It is! It's not exactly new or rare though. It's the simplified version of the "Guidance scale" or "Image strength" in Stable Diffusion when used with starting images. And Midjourney has "Image weight" that also determines how much influence the image prompt has on the final result.
All the LLMs seem to have some kind of "slider" like this, where you can pick how accurate vs creative it is. This seems like a very sensible evolution to me, one where people are immediately aware that LLMs can lie (the trade off between creativity and verity should make this clear).
"Reference strength" is such an important, simple concept. I love that solution.
It is! It's not exactly new or rare though. It's the simplified version of the "Guidance scale" or "Image strength" in Stable Diffusion when used with starting images. And Midjourney has "Image weight" that also determines how much influence the image prompt has on the final result.
All the LLMs seem to have some kind of "slider" like this, where you can pick how accurate vs creative it is. This seems like a very sensible evolution to me, one where people are immediately aware that LLMs can lie (the trade off between creativity and verity should make this clear).