05 / 10

    AI Image Generation

    Discover the world of AI images — from DALL·E to Midjourney and Stable Diffusion. Learn to write prompts that create exactly what you envision.

    ~10 minDALL·EMidjourneyStable Diffusion

    Overview of Key Tools

    Each tool has its strengths. The choice depends on your use case, budget, and desired level of control.

    DALL·E 3

    OpenAI

    Text in images, quick prototypes

    Integrated in ChatGPT. Strong at following prompts precisely and rendering text in images.

    Midjourney

    Midjourney Inc.

    Artistic visuals, marketing

    Best aesthetic quality. Works via Discord or web. Ideal for artistic and marketing visuals.

    Stable Diffusion

    Stability AI

    Local running, full control

    Open-source model you can run locally. Maximum control and customizability.

    Adobe Firefly

    Adobe

    Commercial safety, integration

    Trained on licensed content. Safe for commercial use, integrated in Photoshop.

    Anatomy of an Image Prompt

    A good image prompt has several layers. The more details you add, the more precise the result.

    🎨

    Specify the style

    Photorealistic, illustration, watercolor, 3D render, pixel art...

    📐

    Describe composition

    Close-up, wide shot, bird’s eye view, symmetrical...

    💡

    Describe lighting

    Golden hour, studio light, neon glow, dramatic shadows...

    Add mood

    Calm, energetic, mysterious, nostalgic, futuristic...

    Weak vs. Strong Prompt

    Weak prompt

    Draw a cat.

    Strong prompt

    Photorealistic portrait of a ginger cat sitting on a windowsill on a rainy day. Soft natural light from the window, bokeh background, detailed fur, Canon EOS R5, 85mm f/1.4, cinematic color grading.

    💡 Tip

    Add technical parameters like lens type, lighting, or film look — they significantly improve output quality.

    Negative Prompts

    Tell AI what you don’t want in the image. Negative prompts significantly improve quality, especially in Stable Diffusion and Midjourney.

    Example

    Prompt: Professional product photo of a white flower pot on a marble table, studio lighting

    Negative: blurry, low quality, watermark, text, distorted, extra objects, busy background

    Practical Tips

    1

    Iterate

    Start simple and gradually add details. Each variant teaches you something new.

    2

    Combine styles

    “Watercolor in Studio Ghibli style” or “3D render in Pixar style.”

    3

    Use references

    Upload an existing image as an inspiration reference (image-to-image).

    4

    Test aspect ratios

    16:9 for banners, 1:1 for social media, 9:16 for stories.

    Try it yourself

    Open ChatGPT (DALL·E) or Midjourney and try this prompt:

    Minimalist poster for a Japanese-style cafe. Natural colors, subtle paper texture, matcha latte cup in the foreground, misty bamboo forest in the background. Style: graphic design, flat illustration, 3:4 aspect ratio.

    Open ChatGPT

    Key Terms

    Image prompt

    A text description used by AI to generate an image.

    Negative prompt

    What you don’t want in the image — e.g. “no text, no watermark”.

    Seed

    A number determining the initial noise. Same seed + prompt = same result.

    Aspect ratio

    The proportions of the image, e.g. 16:9, 1:1, 9:16.

    Upscaling

    Increasing the resolution of a generated image without quality loss.

    Inpainting

    Editing a specific part of an image while keeping the rest intact.

    Lesson Summary

    • DALL·E, Midjourney, Stable Diffusion, and Adobe Firefly — each tool has its advantages and ideal use cases.
    • The more detailed the prompt (style, lighting, composition, mood), the better the result.
    • Use negative prompts, iterate, and experiment with different styles.