KonoSuba

God's Blessing on This Wonderful World!

Authors: Juusoz, 金Goldkoron, Khalil, LightUK and SamuraiCat

Recommended Settings

This model performs best with the following settings:

• Image Size

  1024x576 for wide 16:9, 768x768 for square, and 640x1024 for portrait

  Feel free to experiment with higher resolutions, Juusoz made all the examples at higher than recommended resolutions

• CFG

  9-12

• Clip Skip

  1

• Steps

  +30 minimum, +70 can give nice results

Prompt Format

The prompts we trained on are available in the corresponding T{version} Konosuba prompt list.txt file.

T2.11 KonoSuba prompt list.txt, T2.8 KonoSuba prompt list.txt

Format: Parent tag, description prompt 1, description prompt 2, etc;

Example: Megumin, holding staff, casting, close-up; sparkle; night;

The parent tag serves as the primary label for the overall theme of the image and acts as the main subject. The description prompts are a comprehensive list of attributes, objects, actions, and other elements that describe the subject in detail. Each subject is separated by a semicolon (;) while the individual attributes, objects, actions, and elements are separated by a comma (,).

Just because we haven’t trained on something doesn’t mean the base AI model doesn’t already know what it is, so feel free to get creative and try new things!

Socials

Website | Discord | Patreon