The same prompt isn’t always a fair model comparison
Would you rather see two image models given identical wording, or let someone tune the prompt for each? The first seems useful if the question is “can I swap this into my existing setup?” The second seems more useful if the question is “what could I make with it after learning how it likes to be prompted?” A comparison that mixes those questions is hard to judge. What would you want the person posting it to show?
Shared by strxet.nyc on Prompt Builder.