An architect by trade, Harshdeep Arora started experimenting with generative AI as soon as it became available. He envisions that AI-generated art will redefine concepts of authorship and unlock new depths of artistic expression. Through well-trained AI, intricate designs with surprisingly detailed scenes are now being created, showcasing new dimensions of creativity.
The art industry stands at the cusp of a potentially pivotal juncture, as the advent of AI technology
holds the promise of fundamentally revolutionising the way it currently operates.
Training AI for artistic purposes requires employing advanced deep-learning techniques alongside extensive datasets of artistic content. Through training, models are taught to identify patterns, styles, and visual components from established artworks. By amalgamating these acquired features, AI can create art, often using generative adversarial networks (GANs) or similar generative models. An average user may not need to train their own personal model due to the extensive GPU processing required. Leveraging and combining different pre-existing models can achieve excellent results.
Often drawing inspiration from European art, especially the Baroque and Art Nouveau, Arora works with established models on Stable Diffusion and Midjourney, integrating LORAs as needed to achieve specific aesthetic goals in various projects. To foster uniqueness, the artist introduces elements of randomness into the training data, particularly during the initial testing phase.
“Cotton Candy Express | A Pink Paradise Journey through the Japanese Alps” series, one of many in Arora’s portfolio, combines all his favourite subjects: architecture, nature, abstract concepts, and futurism.
