AI Returns the Thrill to Painting
...... by Bill Hudson
Last month we discussed AI as both a threat and a tool for artists. It is now possible for anyone (even those with zero art talent) to generate an incredible image with a unique “prompt” and the luck of timing. Due to the randomness and complexity of the AI image generating process, no two images will ever be the same.
For me, the thrill of painting is directly proportional to the quality of my reference. If, for example, I have my own photograph of a small harbor, with old fishing vessels at the wharfs near wooden fish shacks, on a misty morning…well, I can’t wait to get to my studio. I love every moment painting that special scene while reflecting on childhood summers on the Chincoteague Bay, or early adult years with my small crabbing operation in San Pedro, California.
So, my last two paintings have been pure joy because of AI, or more specifically, Midjourney. With simple “prompts” and a few iterations, I’ve generated reference images far more inspiring than any of my thousands of photographs taken over many years on many trips. With Midjourney, I sit at my desk and take my camera back in time to any imagined location. Then I specify essential elements including the time of day, season, sky conditions, weather, birds in the sky, etc. Believe me, it’s exciting to generate a near-perfect reference composition and begin painting.
However, a few words of caution. AI is not perfect. In fact, it generates many obvious errors that require recognition, correction, or deletion. To illustrate…I love bluegrass music and playing a 5-string banjo. But whenever I prompt an image of five bluegrass musicians playing in the woods near a stream…the banjo player is usually playing a 4-string plectrum banjo; the mandolin has 4 strings rather than 8; and some of the players are standing on top or in the stream with parts of multiple instruments hanging around their necks.
Or, when I request a west-coast harbor scene from the early 1900s with classic Monterey Clipper fishing vessels, I may get boats that lack the double-ended hull with the graceful bow common to sailboats known as feluccas.
But having to make some technical adjustments seems a small price to pay for the benefits and excitement of AI image generation.
|