Read AI Image Alchemy free: the opening pages

Part 1: Foundations of AI Image Prompting

Decoding the Prompt: Anatomy of a Successful Request

Let’s be honest. You’ve probably typed a simple request into an AI image generator, “a cat,” “a landscape,” “a portrait”, and been…underwhelmed. The results were okay, perhaps, but rarely what you envisioned. That gap between your mental image and the AI's output? That’s what we’re going to close. This isn't about the AI being “bad”; it’s about learning to communicate your ideas with precision. We're not teaching a computer to think like an artist; we're providing incredibly specific instructions.

This section breaks down what makes a prompt effective. We’ll move from vague ideas to detailed requests, giving you a framework you can apply to any AI image generator. Forget abstract concepts; we’re focusing on practical techniques you can use immediately.

Why Prompt Anatomy Matters

Think of an AI image generator as an exceptionally skilled, but incredibly literal, artist. Tell it "draw a house," and it will. But it has no idea what kind of house you want. A cozy cottage? A futuristic skyscraper? A haunted mansion? The more information you provide, the closer the result will be to your vision. A well-structured prompt provides clarity and direction, leading to consistent, high-quality images. It's the difference between asking for "food" and requesting "a medium-rare ribeye steak with roasted vegetables and a side of béarnaise sauce."

The Five Core Components

Every successful prompt, regardless of the AI model, relies on five core components. We’ll call this the “SSSDD” framework: Subject, Style, Setting, Details, and Quality. Understanding these components is the first step toward consistently creating images you love.

• Subject: What is the primary focus of the image? (e.g., a cat, a spaceship, a portrait of a woman)
• Style: What artistic style should the image be in? (e.g., photorealistic, impressionistic, cyberpunk, anime)
• Setting: Where and when does the scene take place? (e.g., a futuristic city, a medieval forest, a sun-drenched beach)
• Details: Specific characteristics or elements to include. (e.g., fluffy fur, glowing neon lights, intricate patterns)
• Quality: Keywords that influence the overall rendering quality. (e.g., highly detailed, 8k, cinematic lighting)

Let’s illustrate this with an example. We'll start with a simple idea and build it up.

Initial Idea: A cat.

This is…not very helpful. Let's apply the SSSDD framework.

Subject: A fluffy Persian cat Style: Photorealistic Setting: Lounging on a velvet chaise lounge in a sun-drenched Victorian library Details: Golden eyes, long white fur, antique books in the background Quality: Highly detailed, 8k, cinematic lighting

Now we have a much more descriptive prompt. The difference in output will be dramatic.

From Vague to Vivid: A Worked Example

Let’s trace the evolution of a prompt, showing how adding each component improves the result. We’ll use Midjourney for these examples, but the principles apply to all platforms.

Step 1: The Bare Minimum, “cat”

As you can see, it’s…a cat. Generic, unremarkable, and lacks any personality.

Step 2: Adding Subject & Style, “Persian cat, photorealistic”

Better! We've established the breed and the desired style. However, it’s still isolated and lacks context.

Step 3: Incorporating Setting, “Persian cat lounging on a velvet chaise lounge in a Victorian library, photorealistic”

The setting adds depth and narrative. The image is more interesting, but still feels a bit flat.

Step 4: Adding Details, “Persian cat lounging on a velvet chaise lounge in a Victorian library, golden eyes, long white fur, antique books in the background, photorealistic”

Now we’re getting somewhere! The details bring the image to life. The cat has distinct features, and the background is richer.

Step 5: Enhancing Quality, “Persian cat lounging on a velvet chaise lounge in a Victorian library, golden eyes, long white fur, antique books in the background, photorealistic, highly detailed, 8k, cinematic lighting”

The final result is a stunning, detailed image that closely matches a specific vision. Notice how each addition built upon the previous one, creating a more compelling and refined image.

Action step: Take a simple idea (e.g., "a tree") and work through these five steps, adding one component at a time. Observe how the image changes with each addition.

