character-consistency-prompt

Character Consistency Prompt

What This Skill Does

Writes a set of Midjourney prompts designed to maintain consistent character appearance — face, build, clothing, and presence — across multiple generated images for use in editorial series, podcast artwork, YouTube thumbnails, or visual narratives.

When To Use This Skill

• You are creating a series of images that all feature the same fictional character (for a podcast cover, YouTube series, editorial illustration run)
• You need a character to appear in multiple different scenes or contexts without changing their appearance
• You are illustrating a newsletter, blog, or social media series with a recurring visual persona
• You have already generated a character you like and want to generate additional images of the same person in new situations

What You Need To Provide

Required:

• A detailed description of the character: apparent age, gender presentation, build, distinctive physical features (hair, facial features, any identifying marks or accessories)
• The character's typical clothing or visual identity
• The style you want the images to be in (photorealistic, illustrated, graphic, painterly)

Optional:

• A reference image description (if you have already generated a version of this character you want to match)
• The scenes or situations you need them depicted in (list the images you need)
• Midjourney version (v6 has better face consistency than earlier versions; --cref flag is available in v6 for image-based character reference)
• Aspect ratio for intended use
• Any seed number from a prior successful generation

How the Assistant Approaches This

1. Builds a dense, precise character anchor string — a consistent block of descriptive text that will be used verbatim in every prompt in the set, covering the character's appearance in enough detail that Midjourney has minimal room for variation
2. Identifies the key consistency anchors: face structure, hair, signature clothing item or accessory — these are what must appear in every image
3. Writes a base prompt for the first image, then generates scene-specific prompts that keep the character anchor string identical and vary only the environment, action, and lighting
4. Recommends the --seed workflow: generate the first image, note the seed number from the Job ID, and use --seed [number] in subsequent prompts to bias toward the same character rendering
5. If Midjourney v6 is being used, notes where the --cref [image URL] flag should be inserted to use an existing character image as a visual reference
6. Closes with a "Next Step" note: which prompt to run first (always the base portrait), how to record the seed number after the first generation, and whether to run cinematic-shot-prompt if the character also needs to appear in more complex scene compositions