Luna SDXL

CHECKPOINT
Original - TenStar Fund


Updated:

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

Version Detail

SDXL 1.0

Project Permissions

Reprinting is strictly prohibited

    Use Permissions

  • Use in TENSOR Online

  • As a online training base model on TENSOR

  • Use without crediting me

  • Share merges of this model

  • Use different permissions on merges

    Commercial Use

  • Sell generated contents

  • Use on generation services

  • Sell this model or merges

Related Posts

Describe the image you want to generate, then press Enter to send.