Skill3 steps

Consistent Character

Keep the same face, outfit and style across every image.

How it runs

The right model for every step.

Give the skill one to four reference images and a list of scenes. It locks the identity of the character, generates each scene with the same face, wardrobe and styling, and checks the likeness before returning the set.

Every intermediate result is kept with the run, so you can inspect a step, reuse its output or rerun it with different settings.

Input
- One to four reference images
- Scene prompts
01
Learn the characterFLUX.2 [pro]
02
Generate each sceneFLUX.2 [pro]
03
Check the likenessFace match
OutputA matched image set, one per scene

Run it

Call the skill by name.

Send the inputs, get back a run you can wait on or follow with a webhook. The same skill is available in the studio and to AI agents through the MCP server.

You get back

A matched image set, one per scene

run.sh

curl https://api.monogram.dev/v1/skills/consistent-character/runs \
  -H "Authorization: Bearer $MONOGRAM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "reference_urls": [
        "https://files.example.com/ava-1.jpg",
        "https://files.example.com/ava-2.jpg"
      ],
      "scenes": [
        "Reading in a deco library",
        "Stepping off a 1930s train"
      ]
    }
  }'