Prompt Structure: The "ABCD" Framework (A Simplified Approach)

While SSSDD provides a comprehensive breakdown, sometimes a more streamlined approach is helpful, especially when starting out. The “ABCD” framework is a simplified version, focusing on the most impactful elements:

• Action: What is happening in the image? (e.g., running, singing, exploding)
• Bias (Style): What style are we leaning toward? (e.g., anime, realistic, cartoon)
• Characters/Creatures: Who or what is in the image? (e.g., a knight, a dragon, a spaceship)
• Details/Decor: What extra elements are present? (e.g., glowing swords, rain, ancient ruins)

Example:

“A knight running through a forest, fantasy art, a majestic dragon flying overhead, glowing swords and ancient ruins.”

This framework is excellent for quickly generating ideas and building a base prompt. You can always add more detail later.

Comparing Weak vs. Strong Prompts

Weak Prompt	Strong Prompt	Expected Outcome
"A flower"	"A vibrant sunflower in a field of wildflowers, golden hour lighting, photorealistic, highly detailed"	Weak prompt will produce a generic, unremarkable flower. Strong prompt will deliver a beautiful, realistic sunflower bathed in warm light.
"A city"	"A futuristic cyberpunk city at night, neon lights, flying cars, rainy streets, highly detailed, 8k"	Weak prompt will likely generate a bland, generic city. Strong prompt will create a visually striking and immersive cyberpunk scene.
"A portrait"	"A portrait of a young woman with piercing blue eyes, Renaissance painting style, dramatic lighting, detailed clothing, soft focus"	Weak prompt will result in a standard, uninspired portrait. Strong prompt will produce a captivating and artistic portrait reminiscent of Renaissance masters.

Pro tip: Keep a running document of prompts that work well for you. This will serve as a valuable resource and save you time in the future.

Checklist for a Successful Prompt

Before you hit "generate," run through this checklist:

• Subject is clearly defined: What is the main focus of the image?
• Style is specified: What artistic style do you want?
• Setting is established: Where and when does the scene take place?
• Details are included: What specific elements should be present?
• Quality keywords are added: Are you using keywords like "highly detailed," "8k," or "cinematic lighting"?
• Prompt is concise and readable: Avoid overly long or convoluted sentences.

Beyond the Basics: Experimentation is Key

This framework is a starting point, not a rigid set of rules. Don’t be afraid to experiment with different combinations of keywords, styles, and details. The more you practice, the better you’ll become at crafting prompts that consistently deliver the results you want. Remember, AI image generation is a creative process. Have fun, explore, and see what you can create!

Workthrough: From Vague Idea to Detailed Prompt

Let's take the simple idea of "a cat" and transform it into a compelling, detailed prompt.

Step 1: Subject, Specificity is Key

Instead of just "a cat," let's be more specific. What kind of cat? "A fluffy Persian cat." This immediately narrows the focus and provides more information for the image generator.

Step 2: Action, What is the Subject Doing?

Cats are known for lounging. Let's add that: "A fluffy Persian cat lounging." Now we have a sense of pose and activity.

Step 3: Setting, Where is the Action Taking Place?

A simple room isn’t very inspiring. Let’s create a more evocative setting: "A fluffy Persian cat lounging on a velvet chaise lounge in a sun-drenched Victorian library." This adds context, atmosphere, and visual richness.

Step 4: Style, Artistic Direction

What kind of feel are we going for? "A fluffy Persian cat lounging on a velvet chaise lounge in a sun-drenched Victorian library, dramatic lighting." We’ve introduced a stylistic element that influences the overall mood and appearance.

Step 5: Details, The Finishing Touches

Now, let's add even more detail to enhance the image: "A fluffy Persian cat lounging on a velvet chaise lounge in a sun-drenched Victorian library, dramatic lighting, highly detailed, ornate bookshelves filled with ancient tomes, dust motes dancing in the light, warm color palette."

Final Prompt: "A fluffy Persian cat lounging on a velvet chaise lounge in a sun-drenched Victorian library, dramatic lighting, highly detailed, ornate bookshelves filled with ancient tomes, dust motes dancing in the light, warm color palette."

