With its deep symbolism, swirling colours and captivating imagery, Van Gogh’s “The Starry Night” exemplifies that two-dimensional art remains a human specialty incomparable with AI amateurism.
After an emotional outburst when he cut off a piece of his left ear, Van Gogh voluntarily admitted himself in May, 1889, to the Saint-Paul-de-Mausole asylum.
It was there that he created some of his most famous works, including The Starry Night in June, 1889.
Van Gogh had also interpreted death as another existential dimension, and he metaphorically linked this other dimension with the stars. The cypress tree in the foreground is also traditionally associated with death and mourning. But why would he include these two reminders of mortality in an otherwise serene image?
Van Gogh had become increasingly isolated, feeling that he had failed in life and his career; even he thought that Starry Night was a failure. During the time he shared with this painting, Van Gogh saw no hope in his future.
Despite the troubling times, Van Gogh still included in his painting a tranquil village beneath a turbulent night sky. One can understand now that perhaps Van Gogh recognised his emotional hardship and still attempted to find hope and motivation in the natural beauty of the world.
Upon a closer look, each individual brushstroke comprising The Starry Night can still be perceived, adding layer upon layer of texture and movement. On the other hand, Stability AI’s artwork, in trying to mimic Starry Night, depicts the sky with similar rings of hues which surround a bright yellow centre, but instead of showing nocturnal stars, the AI picture presents us with a daytime scene. Other similarities in both the AI-generated image and the real thing include the gentle blue colour of the distant background mountains and the flame-like trees.
To commend the AI’s successes, it is very capable of capturing our initial curiosity with its contrasting themes.
Witness a serene village and the lively hills; the dim, night-time aura and the rising sunlight, the melancholic green and the calming blue, the dead tree upfront and the blossoming flowers at its feet.
However, the painting in general pales in comparison to Van Gogh, with colours literally pale and washed out, lacking the vibrancy, motion and perspective of the Van Gogh original.
The AI also attempts a lackluster interpretation of the nuances of painting, shown in its unnaturally shaped brushstrokes on the trees and fields, and a lack of small details in light and shadow, giving the image an airbrushed and flat appearance.
On top of that, Stability AI had trouble imitating Van Gogh’s use of light, shading and lines. Two examples are the lack of shadows which jumble the rolling hills together into one inseparable mass, along with the rigid strokes of colour in the sky which are incongruous with the fields’ wavy strokes.
There are also some telltale indications of the image’s AI origin. Stability AI cannot be specific in what it generates, because its source of millions of images will undoubtedly include pictures from various categories and techniques.
This causes the product, which is supposed to be focused on Van Gogh, to be a massive, ineffective amalgamation of other artists, genres and styles.
Ultimately, paintings are the culmination of a series of thoughts and emotions portrayed on a contained canvas, an occasion to which AI cannot rise.
Unless AI evolves to have its own consciousness, its art will remain unimaginative, uninspiring and confusing. While a painting’s complexity is perfect for the tangled-up emotions and stories of Van Gogh, its nuances are still just too difficult for a machine to manage.
