Video generation
$0.1960/video secondGrok Imagine Video 1.5 Reference-to-Video
Generate video from one to seven reference images and a tagged text prompt with duration and aspect-ratio controls.
Loading tools…
Generate video guided by multiple image, audio or other supported references.
Use verified tools with pay-per-use pricing. Provider access is included. Tools that work in your own apps ask you to connect that account.
Connect your first agent to receive $1 trial credit.
Video generation
$0.1960/video secondGenerate video from one to seven reference images and a tagged text prompt with duration and aspect-ratio controls.
Video generation
$0.0150/1k tokensGenerate video with duration, resolution and native audio controls.
Tell your agent what each reference is meant to control, such as identity, wardrobe, product detail, composition or visual style. Clear roles prevent the references from competing with one another.
Label the authoritative image for identity or product appearance and explain what to borrow from supporting references. State which differences are intentional and which details must remain consistent.
Use the smallest set that resolves the visual brief. Extra references can introduce contradictory faces, palettes, proportions or environments, so remove any image that does not control a specific quality.
Begin with a short action and restrained camera movement. Check identity, object geometry, wardrobe, style and background continuity before attempting a complex sequence.
This example gives two references distinct jobs.
Example prompt to give your agent
Use Scrollport's Reference-to-video capability to create a [DURATION] shot for [USE]. Use [REFERENCE A] as the authority for [IDENTITY OR PRODUCT] and [REFERENCE B] only for [COMPOSITION OR STYLE]. Show [ACTION] in [SETTING] with [CAMERA MOVEMENT]. Preserve [DETAILS], ignore [CONFLICTING REFERENCE DETAILS], and show me the price before the paid run.