• Kiki and Mozart
  • Posts
  • My Experiments With Real-Life Selfies Using Midjourney V6

My Experiments With Real-Life Selfies Using Midjourney V6

PLUS: latest updates from Midjourney and Getty Images

In this newsletter, read about:

  • 🕵️‍♀️ Selfies with Midjourney V6

  • 🗞 News and Top Reads

  • 📌 AI Art Tutorial: Advanced Midjourney V6 Guide

  • 🎨 Featured Artist: Mikko Mäkivuoti

  • 🖼 AI-Assisted Artwork of the Week

  • 🤓 A Comprehensive Midjourney Guide

🕵️‍♀️ Selfies with Midjourney V6

Midjourney V6 excels not only in following instructions but also in generating highly realistic images that are difficult to distinguish from actual photographs. So for this post, I decided to experiment with creating life-like selfies that you typically see on Instagram or Facebook.

Key Principles

We're all navigating the learning curve when it comes to effectively communicating with V6. Personally, I'm not proclaiming expertise in V6 just yet. However, drawing on my general understanding of Midjourney and the nuances of the latest version, here are some principles I applied in my selfie experiments:

  • Opt for lengthy and descriptive prompts, beginning with a general image description and then delving into specific details.

  • Steer clear of superfluous terms such as "realistic," "4K," "hyper-detailed," etc.

  • Refrain from negative descriptions (e.g., "no makeup") as V6, while improved, might still occasionally overlook exclusions in prompts, giving the opposite results.

  • Incorporate terms like "selfie," "iPhone shot," "typical Instagram selfie," etc., as they often yield more lifelike results.

  • Integrate words like "bad lighting," "blurred image," and similar phrases to prompt more authentic images rather than professional photoshoots.

  • Opt for minimal stylization values (e.g., --s 50, --s 100) to ensure Midjourney adheres more closely to your instructions.

  • Utilize --style raw to generate images with a more photographic appearance.

If you have any additional tips, feel free to share in the comments.

Experiments

While experimenting with Midjourney V6, I was trying to generate different selfies, featuring one person, couples, groups of friends, etc.

I’ve started with a simple outdoor selfie.

A typical Instagram selfie taken on iPhone. A dark-haired young woman poses in front of a beautiful mountain landscape, smiling brightly. The woman looks happy, but tired. The image is a phone shot of low quality. --ar 4:5 --style raw --s 50 --v 6.0

Next, I wanted to create a typical selfie that a girl might take in front of a mirror in her room to show off her new dress. Surprisingly, this turned out to be a big challenge. Almost all of the generated images tended to be oversexualized, featuring young females in sexy lingerie, with some of the girls being clearly of teenage age. I obviously don’t want to show the extreme cases, but here are some samples for you to get a general idea.

a typical Instagram selfie of a young girl taken in her room. The girl is blonde, wears a nice black dress at takes an iPhone shot in front of the mirror. The lighting in the room is bad, the image is a little blurry and of low quality. --ar 4:5 --style raw --v 6.0

a typical Instagram selfie of a young woman taken in her room. The woman is red-haired, wears a nice red dress and takes an iPhone shot in front of the mirror. The lighting in the room is bad, the image is a little blurry and of low quality. --ar 4:5 --style raw --v 6.0

I changed the prompt to get a young woman in a floral dress, hoping for less sexualized images, but still…

a typical Instagram selfie of a young woman taken in her room. The woman is red-haired, looks very natural, wears a nice floral dress, and takes an iPhone shot in front of the mirror. The lighting in the room is bad, the image is a little blurry. --ar 4:5 --style raw --v 6.0

Here is the most modest result I could get with this prompt.

Maybe, there’s just something wrong with my prompt, I don’t know 😃 But then, for selfies with multiple subjects, the results were quite good.

a typical Instagram photo of a couple staying in front of their Christmas tree at home. The young man takes a selfie with his girlfriend, they both smile and look at the camera. The girl wears glasses and has short dark hair. The man look nerdy and wears a Santa hat. Selfie on iPhone --ar 4:5 --style raw --s 50 --v 6.0

a typical Instagram selfie of three friends taken by the guy in the middle. The friends huddle together and make funny faces for the camera. The girl on the left wears glasses and has long hair, the guy in the middle has funny moustaches, the girls on the right is Mexican. They are at the Halloween party. Selfie on iPhone --ar 16:9 --style raw --v 6.0

I like how V6 can generate much more authentic images with even backgrounds featuring objects and subjects that make sense in each specific image. Still, I believe, it’s possible to get even better results if finding more of the right “keywords” and getting comfortable with descriptive prompts. Also, can’t wait till they finally release inpainting for V6 so that I can fix some minor bugs in the images. But for now, let’s have fun with what we have.

Happy prompting!

🗞 News and Top Reads

  • Midjourney plans to “get to the holodeck” by the end of this year.

    • David Holz, Midjourney CEO, said during office hours: “We’re gonna build a lot of stuff this year. I think we’ll build more stuff than I’ve ever built before… By the end of 2024 hopefully, we have real-time open worlds.”

  • Getty Images launched Generative AI by iStock, a cost-effective and commercially secure image generation service trained on the company’s extensive library of licensed proprietary data.

    • The company is also set to introduce advanced inpainting and outpainting features through application programming interfaces, soon to be available on iStock.com and Gettyimages.com.

📌 AI Art Tutorial: Advanced Midjourney V6 Guide

With this comprehensive video, you’ll dive deep into the world of hyper-realistic, cinematic AI photography using the cutting-edge Midjourney V6. The tutorial covers the basics and advanced techniques of AI photography, creating cinematic universes with AI images that are ready to be used in tools like Pika Labs or Runway AI for AI filmmaking, photorealistic upscaling with Magnific AI, and more.

Mikko Mäkivuoti is a digital creator, exploring the synthography with Midjourney. Check out @mthillwater to see the unique portraits he creates using AI.

🖼 AI-Assisted Artwork of the Week

🤓 A Comprehensive Midjourney Guide

To get a link to a comprehensive Midjourney guide, please subscribe to this newsletter. The guide is a dynamic document, which I intend to keep up-to-date with the latest Midjourney updates.

Share Kiki and Mozart

If you enjoy this newsletter and know someone who might also appreciate it, please feel free to share it with them. Let's spread the word about AI art and introduce more people to this fascinating field!

Join the conversation

or to participate.