How to Create Lifelike People in Midjourney

In this newsletter, read about:

• 🕵️‍♀️ Creating Lifelike People in Midjourney

• 🗞 News and Top Reads

• 📌 AI Art Tutorial: Consistent Styles for Storytelling

• 🎨 Featured Artіst: Richard Nadler

• 🖼 AI-Assisted Artwork of the Week

• 🤓 A Comprehensive Midjourney Guide

🕵️‍♀️ Creating Lifelike People in Midjourney

AI image generators, like Midjourney, often produce images of people that might appear somewhat artificial, you know, not exactly like the ones you see on the streets. Despite their impressive realism, they tend to resemble models from professional photoshoots rather than everyday individuals.

The reason behind this lies in the specific data used to train these AI models. They primarily learn from beautiful photos captured by skilled photographers and meticulously processed in tools like Photoshop to achieve perfection.

However, generating more natural photo portraits of real people is indeed possible in Midjourney with a few clever tricks. Let's delve into the parameters, settings, and keywords that can help us achieve this goal.

1. Use Style Raw

In today's experiments, our goal is to capture close-up portraits of individuals that appear authentic and lifelike. Simply mentioning the word "photo" in the prompt typically yields photorealistic outcomes. However, in the Midjourney versions 5.1 and 5.2, we can further enhance the realism by utilizing the Raw style (i.e., --style raw), resulting in even more photographic images. From my experience, it is an essential parameter when aiming to create a natural-looking depiction of a person.

2. Minimize stylization

Next, I recommend using stylization at a minimum (e.g., --s 50) if you want to get a resemblance of an everyday person. Usually, I love to experiment with high stylization values that produce stunning artistic images, but in this particular case, we want “less artistic” images and a closer match to the prompt.

As you can see, using just these two parameters already produces fairly natural images. However, let's explore how we can further enhance the authenticity of our characters.

3. Include negative prompts for makeup, filters, etc

It is usually helpful to include a negative prompt with keywords such as makeup, filters, Photoshop, etc. Essentially, we are seeking unedited and unfiltered images of characters without makeup. However, it is crucial to include these keywords as a negative prompt, i.e., after the --no parameter, to avoid getting the opposite results. If you include "without makeup" in your prompt, Midjourney is likely to overlook the "without" and generate images of people with heavy makeup.

To include multiple components in a negative prompt, separate them with commas after the --no parameter. (e.g., --no makeup, filters, photoshop).

4. Use the “selfie” keyword

Leveraging the “selfie” keyword can give another boost to the realism of your images and characters. Simply replace “a photo” or “a close-up photo” with “a selfie.”

5. Add additional keywords for a “natural” photo

Finally, you can utilize supplementary keywords to enhance the authenticity of the image. These may include terms such as "natural," "everyday life," "authentic," and so on. The positive influence of these words is not always apparent, but at times, they aid in achieving even more realistic outcomes.

It is astonishing how quickly we adapt to the amazing ability to generate photographic images from mere words in less than a minute. Of course, AI image generators still have several limitations, but by employing a few simple techniques, it is possible to transform the "default" output into the desired result, as demonstrated in this guide where we progressed from artificial-looking characters to remarkably genuine, lifelike individuals. I am confident that with continued advancements in generative AI, this process will soon become much easier.

Happy prompting!

🗞 News and Top Reads

• US District Judge William Orrick, the first to review the artists against AI lawsuit, does not support the artists' claims.

  • He expressed his inclination to dismiss most of the lawsuit during a hearing in San Francisco. However, he did allow the artists to file a new complaint.

  • Orrick stated that the artists need to clearly state and differentiate their claims against Stability AI, Midjourney, and DeviantArt.

  • He also emphasized the need for them to provide more facts regarding the alleged copyright infringement, as they have access to Stability's relevant source code.

  • Additionally, the judge stated that the artists are unlikely to succeed in their claim that the generated images using their names violated their copyrights.

• Apple is creating an AI framework, Ajax, and a chatbot service, Apple GPT, as part of its push to compete in the large language model space.

• Netflix is ready to pay up to $900K/year to a new AI Product Manager.

  • According to the job description, the Product Manager position will focus on increasing the leverage of Netflix’s Machine Learning Platform.

  • If you follow the company’s research updates, you’ll also notice that Netflix is expanding its use of AI to streamline and optimize film and TV production.

📌 AI Art Tutorial: Consistent Styles for Storytelling

Christian Heidorn shares a great video tutorial on how to use the High Variation Mode in Midjourney to keep your images in a consistent style. This trick can be used to create very similar environments for different characters, with regard to style, colors, and general visual effects. This is a very handy tool for storytelling.

🎨 Featured Artist: Richard Nadler

Richard Nadler is a generative artist who explores identity through potent colors and provocative textures. His work spans latent diffusion, GAN, and code-based techniques. He is most known for his Japanese GenArt Collection, which explores his lifelong fascination with the culture of Japan, a place he would frequently travel to with his late father. Based in Munich, Germany, Nadler splits his time between the digital world, a growing family, and his journey towards unofficial sommelier status.

🖼 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!