Monday 15 January 2024

A Tale of Two Worlds - AI-Generated Imagery vs A Photograph



I thought of using the concept of a recent photo shoot (Kat & The Harlequin) to generate a MidJourney image using these prompts:

[editorial fashion photography, elevated angle, beautiful Hispanic woman holding a red and white Venetian mask, wearing a Goth outfit, standing in a dark NYC alley, serene yet haunting. --ar 128:85 --style raw --stylize 20 --v 6]

In so doing, I reversed the order I sometimes use by generating AI images initially for some ideas for forthcoming photo shoots. 

Then, to go full bore artificial intelligence, I asked ChatGPT to generate a blog post based on that idea. Here it is with some edits of my own:

"The convergence of traditional photography and artificial intelligence has given rise to intriguing possibilities. This blog post delves into the comparison between an AI-generated image based on a photo shoot featuring Kat, as a dark-haired beauty holding a Harlequin Carnaval mask in a New York City alley, and attempting to capture the essence of this scene.

The actual photo shoot was in Cortland Alley, during which Kat was the muse, holding a Harlequin Carnaval mask. The grit of the alley was the backdrop, creating a sort of visual discordance between it and the sophistication of the mask.

Solely using my text prompts, the AI-generated image did replicate the style of the photograph using its algorithms. Though the resulting AI image may show technical proficiency, it obviously lacks the human connection I experienced with Kat whilst photographing her, and the raw authenticity of a real moment.

During the photo shoot, my experience, my "eye", intuition, and ability to interpret the scene, as well as Kat's presence, contributed to the result. The AI-generated image could not really capture the intangible aspects that made the real photograph resonate."

Call Me KIJU

Here are impromptu street portraits of Kiju on Crosby Street in Soho, NYC. Kiju is an alternative rock performer.