AI generated imagery

The above images demonstrate two stylistic options I explored using MidJourney in developing imagery to accompany an entry on the Input Output Lace wallet blog. The challenge was to create images that represented concepts from the text they would accompany, while also hewing to brand and accommodating the addition of Lace's established 'portal' graphic, which was composited traditionally — in Photoshop.

The images above were generated during the discovery phase of the Wildebeest identity project, which has its own dedicated page in this portfolio. Generating sets of related images has been a recurring theme in my experience with AI. The skill, in this case, is finding ways to keep the generator on track while exploring options related to material, lighting, and context.

The images in the next section fall into two sets containing several variants each. These were created in order to generate different style profile pictures from single portraits for a Lace wallet promotional campaign. One interesting point on this project was that the source portraits themselves were AI-generated 'blends' of stock photos. This taught me that making sets of related images actually works much better (at least in MidJourney) when starting from AI generated images rather than starting from photos.

Perhaps the best thing about creating images through AI is the quality of results it allows you to tap into. Not necessarily without effort, but quickly and with much less trial and error than would have traditionally been possible. The superhero image shown here was born out of a need to create an image based on a screen grab from a video. The original request was to simply clean up the screen capture, however once stakeholders saw what a wholesale re-imagination could mean, they were sold.