First Impressions: DALL-E 3 ChatGPT Integration

Today was a rainy Sunday, and I found myself with an exclusive opportunity - early access to DALL-E 3 integration into ChatGPT. The timing was impeccable. I was in the midst of crafting an entry for a contest hosted by Wirestock.io.

The DALL-E 3 ChatGPT integration promised a seamless bridge from text to visual narrative. To sculpt a detailed and coherent prompt, I employed two powerful tools: custom instructions and Professor Synapse. The former allowed for a precise directive to the AI, while the latter ensured a methodical, step-by-step approach to crafting the narrative.

The result was a haunting scenario: "A moonlit graveyard scene where Jason Voorhees is unearthing a grave, and Freddy Krueger, with his razor-sharp fingers, draws eerie symbols in the air. Tombstones bear the names of past victims, and shadowy spirits rise from the ground. The atmosphere is chilling and foreboding."

With a click, the prompt was in DALL-E 3’s court. The outcome was a blend of eerie and unexpected. The image showcased Jason Voorhees towering over Freddy Krueger, who oddly donned Freddy's iconic glove. The tombstones bore their names, a deviation from the envisioned narrative, yet the sinister ambiance remained.

This was a clear showcase of DALL-E 3’s limitations. It simply didn't understand the assignment. With a disappointing output and the additional limitations of a fixed 1:1 aspect ratio, the one-shot output, and no option to redraw certain parts of the image made it incompatible with my artistic needs.

Unlike DALL-E 3, MidJourney allows for more creative flexibility. This stint with DALL-E 3 was enlightening. But I currently need a creative playground, like MidJourney.

The Wirestock contest was just the right excuse to delve into DALL-E 3’s capabilities. It was a fun ride, even if the image didn’t quite match the initial prompt. I was honored to have gotten early access and get a clear peek into the AI future.