When it comes to prompting AI image generation models like Midjourney, we often think of it as a directed approach. We have a specific idea in mind and craft our prompts to guide the model towards realizing that vision. But what if we took a step back and approached it from a different angle? What if, instead of trying to control the output, we embraced the unknown and let the model surprise us?
This is the concept of open prompting, where we craft deliberately vague or abstract prompts to explore the latent space landscape and discover unexpected images that emerge. It’s like archaeology or mining for images, uncovering novel and unseen visuals that no one else has ever imagined.
I like to think of the latent space as a superposition, where all possible images exist in potential. In this analogy, prompting is like observing a location in the latent space, causing all those possibilities to collapse into a single image. Open prompting feels like a rich avenue for creativity, shifting the artistic influence away from controlled creation to a curated and discerned collection of artifacts that align with our aesthetic vision and artistic merit.
The possibilities are endless, and it’s exciting to think about the amazing images buried deep within the model that have never been, and may never be seen by human eyes. So, how do you approach image prompting, particularly with Midjourney? Do you prefer the directed approach or the open-ended exploration of the latent space?
Take a look at the set of images I uncovered from the latent space using similar vague and open prompts. They’re all so different and interesting, and it’s amazing to think about the possibilities that lie within.