6 Ways to Control Colors in AI Images

In this newsletter, read about:

• 🕵️‍♀️ Controling Colors in AI Images

• 🗞 News and Top Reads

• 📌 AI Art Tutorial: InsightFace Bot

• 🎨 Featured Artist: doopiidoo

• 🖼 AI-Assisted Artwork of the Week

• 🤓 How to Get Started with Generative AI?

🕵️‍♀️ How to Control Colors in AI Images

Have you ever tried generating an image where you want specific objects to have specific colors? If so, you probably know that it can be quite challenging for image generators like Midjourney and others.

For example, if you request an image of a woman in a red dress and black shoes, you'll likely get results where the colors are misplaced and mixed up. Moreover, the background and other objects may also appear in red and black colors. That's because AI image generators find it challenging to understand how to assign colors to specific objects. When you include a color keyword in your text prompt, it typically means that any of the objects in the image might be depicted in that color.

While achieving 100% control over the colors in AI-generated images can be challenging, there are methods that can provide you with a significant degree of control. I would be happy to share some tips with you.

Use general color palettes

First of all, you can use general color palettes in your text prompts. Midjourney responds quite well to the keywords like:

• pastel colors,

• autumn colors,

• tropical colors,

• cold colors,

• sepia,

• greyscale,

• shades of [color], etc.

If you are chasing a specific color assignment, you can use this color scheme method and reroll the results until you see the permutation you are after.

Create dominant colors

When relevant, you can explicitly create dominant colors in your text prompt. Begin by including the color keyword and assign it a text weight greater than 1. This way, the output images will adhere to the corresponding color scheme.

Use text weights to balance colors

Text weights can go beyond just creating a dominant color; they can also be used to remove certain colors from an image and balance other colors. If you wish to remove a specific color from your image, assign a negative weight to it. However, it's important to ensure that the sum of all text weights remains positive. To achieve this, assign positive weights to other colors that you prefer to have in your image.

For instance, in my text prompt below, I specify "green and yellow:: red::-0.3". This means assigning a weight of 1 to green and yellow, and a negative weight of -0.3 to the red color. As a result, the images generated are highly dominated by green and yellow. You can adjust these weights according to your specific color preferences for an image.

Pay attention to word order

In text prompts, the order of words carries significant importance. AI tends to pay more attention to the words positioned earlier in the prompt. Therefore, if you have two objects that you want to specify colors for, it is advisable to start with the object that occupies more space in the image.

For instance, let's consider an example where you want to depict a woman in the image wearing a black jacket and a brown shirt. Since a jacket typically covers more surface area than a shirt, starting with the shirt color may not yield optimal results, as shown below.

By simply swapping the positions of "black jacket" and "brown shirt" in the text prompt, you significantly increase your chances of getting the requested output.

Define material instead of color

Indeed, in certain cases, when you desire contrasting colors in your image, specifying a strong color preference for one object may have a negative impact on the overall image. As we observed earlier, the keyword "black jacket" worked well when aiming for autumn colors. However, if your intention is to have pastel colors in the background, the inclusion of the word "black" could potentially hinder the desired outcome.

As you see, the background here is quite dark, and not exactly in the pastel colors.

In such scenarios, a useful approach is to employ a specific trick: instead of directly specifying a color, focus on indicating the material commonly associated with certain colors. For instance, when referring to clothes, you can consider:

• Cotton, typically associated with white color.

• Wool, often linked to gray, brown, or black.

• Leather, commonly connected to brown or black shades, and so on.

By utilizing material references, you can provide indirect guidance to AI image generators, enhancing the likelihood of achieving the desired color outcomes.

As you see, this trick worked for the image in autumn colors. Unfortunately, it’s not always the case.

When requesting a leather jacket in an image that predominantly features pastel colors, the jacket often ends up adopting the same color scheme. Fortunately, there is another trick we can employ to address this issue.

Use the reference image method

This technique is slightly more advanced and requires basic skills in Procreate, Adobe Express, or Photoshop. Follow these steps:

1. Begin by selecting an AI-generated image that closely resembles what you're looking for but requires some color adjustments.

2. Utilize image editing tools such as Procreate, Adobe Express, or Photoshop to fix the colors in the selected image. The final result doesn't need to be perfect or visually polished; the focus should be on achieving the right colors in the desired areas.

3. Once you have made the necessary color adjustments, upload the modified image from Step 2 as a reference image to Midjourney or another AI image generator. Generate a new set of images using the same prompt, taking into account the changes made in the reference image.

Now let's revisit the previous example.

While the original image below is appealing, our goal is to have a black leather jacket, maintain bright pastel-colored houses in the background, and potentially introduce more green color to the scenery.

Step 1

Step 2

Step 3

Here we go!

While achieving 100% control over colors may be challenging, utilizing certain techniques might bring us closer to the desired results. From defining dominant colors and utilizing text weights to carefully considering word order and employing image editing tools, these strategies enhance our ability to control colors in AI-generated images. As AI continues to advance, so does our potential for precise color manipulation.

🗞 News and Top Reads

• Google is starting to open up access to new generative AI capabilities in Search.

• Clipdrop by Stability.ai presented Reimagine XL – an upgraded image reimagination tool.

  • Simply upload an image and the tool will generate unique and diverse variations of your image.

• ChatGPT is gaining popularity in the US.

  • Pew Research Center survey conducted in March showed that about six-in-ten U.S. adults (58%) are familiar with ChatGPT, though only 14% have tried it themselves.

• AI experts and public figures have published a new statement on AI risks.

  • They suggest that mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.

📌 AI Art Tutorial: InsightFace Bot

In this tutorial, Olivio Sarikas shows how to use a free InsightFace bot that allows adding your face to any image – Midjourney-generated or any image you find online. The tool works great with different styles.

🎨 Featured Artist: doopiidoo

doopiidoo is a successful AI creator with works featured in exhibitions all over the world, including Paris, New York, and Rome.

🖼 AI-Assisted Artwork of the Week

🤓 How to Get Started with AI Art?

1. DALL-E: Creating Images from Text – introduction to text-to-image generation.

2. The DALL-E 2 Prompt Book – a guidebook by OpenAI that explains how to effectively right prompts to generate images across different domains (e.g., photography, illustration, art history, 3D artwork).

3. All-In-One Guide For Midjourney: The Art Of Prompts – a comprehensive guide on creating effective Midjourney prompts by Karol Andruszków.

4. Best Midjourney Prompts – a guide that covers the basics of Midjourney prompts (e.g., which keywords to use to create abstract art, surreal art, minimalism, etc) as well as some more advanced options (e.g., keywords related to camera lenses and filters, imitating certain artists and photographers without using their names). Finally, they provide a list of 600+ creative text prompts for image generation.

5. Stable Diffusion Prompt Book – a prompt book prepared by OpenArt. The book discusses ideal prompt format, using modifiers to change the style, format, or perspective of the image, applying ”magic words” to improve image quality, adding negative prompts, and adjusting Stable Diffusion parameters.

Share Kiki and Mozart

If you like this newsletter and know somebody who might also like it, feel free to share this newsletter. Let’s have more people learn about AI art!

If you have been forwarded this email and you like it, please subscribe below. And welcome to the world of AI art!