Referencing Multiple Input Images in Prompts
When your generation request contains two or more reference attachments (such as two different characters, a character and a weapon, or a character and a vehicle), use explicit @Image tags to instruct the AI model.
Tagging Syntax in Studio & Agent Mode
Type @ inside the composer to trigger the auto-complete dropdown, or use explicit tags:
@Image1: Refers to the first attached image.@Image2: Refers to the second attached image.@Image3: Refers to the third attached image.
Structuring Multi-Character Action Prompts
When two characters interact in a scene, define their roles clearly in the prompt:
Live-action color cinematic film. Two characters in an intense standoff. @Image1 (Detective Marcus) stands on the left in a rain coat, pointing a flashlight. @Image2 (Elena) stands on the right, looking back over her shoulder. Anamorphic lens flare, dark alleyway.Automatic [Characters] Section Formatting
Agent Mode automatically structures multi-image attachments into a dedicated [Characters] block before dispatching to the rendering cluster:
[Characters]
Character 1 - Image 1 - Male detective in trenchcoat (@Image1)
Character 2 - Image 2 - Female operative in black jacket (@Image2)