Luna SDXL is a model with several versions
for high-quality, stylish character images (e.g., realistic, anime, semi-realistic)
Hey everyone! 👋
Here's a handy guide to creating stunning images with an SDXL checkpoint—ideal for artistic, feminine, and sensual representations in a clear, simple format.
I've already shared several versions, ranging from realistic to anime to semi-realistic styles.
Now I'll show you how to build precise and creative prompts using a modular system.
🎨 AI Prompt Builder Guide for High-Quality Character Art
Create stunning, ultra-detailed character art using this modular prompt system. This guide breaks down how to construct powerful prompts step-by-step for your text-to-image AI workflows.
🧱 1. Base Quality & Style
Start your prompt with high-level artistic and technical attributes. These control the overall fidelity and mood of the image.
(masterpiece, best quality, ultra-detailed, highres, cinematic lighting, soft lighting, volumetric light, golden hour, intricate details)
Optional extras:
(dynamic angle, dramatic shadows, depth of field, lens flare, ambient haze, 64k texture)
🧝 2. Main Character Description
Describe your subject with appearance, outfit, and distinguishing features. This part determines the focus of the artwork.
(full body of a youthful [character type], [hairstyle], [eye color], [skin type]), ([outfit], [accessories], [distinct features – wings, glowing eyes, tattoos, etc.])
Character examples:
valkyrie goddess
cyberpunk mercenary
fantasy elf
dark mage
gothic vampire
futuristic android
fashion model
Hair examples:
long platinum hair, curly blonde, twin braids, pink ombré, short bob cut, windswept silver
Eyes:
glowing blue, deep violet, heterochromatic, fiery red
Outfits:
ornate golden armor, sleek bodysuit, black lace dress, futuristic battlewear, gothic lolita
Accessories:
crystal crown, choker, thigh straps, belts, runes, earrings, capes
🏞️ 3. Background / Environment
Set the scene. Use location, atmosphere, and specific visual details to anchor your character in a world.
(background: [location], [lighting/atmosphere], [environmental elements])
Scene examples:
enchanted forest
glowing meadow
fashion boutique
ancient ruins
futuristic city alley
moonlit cathedral
Lighting & mood:
golden sunbeams, volumetric fog, glowing embers, particles, reflections, soft haze
😌 4. Expression & Mood
Capture the emotional essence of the character by defining facial expression and gaze.
(expression: [emotion or vibe]), (eye contact, soft smile, subtle blush)
Examples:
serene, playful, confident, seductive, mischievous, shy, intelligent, peaceful
🎭 5. Motion / Energy / Style Enhancers
Add cinematic flair by describing dynamic posing, movement, or magical elements.
text
KopierenBearbeiten
(dynamic pose, flowing hair, floating fabric, sparkles, energy trails, glowing symbols)
🚫 6. Negative Prompt
These terms help avoid common visual flaws or unwanted styles.
EasyNegative, bad-hands-5, score_6, score_5, score_4, bad anatomy, watermark, signature, extra limbs, lowres, blurry, worst quality, distorted face, big breasts, nsfw
Optional adds (if aiming for realism):
source_anime, source_cartoon, source_pony
🛠️7. SDXL Optimal Generation Settings
Sampling Method (Sampler): DPM++ 2M Karras (recommended for balance of detail and smooth gradients) Alternative: DPM++ SDE Karras (for softer artistic looks)
Sampling Steps: 25–40 steps (can go higher, but 25 is usually optimal)
CFG Scale (Classifier-Free Guidance): 7.0–8.5
Lower (6.5) = more adherence to the image model
Higher (8.5) = more literal prompt-following, may introduce artifacts
Image Size (Resolution for 3:4 portrait art):
Base Size: 832x1216 (ideal for SDXL, balanced memory use)
Max Quality (with good VRAM): 1024x1536 (sharp details, especially in faces and hands)
Tip: Keep multiples of 64 for SDXL compatibility
🔁 Full Prompt Example
(masterpiece, ultra-detailed, cinematic lighting, golden hour), (full body of a youthful valkyrie goddess, curly blonde hair with twin braids, bright blue eyes, porcelain skin), (ornate gold and white armor, glowing crystal crown, soft pink wings, flowing cape), (background: ancient temple ruins in a lush meadow, sun rays, glowing haze, drifting particles), (expression: serene, wise and confident smile, direct gaze), (dynamic pose, hair floating slightly, soft sparkles around)
Negative prompt: EasyNegative, bad-hands-5, score_6, score_5, score_4, watermark, extra limbs, lowres, distorted anatomy, big breasts, signature
Have fun and post your pictures.
wishes your Greifer