A Set of Images That Belong Together

A Set of Images That Belong Together

Locking the parts that must not vary so the variation lands where you want it.

leonardo-ai image consistency seed

What it is for

A set that belongs together is made by deciding what must NOT vary before generating anything. Lock those, and the variation you do get lands where you wanted it instead of arriving everywhere at once.

The skill

A set falls apart because everything varies at once. Fix most of it.

Lock, and do not touch between images:
- **The seed** while you are still tuning the prompt. Changing prompt and seed
  together tells you nothing about which change did what.
- **The model and the preset.** Switching either is a different set.
- **The lighting clause**, word for word. Light direction is what the eye reads
  as "same place".
- **The aspect ratio.**

Vary one thing per image: the subject, or the action, or the angle. Not two.

The prompt structure that holds a set together:

    <subject>, <action>, | <fixed style block> | <fixed light block> | <fixed lens block>

Write the three fixed blocks once and paste them unchanged. Rewriting them from
memory each time is why sets drift — a synonym is a different prompt.

When one image in a set is wrong, regenerate it with the same seed and a single
changed word. Regenerating with a new seed gives you a good image that does not
belong.

How to install it

1. Open the tool and start a new conversation, project or template.
2. Paste the text below wherever the tool keeps standing instructions
   — usually called system prompt, instructions, or persona.
3. Save it so it applies to every run rather than a single message.

Did it work for you?

Where it came from

Skills app Original