Sign In

Vete ┃ Ghostlight ILL

Updated: Feb 2, 2026

base model

Verified:

SafeTensor

Type

Checkpoint Merge

Stats

221

0

Reviews

Published

Nov 25, 2025

Base Model

Illustrious

Hash

AutoV2
0969CBCAA7
default creator card background decoration
Vetehine's Avatar

Vetehine

---

Creator: Vetehine

Discord: Vetehine

Linktree: https://linktr.ee/Vetehine

---

Please let me know what you think about this Model in the comments.

---

Vete ┃ Ghostlight ILL is a semi-realistic model focused on soft illumination, atmospheric depth, and controlled realism with an ethereal edge. It is designed to sit precisely between anime stylization and photoreal influence, producing images that feel luminous, calm, and emotionally grounded without drifting into hard realism or exaggerated anime traits.

This model emphasizes natural light behavior. Highlights are gentle and diffused, shadows remain soft and readable, and midtones are preserved to maintain depth and skin realism. Glow is used subtly and intentionally, creating a quiet, ghostlike presence rather than strong neon or harsh contrast. Light wraps smoothly around faces, hair, fabric, and environmental elements, giving scenes a floating, cinematic quality.

Ghostlight ILL excels at clean facial structure, accurate anatomy, and stable proportions. Skin rendering is smooth and natural, avoiding waxiness while keeping a soft finish. Eyes remain expressive and luminous without oversharpening. Hair maintains separation and shine under varied lighting conditions, from daylight softness to night scenes and low-key illumination.

Stylistically, the model is highly flexible. It performs well in calm interiors, soft fantasy, winter scenes, gentle gothic moods, night city environments, and emotionally focused portraits. Color handling stays neutral and balanced, allowing cool blues, warm light, whites, and dark tones to coexist without color bleed or bias.

Ghostlight ILL can be used directly as a standalone model for polished semi-realistic output, or merged as a refinement layer to introduce softness, light control, and atmospheric clarity into other bases. It is stable, clean, and intentional, built for creators who want realism without harshness and glow without noise.

---

Recommended Danbooru Tags

For Soft Semi-Realistic Output

(semi_realistic:1.2), soft_lighting, diffused_light, natural_shadows, gentle_glow, detailed_face, realistic_skin, smooth_shading, cinematic_light, depth_of_field, subtle_bokeh, glossy_hair, clean_render, highres

---

For Night and Atmospheric Scenes

soft_rim_light, low_key_lighting, ambient_light, cool_tone, volumetric_light, light_fog, reflective_surfaces, calm_mood, cinematic_depth

---

AI Image Generation Settings Guide

https://moescape.ai/posts/ai-image-generation-settings-guide

---

Recommended Settings -

Clip Skip - 2

DPM++ 2M Karras

Sampling Steps - 25-50 (Recommended: 37 Steps)

Cfg - 6-7

---

Recommended Settings -

Clip Skip - 2

Euler Ancestral (Recommended: 30 Steps)

Sampling Steps - 25-35

Cfg - 6-7

---

Recommended Settings -

Clip Skip - 2

Restart

Sampling Steps - (Recommended: 66 Steps)

Cfg - 6-7

---

Hires-

R-ESRGAN 4x+ Anime6B, 4x-AnimeSharp

Steps 10 - 20

Denoising 0.2 - 0.3

---

Note: Using LoRAs/Spells may affect model performance or stability. Apply them with your own settings and discretion.

---