Summary

The author shares a step-by-step guide to adding more realism, personality, and life to Midjourney AI-generated portraits.

Abstract

In this article, the author builds upon their previous work on creating life-like facial generation in Midjourney, aiming to push the boundaries even further. The author presents a new technique that allows for tighter control over the output while keeping the face nearly identical to the original. The step-by-step guide involves adding specific suffixes to the prompt and using the VARY (REGION) section to trace around the eyes, nose, and lips. The author demonstrates the effectiveness of this technique through a series of generated images and animations.

Opinions

The author believes that their new technique provides more control over the output and keeps the face nearly identical to the original.
The author suggests that Midjourney can create life-like images, but there is a slight hint of flatness in the original images.
The author emphasizes the importance of tracing precisely around the eyes, nose, and lips in a tight semicircle, not around the full face.
The author is confident in the effectiveness of their technique, as they have achieved nearly identical results repeatedly.
The author encourages readers to use this technique to create even more realistic imagery.
The author plans to continue pushing the boundaries and exploring Midjourney.com for work that may have employed this technique.
The author invites readers to share their own discoveries and learn more through their Udemy course.

How to Add Realism, Personality, and Life to Midjourney AI Portraits

My Step-by-Step exploration into the creation of life-like AI imagery

FIRST IMAGE has more personality, 2nd was original Midjourney image — Created by the Author in Midjourney

Recently, I wrote an article detailing ‘hacking the code’ of life-like facial generation in Midjourney. But I always strive to push myself further. The images created were certainly more lifelike, but in their naturalness, the faces had slight distortions. In the days to follow, I questioned, “could I keep the ‘look’ of the original image and just add more personality to the original faces.” The answer again? Yes I could and with even more control over the output. Take a look at the difference in this animated GIF.

Images created by the Author in Midjourney

Here’s another one…

First, If you want to read the techniques I employed in the last article you can read it here…

Creating Life-Like and Natural-Looking Midjourney Faces in a Snap

Thank you Midjourney “Vary (Region)” option, you helped me crack the code

davidwlitwin.medium.com

But this new technique gives me far tighter control over the output, and keeps the face nearly identical to the original as you can see above. The step-by-step in this story is even easier than the steps in the last one. Ready? Let’s begin.

Step-by-step guide to more AI personality

First, like my original story, I start by adding the following suffix to my prompt: “shot with Canon R5, 70–200mm, cinematic studio lighting, strong shadows, photographed by Bert Stern, 8k UHD -s 800”

That produced the following image

Image created by the Author in Midjourney

As stated in my other story, Midjourney is able to create some pretty life-like images, but there is just that slight hint of flatnesss in the image above. I used the VARY (REGION) section and AND, HERE IS THE TRICK, traced precisely around the eyes, nose, and lips in a tight semicircle (AND NOT AROUND THE FULL FACE).

Then, I added the following text at the FRONT of the prompt: “add more realistic detail, personality and light sourcing to this face::5 beautiful facial features, happy bright eyes.” That produced the following four generations.

If you read my last article you’ll see how dramatically more controlled these generated images are, compared to the techniques I employed previously. All of these could have worked, but the top right seemed to really capture the full personality.

And just like that, in minutes, I was done. Here’s an animation to see the difference.

Can it be that simple? I have done this repeatedly and have achieved nearly identical results. Therefore I believe the answer is a resounding “yes!” I hope that this technique inspires you to create even more realistic imagery.

Full image for title photo, Created by the Author in Midjourney

I will continue to push the boundaries as I look on Midjourney.com to see if I can spot any work that may have employed this technique. Please reach out to me with your own discoveries.

And if you are READY TO LEARN MORE: I’ve added dozens of other strategies in my Udemy Course. You can join this class HERE at 85% OFF the regular $84.99 cost!

Live inspired.

This story is published on Generative AI. Connect with us on LinkedIn to get the latest AI stories and insights right in your feed. Let’s shape the future of AI together!