The direction and training of AI are essential for achieving satisfactory results. These were some of the first images generated, using prompts that were simple and direct, without much detail. The intention was to see how Midjourney would interpret straightforward instructions and what kind of outputs it would create from basic, unrefined prompts, with no specific guidance or references.
What was particularly interesting and challenging was that Midjourney didn’t have sufficient information about the Apple Vision Pro, which posed a significant challenge in helping the AI understand this new Apple product. Since the Apple Vision Pro had only been recently launched, Midjourney lacked a reference database for it. To overcome this issue, I began training the AI by feeding it reference images to match the product I was requesting. This greatly sped up the process, as instead of relying on trial and error, I effectively taught Midjourney what I wanted it to generate by using visual references.