Melodious AI
Make a video

Photos and @ mentionsBeta

Put yourself, a place or a look in your video. Add photos as characters, places and looks, and pull in saved assets by typing @.

Last updated

Photos tell the director who and what to keep consistent. Every keyframe is drawn with your reference in hand, so the same face holds from the first shot to the last.

Add photos

Upload

Press + then Upload photos and pick one or more. With the box focused, ⌘ I (Ctrl I on Windows) does the same.

Set each one's role

Each photo arrives as a Character. Tap its label to change it to a Place or a Look.

Start

Your photos carry into the project. The brief opens with them already filled in as your cast, locations and look.

Use what you've saved

Press + then Your assets to see your saved characters, places and looks, with a search and Browse all. Or type @ in the box to open the same picker without leaving the keyboard. A picked asset sits beside your photos with its role, and is never uploaded twice. Assets in the sidebar holds everything you've saved.

Limits

You can cast up to 4 characters and add up to 14 references to one video. Past 4 characters the model starts blending faces, so Melodious stops you rather than ship something inconsistent.

Photos that hold

Use a clear, well-lit, front-facing photo, one person per photo. A group photo is kept as a group: the model can't pull one person out of it.

You can also attach a photo to a single shot from its card. See The storyboard.

Was this page helpful?

On this page

We use analytics and support tools (GTM, Plausible, PostHog, Crisp) to improve Melodious AI. Manage this in Privacy Settings anytime.