All the major AI players seem to be building their release schedule around OpenAI – waiting for the behemoth to announce something and then Midjourney or Google will counteract it with their own developments. A little more than a day after OpenAI released GPT-4, for example, Midjourney unveils v5 of their service. Seems convenient. And it leads me to believe that these teams have the next few iterations already complete.
WTF? Midjourney Gets Too Real
On Wednesday, Midjourney announced version 5 of its commercial AI image-synthesis service, which can produce photorealistic images at a quality level that some AI art fans are calling creepy and "too perfect."
"MJ v5 currently feels to me like finally getting glasses after ignoring bad eyesight for a little bit too long," said Julie Wieland, a graphic designer who often shares her Midjourney creations on Twitter. "Suddenly you see everything in 4k, it feels weirdly overwhelming but also amazing." – Ars Technica
Perhaps the biggest news about this Midjourney upgrade is that it can properly generate hands (something that even human artists tend to struggle with).
Over the past year, the idea that AI art generators can't render hands correctly has become something of a cultural trope. Notably, Midjourney v5 can generate realistic human hands fairly well. "Hands are correct most of the time, with 5 fingers instead of 7-10 on one hand," said Wieland. – Ars Technica
Honestly, I’ve never had to learn software as quickly as I have this past year with Generative AI. I just got around to really understanding how to operate in Midjourney v4, and now a lot of that knowledge may be chicken scratch. Even the comprehensive Midjourney Style Guide I shared in Everydays 181 is obsolete until its creator updates it with the new changes. The differences between these models tend to be so great that you’re better off forgetting the tricks that worked in previous versions, and develop a new baseline with the newest version.
It’s a new way of pushing user experience updates. Adobe Photoshop always gave us at least a year with new software until they made major changes. Google has always rolled out new features without disrupting the core functionality – for instance, there are many features most of us haven’t gotten around to adopting like Google Lens, but ultimately that doesn’t ruin typical Google Search.
Generative AI is breaking and building the human-AI interaction and user experience constantly, which frankly is hard to keep up with even for a quick-learning techie like me.
But OpenAI, Midjourney, Google, Microsoft, etc. aren’t going to slow down for us to keep up with the changes. We must forge ahead and be as agile as possible in learning AI.
With that being said, below you’ll find my first experiment working with Midjourney v5. I’m impressed with the results. Kind of reminds me of a PS5 game.