In continuing my experiments with Midjourney AI, for fun I tested entering the tagline of Lucept: “the future of architectural lighting”

Much of the magic behind this system is the sample set it was trained on. But still, the poetry of what text phrases a user submits as the query dramatically alters the result. It creates such generative, expansive, random and imaginative ranges of imagery.

So I was curious as to what results such a simple phrase as “the future of architectural lighting” would return – the phrase could be interpreted in any number of ways. Yet it generated a fascinating 4-up sample set, and then I “rerolled” the query two more times to generate even more wild variations. Here’s the result:

I tried generating the next level of variations for one of the initial sets, with mixed results:

And finally I upscaled one of the previous images: