top of page
Writer's pictureGerard Kunkel

Generative AI then and now

NMP Partner Patrick Donoghue introduced me to Midjourney during its limited beta release in early 2022. It was exhilarating to experiment with this new GenAI tool, converting ideas expressed in words into images—though the early results were often impressionistic blurs with distorted shapes and figures. However, as with all new software, it’s important to consider the potential for advancement if the underlying idea has value in the marketplace.

Midjourney officially launched in February 2022 with Version 1 to a private group of beta users. By April 2022, Version 2 was released, which is when we first began to integrate generative AI into our early iterative design process. We experimented with its capabilities to understand its limitations and explore how it might be applied to our work.


At that time, we used Midjourney as a tool to envision a client solution we were developing. While the technology wasn’t capable of producing final-quality images, it was effective for establishing a vision. This type of directional artwork was useful for discussing the general ideas we would later refine into high-fidelity final products.


That was then.


Today's GenAI tools take us further than iterative ideation. Looking back just over two years, it’s remarkable to see how far Midjourney and its competitors (such as DALL-E, Stable Diffusion, Runway ML, and Playform.io) have advanced in such a short time.


To illustrate this rapid progress, I revisited my original text-to-image prompt from April 2022 (Version 2) and compared it with what is generated today in September 2024 (Version 6.1).


The prompt in both cases was: “imagine a warm and happy visit to the cemetery with family, a visit with grandpa and grandma photorealistic.”


This prompt was part of our ideation process for a client looking to build a new AI-powered experience for remembering the deceased. At the time, the technology was just beginning to emerge that could make the client’s vision possible—though it was still a few years away. Now, with the advancement of generative AI and related graphics presentation technologies, the original product ideas are entirely feasible with thoughtful and careful integration.


Generative AI image comparison using the same text prompt: “imagine a warm and happy visit to the cemetery with family, a visit with grandpa and grandma photorealistic.”


Generative AI image created with Midjourney version 2 in April 2022


Generative AI image created with Midjourney version 6.1 in September 2024 using the same prompt used in 2022.


History of Midjourney version releases

  • Midjourney V1: February 2022

  • Midjourney V2: April 12, 2022

  • Midjourney V3: July 25 2022

  • Midjourney V4: November 2022

  • Midjourney V5: March 2023

  • Midjourney V6: December 2023

  • Midjourney V6.1: July 2024

Comments


bottom of page