How to Get More Reliable, Coherent, and Beautiful Results in Midjourney

In this newsletter, read about:

• 🕵️‍♀️ Improving the Quality of Your Midjourney Images

• 🗞 News and Top Reads

• 📌 AI Art Tutorial: 5 Ways to Avoid Symmetry in Midjourney

• 🎨 Featured Artist: Sougwen Chung

• 🖼 AI-Assisted Artwork of the Week

• 🤓 A Comprehensive Midjourney Guide

🕵️‍♀️ Improving the Quality of Your Midjourney Images

Today, I want to share a few tips on getting more coherent, original, and aesthetically pleasing results in Midjourney.

We’ll start with the following prompt.

The prompt is short and simple, without unnecessary noise, and the results are already quite good. But let’s see if we can get even more interesting images using a few tricks.

Invoke Archetypes

First of all, it’s a very useful trick to leverage archetypes instead of writing detailed descriptions in prompts. Specifically, in our case, we can replace “woman wearing a crown” with one word “queen”: 

With this small twist, we were able to get more diverse and captivating images. There are several reasons why this trick works for Midjourney prompts:

• It reduces the number of words, and thus Midjourney will have more space in its memory to consider additional details that you specify in your prompt.

• It invokes specific coherent images that people (and now AI algorithms) have in their memory, like for example, “cat lady”, “mad scientist”, “schoolteacher”, or “gamer.”

• Short descriptions based on archetypes are more usual for image titles and alt text and thus are more familiar to Midjourney’s AI models trained on such data. For example, it’s much more likely for the image to be titled as “a cyclist” than as “a person riding a bicycle.”

Read more on Invoking archetypes in the #prompt-faqs channel of the Midjourney’s Discord server.

Add Details

To further improve the quality of your image, consider adding details on the subject’s emotions, attitude, image style, lighting, etc.

Here I decided to create a “mysterious” queen in the style of “cinematic photography.”

For your images, you may want to explore different:

• Emotions and attitudes: laughing, smiling, sad, worried, surprised, curious, charismatic, fierce.

• Styles: fantasy art, concept art, surrealism, in the style of specific artists.

• Aesthetics: stock photography, fashion photoshoot, portrait.

Improve with Remix and Vary

Once you get an image that you like in terms of general impression and aesthetics, you can experiment with details using Remix and Vary options.

For example, I like the image below, but a right hand doesn’t look right 😃 

So, I leveraged the Vary (Region) option to improve this area. After a few re-rolls, I got the result like this.

The latest Midjourney features provide a tone of options for further experiments and perfection. You may want to change colors, add or remove objects, guide the changes with reference images, zoom out, extend the image vertically or horizontally, change its aspect ratio, etc. The creative process with Midjourney can be quite fun!

Happy Prompting!

🗞 News and Top Reads

• Microsoft is testing a new background removal tool in Paint.

  • The new feature lets you remove an image’s background with a single click. You can also manually select the portion of the background that you want to remove.

  • The tool is currently only available to Windows testers.

• Motorica, a Sweden-based AI company, has officially launched the open beta phase for MoGen, its tool for AI-generated character animation, which is powered by diffusion models.

  • MoGen was trained on a diverse dataset of over 150 styles.

  • Users can modify preset motion paths or create their own within any compatible software, including Maya, Blender, Unity, and Unreal Engine.

• Time has published TIME 100 AI, a list of the 100 most influential people in artificial intelligence.

  • The list features, among others, Sam Altman from Open AI, Elon Musk from xAI 😆, and Clément Delangue from Hugging Face.

  • A few artists have also been included: Stephanie Dinkins, Sougwen Chung, Linda Dounia Rebeiz, Kelly McKernan, and Joy Buolamwini.

📌 AI Art Tutorial: 5 Ways to Avoid Symmetry in Midjourney

In this video tutorial, Nolan from Future Tech Pilot shares 5 tricks to avoid unwanted symmetry and get some interesting images with dynamic framing.

🎨 Featured Artist: Sougwen Chung

Sougwen Chung is an internationally renowned artist and researcher exploring the boundaries of collaboration between human and machine. A former research fellow at MIT, Chung has programmed and built robots named D.O.U.G. (Drawing Operations Unit Generation_X). These machines mimic the artist’s hand-drawn gestures and synchronously draw along with her. The custom robots make marks in a feedback loop with the artist, generating sketches based on neural nets trained on Sougwen’s drawings gestures and biometrics.

🖼 AI-Assisted Artwork of the Week

🤓 A Comprehensive Midjourney Guide

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!