JRD_MaryGold_Style is a soft, painterly LoRA designed to bring gentle textures, smooth gradients, and expressive character rendering to illustration‑focused checkpoints. It enhances faces, shapes, and color transitions without overpowering the base model, making it ideal for character creation, portraits, and stylized scenes.

This LoRA works best on non‑photorealistic or illustration‑oriented checkpoints, where its subtle brushlike qualities and clean detailing can fully emerge. It preserves form and expression while adding a warm, cohesive finish that feels natural and consistent across different subjects.

For sampling, the LoRA performs reliably with Euler (with or without Karras), but reaches its highest stability and refinement with DPM++ 2M SDE Karras and DPM++ 3M. These samplers bring out smoother transitions, cleaner edges, and more controlled shading.

Recommended LoRA weight is 0.8 to 1.0, depending on how strong you want the stylistic influence to be.

Key Features

• Soft, painterly rendering with smooth tonal transitions

• Excellent for character design and expressive portraits

• Works best with illustration‑based checkpoints

• Stable performance across a wide range of prompts

• Strong sampler compatibility, especially with DPM++ variants

• Balanced influence at weights between 0.8–1.0