Here’s an example of how much one generative AI program has changed in a year.
⏩ If you feel like technology in the last year has been moving faster than a running toddler, you are not alone.
Anyone who uses digital tools to create visuals or write content may be experiencing a daily sense of FOMO when it comes to how generative AI tools will affect their creation process.
One thing I have noticed is that a lot of generative AI programs are versioning so quickly that users have to remain agile to prompt desired outcomes. Don’t get set on one way of prompting.
Since commercially available generative AI models are black box and updated at the discretion of their providers, relying on one software/service as a sole source for your generative outputs could come with risks of features changing for better or worse.
To show you what I mean, I decided to share an experiment I did using Midjourney, a natural language to image model that runs on the Discord platform.
🧪EXPERIMENT: Midjourney V1–V5
As an AI image generator, Midjourney offers a choice of versions and features for its users to select in order to generate all kinds of varied outputs.
To test the evolution of the Midjourney software versions, I used the following prompt: “Looking up into the mind of the ocean” as the sole text input and changed just the version in the settings.
Midjourney offers five software versions to chose from, V1 — V5.
📋RESULTS: Stylized-V1 to Realistic-V5 in one year
By scrolling down through the pictures, you can see that the Midjourney model versions interpret my abstract prompt with outputs that evolve from abstract/stylized (V1) to realistic (V5).
The amazing thing is that these updates to Midjourney’s software/service happened over the course of just one year. Not all companies support previous software versions or make updates as frequently.
💭 My thoughts:
This experiment is an example of how, with iteration, the Midjourney generative AI tool has been “improving” with each version by creating more realistic outputs with a simple statement prompt.
An advantage for creators is there is so much room to be human in the abstract places that enthusiastic progress tends to leave in its wake. A push for alternative reality is an obvious focus for the progress of AI so unique humanness will become a hot creative commodity.
I plan to keep experimenting with generative creation tools, but do my diligence to protect my data, ROI and privacy.
This is just one example of the rapid pace of change from one generative AI image software/service. Here are some other image generators to try: DALL-E2, DeepArt.io, DreamStudio and Night Cafe.
I would love to hear your thoughts about the changes you are seeing interacting with generative AI tools in the past year.
OR Check out more of my experiences using Midjourney and other insights on JEFS blog page.
Julia Fletcher is founder of JEFS Storytelling Arts, a graphic design studio, where she uses her unique research skills and artistic talents to create custom visual stories that help clients’ increase engagement and promote the education of their audience.