As 2023 comes to a close, we find ourselves in a world profoundly changed by the tools that we have available. This year has been a wild ride, but the changes we’ve seen pale compared to what we can expect in 2024.
The image that illustrates this post is one of the first I created using Midjourney Version 6, released to the public in Alpha status. It took me perhaps all of 5 minutes to think of the prompt and zero in on the images that I wanted as I went through two additional generation cycles. The quality of the image is spectacular. Its faithfulness to what was asked in the prompt is 95%.
I’ll include the prompt below for you to check out and play around with it.
Detailed children’s book illustration style Christmas Greeting Card image showing a Christmas tree near a lit fireplace and Santa Claus placing presents from a large bag, under the Christmas tree --ar 3:2
As you can see from the prompt, the only thing missing is the bag of presents, and that was actually a choice I made, as I liked this picture better than the ones that included the bag. I’ve added some of the other options I generated in these few minutes just to give you an idea of how all-around good these are. These are variations of the one above, and they all look superb, though I prefer the one above.
We can expect that tools such as Midjourney, Leonardo, Pikalabs, Elevenlabs, LOVO and ChatGPT will continue to improve at an accelerated pace. As developers use AI-powered tools to help in the development of these products, you can expect their evolution to happen at a frantic pace.
When Midjourney was first released, in February 2022, the simple prompt “a red car —ar 3:2” would yield these results:
When version 4 came out, in November 2022, the results for this prompt were incredibly evolved, as you can see below.
If you put this very same prompt to version 6, which we should remember is still in alpha state, you get a level of realism that is mind-boggling.
We can only imagine that when the full feature set of Midjourney becomes available on version 6 of the model, we will be able to Zoom Out and use the Vary Region commands to get a more complete view of the car or correct small imperfections as we can currently do when using version 5.2.
One thing that versions 4, 5, and 6 seem to have in common, though, is their bias towards vintage cars.
I hope you will accompany me in 2024 as I explore these tools and how we can make use of them to create ever more sophisticated art and content.
Happy holidays!