Regional Prompter + ADetailer
Last updated
Last updated
Regional Prompter is an advanced tool that allows creators precise control over their images by dividing a canvas into multiple areas called "regions." Each region receives its own unique prompt, enabling users to define exactly what content appears in each section.
Common uses:
Clearly detailing multiple characters without overlapping attributes.
Creating complex, dynamic scenes.
In Regional Prompter, BREAK is an advanced command for Stable Diffusion used to switch prompts between regions. On Moescape UI, BREAK is automatically added, so manual entry isn't required.
Figure 1a: Shows two equal column regions, illustrating no need to manually enter BREAK.
Figure 1b: After selecting โContinue (2 regions),โ prompts appear separated by BREAK (Common Prompt, 1st Region Prompt, 2nd Region Prompt).
Outside Regional Prompter, BREAK performs an advanced function related to tokens and chunks, suitable only for experienced users.
ADetailer is a Stable Diffusion UI extension designed to detect, mask, and inpaint details automatically. It act like targeted enhancement tools, commonly used for improving faces, hands, and eyes, with additional models available for other anatomical details.
Regional Prompter can be effectively combined with ADetailer, especially when dealing with multiple faces in an image.
Two primary ways to define prompts are Common and Base:
Common Prompts: Apply a uniform prompt across all regions, streamlining creation and minimizing redundancy.
For example, instead of repeatedly specifying "a man and woman," like:
a man and woman, a man with black hair BREAK a man and woman, a woman with green hair
You can simplify it like:
a man and woman BREAK a man with black hair BREAK a woman with green hair
Base Prompts: Similar to Common prompts but allow adjustment using a base ratio. This offers control over how prompts influence the overall image, enabling different weights in specific areas.
Choosing between Common and Base prompts is essential. Typically, Base prompts are best for specific number of characters (like couples). While both can be enabled, complexity can quickly increase.
Model: Nova Orange XL
ADetailer: Toggled ON
Regional Prompter Settings:
Layout Preset: 3 Equal Columns
Divide Ratio: 1,1,1
Split Mode: Columns
Common Prompt:
masterpiece, three girls, best quality
Negative Prompt: (default negative prompts used)
Prompt 1:
a girl, focus face, ((multi-colored hair, rainbow hair)), long hair, smiling, cap, open mouth, white shirt, painting apron, hold paintbrush, blue eyes, pink eye, closed one eye, heterochronia, standing
Prompt 2:
a girl with short bobcut hairstyle, disconnected straight tips, short blunt bangs, high side hair bun, light green short hair, orange eyes, vibrant playful look, orange star jacket, long sleeves, full body
Prompt 3:
an angel with gold halo, blue eyes, midnight blue hair, two angelic wings
Settings: (default settings used)
Sampling Method: Euler Ancestral
Steps: 25
Scale: 6.5
Canvas Size: Panorama (1216 X 832)
Hires.fix Settings: (default settings used)
Upscale By: 2
Hires Steps: 10
Denoise Strength: 0.2
Upscaler: R-ESRGAN 4x Anime6B