AI Generated human portraits are a simply incredible, all the early challenges such as hands and teeth have been addressed in Midjourney models since version 4 was releases. In this post I wanted to focus on creating various human expressions using the Midjourney model v5.1.
Here is a list of expressions that we will experiment with to understand if they work well or not.
- Raising eyebrows
- Rolling eyes
- Looking surprised
- Looking angry
- Looking sad
- Looking happy
- Looking disgusted
- Looking afraid
- Looking confused
- Looking ashamed
As this is such a long list of expressions creating each image would mean creating 23 individual prompts and submitting them one by one to Midjourney. Luckily this is easily handled by Midjourney by using permutations feature which you can read more about here.
Permutations allows you to create one prompt containing all the variations you want and it will automatically work out the resulting prompts for you.
Our base prompt will be: a 21yo woman with a <expression type> expression on her face –v 5.1
Below is the series of images that I found Midjourney correctly represented in the images it produced. In most cases 4 of the 4 images represent the expression, in some cases only 3 of the 4 in the grid of images generated represent the expression correctly. Either way I guess I take them
In the below images you can see the expression is not really represented at all or done correctly with Midjourney. It is disappointing that some of these easy to create expressions are not represented correctly but hopefully as the AI models improve these results will as well.
I got this extensive list put together with the help of ChatGPT, but if you think there are other expressions that have missed please let me know in the comments below and I can happily update the post to include audience submissions. Make sure you follow me on Twitter to stay up to date with the latest from the blog and my AI experiments.