Reference to Video

Give the model more than a prompt—attach images, video, and audio so motion and identity stay on brief.

Recast Meme’s video studio is reference-aware. Depending on the model, you can attach images, clips, and audio to steer the generation instead of hoping a prompt alone holds the look.

This is the technical backbone behind trend remix: keep what matters from the references, then generate a new version for your cast or product.

How it works

  1. 1

    Pick a model that accepts your refs

    Check image, video, and audio limits in the studio for the model you choose.

  2. 2

    Attach references

    Add the plates that define likeness, motion, or sound.

  3. 3

    Prompt lightly and generate

    Describe what should change. Let references carry what should stay.

Multi-modal refs

Images for identity, video for motion cues, audio when the model supports it.

Less prompt wrestling

References reduce how much motion and look you must describe in text.

Built for remix

Same path you use when recasting a trend with your own assets.

FAQ

What is reference to video?
It means the generator conditions on media you provide—not only a text prompt—so outputs follow your subject or motion cues more closely.
Do all models support video and audio refs?
No. Each model lists what it accepts. Switch models if you need a specific reference type.
Is this the same as Recast?
Recast is the product story—trend remix with your cast. Reference to video is the studio capability that makes that remix controllable.

Start on Recast Meme

Generate from the studio, or browse what's trending and recast it.

Related