This prompt is significantly more descriptive than the original "a cat." It provides the image generator with a wealth of information, increasing the likelihood of generating a visually stunning and satisfying image.

Let's look at another example, building from "a dragon":

1. Subject: "A majestic, emerald green dragon"
2. Action: "A majestic, emerald green dragon soaring through a stormy sky"
3. Setting: "A majestic, emerald green dragon soaring through a stormy sky above jagged, snow-capped mountains"
4. Style: "A majestic, emerald green dragon soaring through a stormy sky above jagged, snow-capped mountains, painted in the style of Frank Frazetta"
5. Details: "A majestic, emerald green dragon soaring through a stormy sky above jagged, snow-capped mountains, painted in the style of Frank Frazetta, lightning illuminating its scales."

This illustrates how adding layers of detail transforms a simple idea into a complex and visually rich concept.

Understanding Keyword Weighting and Syntax

While a well-structured prompt is essential, understanding how to emphasize certain keywords can further refine your results. Most image generators allow you to adjust the “weight” of specific terms. This tells the generator to pay more attention to those elements.

Using Parentheses and Brackets:

• Parentheses ( ): Increase the weight of a term. For example, "(fluffy cat)" tells the generator that "fluffy" is more important than other descriptors.
• Brackets [ ]: Decrease the weight of a term. For example, "[small cat]" suggests that the size of the cat is less critical.

Example: "A (fluffy) cat [sitting] on a mat." This prompt will prioritize the fluffiness of the cat while downplaying the action of sitting.

Using Colons for Emphasis:

Some generators allow you to use colons to assign a specific weight to a keyword.

Example: "A cat::2, a dog::0.5." This prompt gives the cat twice the weight of the dog, making it the dominant element in the image.

Important Note: Syntax and weighting options vary between different image generators. Always consult the documentation for the specific tool you are using.

Mastering Descriptive Language: Beyond Basic Adjectives

Strong prompts rely on vivid and descriptive language. Instead of simply saying "a beautiful flower," try using more evocative terms like "a delicate, porcelain-like orchid" or "a vibrant, crimson rose with velvety petals."

Here’s a table of stronger descriptive words to replace common adjectives:

Weak Adjective	Stronger Alternatives
Beautiful	Exquisite, stunning, radiant, captivating
Good	Exceptional, remarkable, outstanding, superb
Big	Immense, colossal, gigantic, vast
Small	Miniature, tiny, petite, diminutive
Happy	Joyful, elated, blissful, cheerful
Sad	Melancholy, somber, despondent, mournful

Using Sensory Details:

Engage the viewer's senses by incorporating details that appeal to sight, sound, smell, taste, and touch.

• Sight: "Gleaming gold armor," "Iridescent scales," "Ethereal glow"
• Sound: "The roar of the waterfall," "The gentle rustling of leaves," "A haunting melody"
• Smell: "The scent of pine needles," "The aroma of freshly baked bread," "A musky, earthy fragrance"

Negative Prompting: Telling the Generator What Not to Include

Just as important as telling the generator what you want is telling it what you don’t want. Negative prompts allow you to exclude specific elements from the image.

Example: "A fantasy landscape, highly detailed, vibrant colors, --no people, --no buildings." This prompt will generate a landscape without any human figures or structures.

Common Negative Keywords:

• "blurry"
• "grainy"
• "low resolution"
• "artifacts"
• "text"
• "watermark"

Experiment with negative prompts to refine your images and eliminate unwanted elements.

Iteration and Refinement: The Prompting Process

Creating the perfect image is rarely a one-shot deal. It often requires iteration and refinement.

1. Start with a base prompt: Begin with a solid foundation as we discussed earlier.
2. Generate an image: Run the prompt and analyze the results.
3. Identify areas for improvement: What aspects of the image are not quite right?
4. Modify the prompt: Adjust keywords, add details, or use negative prompts.
5. Repeat steps 2-4: Continue iterating until you achieve the desired outcome.

Remember, prompting is a skill that improves with practice. Don’t be afraid to experiment, learn from your mistakes, and explore the endless possibilities of AI image generation.

The sample ends here.