Take the face from the input picture

Tested & Verified Photography Commercial License

Basic Prompt

An ultra-naturalistic portrait inspired by an accidental iPhone front-camera selfie: big, gentle doe-like eyes with long lashes, warm sleepy gaze looking slightly off-frame. Straight, medium-long dark brown hair with a middle part and soft layered ends falls naturally past the chest. Small delicate nose, naturally full lips with a soft pout, and a calm, dreamy expression quiet smile, as if lost in thought, utterly unposed. The image retains subtle imperfections: slight motion blur, uneven indoor lighting with mild overexposure on the forehead and cheekbone, and casual framing that feels impulsive and intimate. Enhanced with refined natural shaping around the neckline for graceful definition and subtle volume never retouched, never artificial. Shot on iPhone 14 front camera simulation, shallow depth, soft ambient light from a nearby window, muted skin tones, authentic bedroom ambiance.

To view Detailed Prompt please login

Add to Library

This isn’t a polished portrait it’s a stolen moment of quiet presence, elevated with care but never staged.

No approved reviews yet.

Write a comment

Select a rating

Model Settings & Parameters

These verified parameters and model configurations were benchmarked to produce the sample output images and text with optimal stability and artistic fidelity.

Target Engine Photography

Aspect Ratio 16:9 / 1:1

Stylize / CFG 250 – 750

Output Resolution UHD 4K / 8K

Seed Stability Random / Multi-run

Generation Time < 30 seconds

Known Limitations & Tips

Complex Character Anatomy: In complex multi-subject scenes, run at least 2–3 seed iterations or isolate subjects with regional inpainting to prevent limb or attribute bleeding.
Text Rendering In Art: Place exact desired wording inside double quotes (e.g. "Neon Sign"). Keep rendered words short (1–3 words) for highest accuracy on Midjourney v6 and DALL-E 3.
Lighting Overexposure: If highlights appear clipped, lower your stylize parameter (e.g. --s 150) or append diffused volumetric ambient lighting.