Tem um workflow do ComfyUI? Gere uma documentação legível direto no navegador.Torne um workflow legível.Experimentar agora ↗
I really loved the soft matte coloring and expressive illustration style associated with Secret Sauce.
Fonte do modelo
Descrição da fonte
✨ 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:
Fontes
1 fonteArtefatos de modelo
2 artefatosTrechos de fonte
1 trechoSource context: 184 downloads · Type Checkpoint · Base model Illustrious
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
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
