Do role lines like "you are an expert" still help image and video prompts?
Many prompts start by assigning a role, such as an expert photographer or an award-winning director. It feels like it should help, but people disagree about whether it changes results, especially for image and video models.
The arguments usually go a few ways. Some say a role line sets a style or quality expectation the model picks up on. Others find that concrete details, like lens, lighting and composition, do far more than any title. Some models seem to ignore role framing entirely, while chat-style models may respond to it more. And a role line can waste space in prompts with tight length limits.
Have you tested prompts with and without a role line? On which model did it actually make a difference, and what did you replace it with if it did not?