Deep_Shadows is a fast, high-contrast SDXL checkpoint distilled for 4-step LCM Simple generation. It is built to deliver rich, deep blacks and strong shadow separation without sacrificing fine detail, even at very low step counts.

Unlike many turbo-style models that flatten lighting or crush blacks into featureless voids, Deep_Shadows is intentionally biased toward dramatic tonal range. Shadows retain internal structure and texture, while highlights stay controlled and crisp. The result is imagery with weight and atmosphere — dark clothing folds, under-exposed interiors, night scenes, and chiaroscuro portraits all benefit from this characteristic.

Strengths

• Deep, readable shadows: Black areas hold micro-detail instead of turning into solid voids. Excellent for moody lighting, low-key portraits, and cinematic scenes.

• High contrast with clarity: Strong separation between light and shadow without harsh clipping or loss of midtone information.

• 4-step efficiency: Designed specifically for LCM Simple at 4 steps, making it practical for rapid iteration, batch work, or resource-constrained setups while still producing coherent, detailed images.

• Texture preservation: Skin, fabric, stone, metal, and natural surfaces keep visible grain and structure even under aggressive lighting.

• Stable at low steps: Minimal composition breakdown or anatomical collapse compared to many other low-step SDXL models.

Recommended Settings

• Sampler: LCM r Simple

• Steps: 4 (primary target)

• CFG: 1.0–2.0 (lower CFG generally preferred; higher values can still work but may reduce the model’s natural shadow character)

• Resolution: Standard SDXL ranges (1024×1024, 896×1152, 1152×896, etc.)

• Works well with contrast-enhancing prompts that mention chiaroscuro, dramatic lighting, deep shadows, low-key, rim light, or dark atmosphere.