✨ About the model

I really loved the soft matte coloring and expressive illustration style associated with Secret Sauce.

However, since the original model was Pony-based, I often encountered compatibility issues, visual artifacts, inconsistent anatomy, and especially weak or overly feminine male characters when using it in my usual Illustrious workflow.

So I decided to recreate the qualities I enjoyed most as a new style model.

I trained the style on a carefully curated dataset and then refined the result through several LoRA and checkpoint experiments. The final model combines:

- soft matte digital painting

- clean comic-inspired linework

- expressive faces and eyes

- smooth, fluid body shapes

- better male anatomy and facial structure

- strong female and masculine character generation

- good support for anime, anthro, furry, fantasy, and cute art

- detailed clothing and accessories without overly harsh rendering

I am happy to finally present SauceCraft XL. 💙

This model is designed to retain the softness and charm of the style that inspired it, while providing better versatility, cleaner anatomy, and more reliable results on Illustrious-based generation.

---

🎨 What it does well

- Adult women and men

- Masculine faces and bodies

- Anime and semi-anime characters

- Anthro and furry characters

- Fantasy characters

- Portraits and full-body illustrations

- Cute and kawaii artwork

- Soft NSFW and pin-up compositions

- Clothing, uniforms, armor, and accessories

- Matte coloring with clean, readable shapes

The model can shift between cute, elegant, sensual, comic-like, and more serious illustrations depending on the prompt.

---

⚙️ Recommended settings

Sampler: Euler a

Scheduler: Automatic

Steps: 30–40

CFG: 4.5–6

Recommended resolution: around 896×1152 or other SDXL-compatible resolutions

### Hires.fix

Upscale: 1.5×

Hires steps: 12–18

Denoising strength: 0.30–0.40

Hires CFG: around 4–5

Upscaler: 4x-ClearRealityV1 Soft or another soft illustration upscaler

ADetailer: recommended for portraits, close-ups, or complex compositions.

My usual starting setup:

Sampler: Euler a

Steps: 35

CFG: 5

Resolution: 896×1152

Hires.fix: enabled

Upscale: 1.5

Hires steps: 15

Denoising strength: 0.35

Hires CFG: 4.5

Upscaler: 4x-ClearRealityV1_Soft