General guide of writing prompts for SDXL models

Have you ever looked at those stunning AI-generated artworks and wished you could create something equally impressive? The secret lies in mastering the use of “prompts.” This guide will walk you through the techniques of writing effective prompts to unlock the full potential of AI image generation tools✨


What is a Prompt?

A prompt is essentially the instruction or description you provide to the AI to generate an image. It can be a simple keyword (like “Kafka”) or a detailed scene description (like “a cute anime girl wearing sunglasses on a sunny day”). The more specific and detailed your prompt is, the better the AI can understand and bring your imagination to life.

On PixAI, the most effective way to structure your prompts when using SDXL models is by using descriptive ‘tags’ or short phrases separated by commas. This ensures clarity and precision. For example:

“1girl, long flowing red hair, green eyes, elegant black dress, lace details, standing, balcony, neon lights, cinematic lighting, ultra-detailed”

Enable the autocomplete feature for quicker prompt input!


Commonly Used Prompt Categories for Inspiration

If you need some inspiration, here are popular tags to get you started:

  • Actions: holding, sitting, hand on hip, v, heart hand, tying hair
  • Clothing: dress, bikini, serafuku, kimono
  • Pose: looking at viewer, profile, facing away, selfiemirror
  • Subject Type: 1girl, chibi, fox girl, 2boys
  • Scene:
    • Indoor: library, stage
    • Outdoor: field, cyberpunk city, amusement park
    • Special: onsen, castle
  • Special Effects: blurry, sparkle, motion lines, film grain, cinematic lighting
  • Background: simple background, white background, valentine
  • Camera Angle: cowboy shot, from behind, from side, from above, from below, close-up
  • Artistic Style: sketch, traditional media, realistic, retro artstyle, watercolor

6 Pro Tips for Writing Highly Effective Prompts

  1. Order Matters: Tell a Visual Story

The order of your prompt influences how the AI prioritizes elements. We recommend this structure: Subject — Action — Location — Aesthetic & Style.

Example structure:

“1girl, bangs, long hair, blonde hair, green eyes, hair ribbon, serafuku, blue skirt, white knee socks, indoor shoes, sitting, holding books, full body, looking at viewer, school, classroom, chair, masterpiece, highest quality, beautiful face, highly detailed, watercolor.”

Tags placed at the beginning carry more weight, so always lead with your most important elements.

  1. Emphasis & De-Emphasis: Control Visibility with Symbols
  • Use ( ) parentheses to strengthen a feature
  • Use [ ] square brackets to weaken an element

You can use multiple layers of brackets for stronger effects.

Example:

  • Strengthening: 1boy, short hair, high school uniform, standing, balcony, (cloud), ultra-detailed, cinematic lighting.
  • Weakening: 1boy, short hair, high school uniform, standing, balcony, [cloud], ultra-detailed, cinematic lighting.

Weakening

Original

Strengthening

  1. Be Specific, But Don’t Overwhelm

Focus on the character’s appearance, expression, and pose. Avoid adding too many unrelated elements that distract from the main subject.

✅ Good prompt example:

“1girl, wavy blonde hair, green eyes, floral sundress, sitting, wooden bench, sunny, garden, soft and warm lighting, realistic style”

❌ Overloaded and conflicting prompt (avoid this):

“1girl, wavy blonde hair, green eyes, floral sundress, sitting, v, holding a doll, wooden bench, sunny garden, dragon, flying, castle, beach, palm trees, realistic style”

  1. Avoid Ambiguity

Vague prompts lead to unexpected results. Be clear about specific features.

  • Unclear: “A cool character” (What does “cool” mean?)
  • Clear example:

“1girl, spy, hair bun, black hair, aqua eyes, brown eyeshadow, makeup, red lips, red nails, stiletto nail, diamond manicure, jewelry, shiny skin, silver earrings, strapless dress, black dress, collarbone, black pantyhose, thigh, red high heels, standing, rooftop, from behind, cinematic lighting, ultra-detailed, highres”

  1. Use Negative Prompts

Specify what you don’t want in the image. Most AI tools provide default negative tags, but you can customize them.

Example negative prompts:

“six fingers, watermark, multiple characters”

  1. Experiment and Iterate

Try small adjustments to your prompts and observe the changes:

  • Change expressions: “smiling softly” vs. “serious expression”
  • Adjust lighting: “soft sunlight” vs. “dramatic shadows”
  • Try different styles: “anime-style” vs. “hyper-realistic”

Start Creating!

Writing effective prompts is both science and art. The most important thing is to maintain an experimental mindset—try, adjust, and learn which keywords and structures work best for conveying your creative vision.

Now, open your AI image generation tool, input your inspiration, and start creating unique visual artworks!

Index