5 pulled off the scope, which is honestly wild. The retro posters actually look accurate. But a lot of the pins aren't on the right states, and half the arrows point nowhere near their park. Impressive it even tried, but precision is

Introductie van de bron.