Summary  
This chapter covers how to pass multiple reference images as inputs to a generative video function in order to maintain consistent visual elements across frames.

General domain of usage  
AI-powered video generation

By default, a character's look can shift slightly between generations. **Ingredients to Video** fixes this: you upload up to three reference images — usually a character, an object, and a setting — and Veo blends them into one consistent clip.


**Ingredient** — a reference image (character, object, or style) that Veo uses to keep a visual element consistent across generations, instead of reinventing it from a text description each time.

Definition

**Steps:**
1. Generate or upload a clean, isolated reference image for your character.
2. Open "Ingredients to Video" in Flow or the Gemini app and add the image.
3. Write a prompt describing what the character does.


The first ingredient you add gets the most visual priority — put your main character first if elements compete for attention in the frame.


Note

AI video generation has moved from novelty to everyday creative tool — and Google Veo 3.1 is currently the most capable model you can access for free. This mini-course teaches you to go from a blank prompt box to a finished, publish-ready video clip. Across 10 short lessons, you'll learn to write prompts that actually work, generate built-in dialogue and sound effects, keep the same character consistent across multiple shots, and export the right format for YouTube, TikTok, or Instagram. No editing software or technical background required — just a browser and an idea.

Keeping Characters Consistent with Ingredients to Video