Upload up to 9 reference inputs and steer the motion, framing, and style with stronger visual control.
Reference-Led Restyle
Keep the subject silhouette and composition, then restyle the shot into a neon editorial sequence
Product Direction Lock
Preserve the product form and materials while adding cinematic dolly motion and premium lighting
Character Visual Continuity
Keep the character identity and wardrobe cues, then generate a moody night-time city scene
Reference-guided motion works best when each input has a clear job instead of all of them trying to do the same thing.
Ready to guide motion with reference inputs?
The strongest results come from separating what must stay stable from what you want the model to reinterpret.
Reference-to-video works best when the prompt explains the role of each reference instead of treating them as a generic attachment.
Start with the reference that best captures the subject or product you need the model to keep intact.
Use additional images when you need better control over style, lighting, wardrobe, or secondary composition cues.
Explain what should move, what should remain fixed, and how much reinterpretation you are comfortable with.
Review the outputs and keep the version that balances reference fidelity with fresh motion the best.
You can upload up to 9 reference inputs, which is useful when you need multiple visual anchors instead of a single source image.
Use reference-to-video when one image is not enough to describe the subject, product, or visual language you need to keep consistent.
Yes. It is a strong fit when you want a character, wardrobe, or identity cue to stay recognizable across the new motion result.
Yes. It works well when product shape, material, and styling need to remain anchored while you change camera motion or scene mood.
