Three layers, in this order
Every Image and Video render is sent with three layers:
- The Studio's instruction. One line per model, written by us (for example, "one mark, lots of empty space" for Logo). You cannot turn it off.
- Your store. The skills, rules and connectors saved in your store's settings. They ride with every render, without you choosing them again.
- Your prompt. What you type this time. It is not saved as a skill.
Put the recurring part in a rule
If you type the same thing on every render ("no text in the image", "always our cream background"), it belongs in a rule in your store's settings. It will then apply to every render, and your prompt stays about this one.
A skill is a longer instruction for a kind of work. Your store's saved skills are read again on the server at each render, so an edit to a skill applies to the next render.
What is not sent
- Voice reads only the line you type. Skills and rules are instructions for a picture or a clip, not words to say aloud.
- The browser only sends which skills apply. Their text is read from your store on the server, so nothing outside your store can add an instruction to your render.