A Hopper-esque visit to Chinatown

More about A Hopper-esque visit to Chinatown

Rebooting an old image with ChatGPT
The base image was created about 5-6 years ago with an AI art program called Wombo. I always liked the composition, so I decided to give it to ChatGPT to see what would become of it.
Out of curiosity, I asked Google's Gemini to translate the signs and here is what it said,
"In this image, the signs actually cannot be translated because the text is AI-generated pseudo-script.
While the image beautifully captures a moody, Edward Hopper-esque atmosphere, the characters on the signs are abstract brushstrokes designed to look like East Asian calligraphy (such as Chinese Hanzi or Japanese Kanji/Hiragana) without actually forming real, legible words.
Here is a breakdown of what is happening structurally on those signs:
The Main Vertical Glowing Sign: The three large calligraphic shapes mimic real characters (the middle one heavily resembles the character 休, meaning rest), but the strokes are distorted and blended into fictional symbols.
The Top Horizontal Sign: This sign uses stylized, chaotic brushstrokes that mimic a restaurant or shop name but dissolve into abstract lines upon closer inspection.
The Lit Vertical Board (Middle): This features a mix of mock seal-scripts, vague number-like shapes (like an "8"), and illegible text columns meant to simulate a menu or notice.
The Small Hanging Sign Under the Awning: This resembles a mix of distorted cursive English letters and wavy lines rather than any coherent alphabet.
It's a classic case of AI prioritizing visual texture and mood over linguistic accuracy—it perfectly sells the vibe of a neon-lit nighttime storefront, even if the signs themselves are beautifully painted gibberish!"