"The character from this image", "the tone of that voiceover on the left" — those need something you can actually point at. DOTV's input bar gives you two ways to point: reference images (the whole image goes to the model) and @ mentions (naming a node inside your sentence) — consistent characters and a consistent style in an AI video come down to carrying the right references into every shot. This section covers how to use both, what you are picking when you click on the canvas, and how "Image 1 / Image 2" line up.