G1Joshi

midjourney

Midjourney AI image generation. Use for creative AI.

G1Joshi 12 3 Updated 6mo ago
GitHub

Install

npx skillscat add g1joshi/agent-skills/midjourney

Install via the SkillsCat registry.

About this skill

This skill enables high-quality AI image generation with advanced features like character consistency and style referencing. It solves the need for sophisticated aesthetic control and rapid visual ideation. Developers should use it when applications require artistic, high-fidelity imagery or complex editing capabilities like in-painting and out-painting.

SKILL.md

Midjourney

Midjourney is a closed-source image generator known for its artistic quality and distinct "Midjourney Look". v7 (2025) adds 3D modeling and higher photorealism.

When to Use

  • Marketing/Art: Unbeatable aesthetic quality out of the box.
  • Ideation: Rapidly generating mood boards.
  • Web Interface: Now usable via web (no longer Discord-only).

Core Concepts

Parameters

--ar 16:9 (Aspect Ratio), --stylize 100, --niji (Anime style).

Zoom / Pan / Vary

In-painting and out-painting features to edit generated images.

Character Reference (--cref)

Keep the same character face across different images.

Best Practices (2025)

Do:

  • Use the Web Editor: Much better than Discord for editing/masking.
  • Use Style References (--sref): Pass an image URL to copy its art style.
  • Use Permutations: Generate 4 variations of a prompt with {red, blue} syntax.

Don't:

  • Don't use for text: While v7 is better, SD3/Flux are often better for strict text rendering.

References