Adding Reference Images

Reference images are one of Video Series' most important features. They solve the core challenge of multi-scene AI video: keeping characters, products, and visual elements consistent from one clip to the next.

When you upload a reference image in the “Carried through every scene” section, Qolaba's AI analyses it, generates a detailed text description, and injects that description into every scene's generation prompt. This means your character's face, clothing, and build, or your product's shape, colour, and branding, remain visually consistent across all clips.

Feature	Detail
Multiple references	Upload several images, one for a character, one for a product, one for an environment. Each gets its own label and AI-generated description.
Editable labels	Name each reference (e.g., “Sam”, “Classic watch”, “Product”). Labels appear in scene prompts for easy identification.
AI-generated descriptions	The system automatically writes a detailed visual description. You can edit it to emphasise or de-emphasise specific features.
Set once, applied everywhere	Reference images are set at the series level, not per scene. Every clip in the series inherits them automatically.
Source flexibility	Upload from your computer (up to 20 MB), or choose from your Qolaba image generation history (All / Generated / Uploaded tabs).