At Ocean CGI we’re always looking for ways to streamline workflows, allowing more time to be creative. Although no matter how efficient systems and processes become, the product is ultimately created by an artist. For now.
AI has developed rapidly in recent years and its progress in understanding what makes an ‘image’ is startling. Using Midjourney AI and inputting only a few text prompts I generated these visually striking concepts, using algorithms and technologies I won’t pretend to understand.
The images always have an impressionist style and don’t really make sense when viewed close up, but as a way to develop concepts, the overall effect is impressive. Prompts can be detailed; I described the bamboo forest house with around 140 words. Whereas “LS Lowry painting of Barnstaple, Devon” or “Frank Lloyd Wright vs Anthony Gaudi Building at Sunset” yielded just as impressive a result.
From pressing enter, it takes around 1 minute to produce four images. You then have the option of upscaling or creating four more variations of each image. The AI even learns from this, as millions of users selecting the best results teaches the AI what works and what doesn’t.
The current application leans toward concept artists, but architecture students and product designers to name a few will be utilising this and other AI systems more and more. Arch-Viz might just be safe. For a while.