Here is something most people discover the hard way about describing colors in AI prompts.
You paste your brand hex code into a Midjourney prompt. You run it. The color that comes back looks nothing like your brand color. You try again. Still wrong. You add 'exact color' to the prompt. Still wrong.
This is not a Midjourney bug. It is a fundamental misunderstanding of how AI image generators process color information.
Midjourney does not read hex codes. It reads language. And there is a specific kind of color language that produces accurate, consistent, on-brand results every time.
Midjourney does not read hex codes. It reads language.
This guide teaches you exactly that language.
Why hex codes do not work in AI prompts
Hex codes are instructions for screens. They tell a monitor exactly which wavelength of light to emit at which pixel. Midjourney is not a monitor. It is a language model that has learned to associate descriptive words with visual concepts.
When you write #8B1A4A in a prompt, Midjourney sees a string of characters with no visual meaning attached. It ignores it or interprets it randomly.
When you write 'deep burgundy with subtle violet undertones, rich and saturated, matte finish' - Midjourney has thousands of training images associated with exactly that description. It knows what that looks like. It renders it accurately.
The translation from hex code to descriptive language is the entire job. And once you learn it, your AI images start looking dramatically more on-brand.
The translation from hex code to descriptive language is the entire job.
The 4 dimensions of color description
Every color can be described across 4 dimensions. Use all 4 and your color renders accurately. Skip any of them and the AI fills in the gap randomly.
A hex code alone renders randomly - naming all 4 dimensions renders it accurately.
Dimension 1 - Color family
The base color name. Not just 'blue' or 'red' but a specific named shade within that family.
Navy with cool gray undertones vs navy with warm indigo undertones - completely different feels
Red with blue undertones (crimson) vs red with orange undertones (scarlet) - different moods entirely
White with warm cream undertones vs white with cool blue undertones - different brand personalities
Always state the undertone explicitly. 'Deep teal with subtle blue-green undertones' is twice as accurate as just 'deep teal'.
Dimension 3 - Saturation and depth
How pure and intense is the color? How light or dark?
Saturation descriptors:
Vivid, saturated, pure, intense - high saturation, bold and energetic
Muted, dusty, desaturated, toned - low saturation, sophisticated and calm
Pastel, soft, pale, washed - very low saturation, gentle and delicate
Depth descriptors:
Deep, dark, rich, moody - low lightness
Medium, balanced, true - mid lightness
Light, pale, soft, airy - high lightness
Combining them: 'deep muted burgundy' tells you exactly where in the color space this lives. 'Pale vivid coral' tells you something completely different.
Same hue, five saturation levels - the word you choose changes the whole feel.
Dimension 4 - Finish and how light interacts
The same color looks completely different depending on the surface it is on. This dimension is what makes colors feel real and touchable in AI images.
Matte finish - absorbs light evenly, no reflections, flat and sophisticated
'deep burgundy with subtle violet undertones, rich and saturated, dark and moody'
Brand color consistency across multiple images
The biggest challenge with AI brand photography is not getting one image right - it is getting 20 images that all feel like the same brand.
The biggest challenge with AI brand photography is not getting one image right - it is getting 20 images that all feel like the same brand.
The solution is a brand color vocabulary. A set of 4 to 6 specific color phrases that you use consistently across every prompt. Once you have these phrases they become part of your standard prompt template and every image you generate shares the same color language.
What a brand color vocabulary looks like:
Primary: 'deep burgundy with subtle violet undertones, rich and saturated, matte velvet finish'
Secondary: 'warm champagne with soft golden undertones, light and luminous, satin sheen'
Accent: 'antique brass with matte oxidized finish, aged warm metal'
Background: 'soft warm cream with peachy undertones, pale and inviting, matte'
Avoid: 'neon colors, cool blue-white whites, pure black, any metallic chrome'
Drop this vocabulary into every prompt and your images start looking like a coherent brand campaign rather than a random collection of AI outputs.
Get your color vocabulary in seconds
Building a brand color vocabulary manually takes time. Our free Brand Color Palette Translator does it automatically.
Enter your brand hex codes, select your brand personality and content goal, and the tool translates everything into prompt-ready color language, including material suggestions, lighting recommendations, mood keywords, and two complete sample prompts using your palette.
Already know your color language and ready to build a full prompt? The Midjourney Prompt Builder has a dedicated color and materials layer where you can apply your vocabulary directly.