Reference to Video Generator

What Is Reference to Video?

Reference to video is a multimodal AI video workflow: upload the images, short clips, and optional audio that define your intended subject, style, motion, or sound, then add a prompt that directs the final shot. Story321 uses MiniMax H3 Reference to Video and requires at least one image or video reference, so a soundtrack alone cannot create a clip.

Reference to Video Features

• Use three kinds of creative reference

  Reference to video accepts up to nine images, three video clips, and three audio clips in one request. Combine a product image, a camera-motion sample, and a sound reference when each asset has a clear job.

• Keep output settings deliberate

  Choose a 5–15 second result, 768P, 2K, or 4K output, and adaptive, landscape, portrait, square, or ultra-wide framing. The default reference to video request is a five-second 2K adaptive clip.

• Prompt expansion is on by default

  Describe the intended action, timing, camera behavior, and details that must remain stable. Prompt expansion is enabled by default to help turn a concise production direction into a fuller generation instruction.

• Safety checking stays enabled

  The reference to video form always sends the provider safety setting enabled. This cannot be disabled from the browser or by directly calling the Story321 generation route.

• Upload before generation

  Story321 uploads local reference files before the request reaches MiniMax H3. The generation payload contains only the supported reference URL arrays and output settings.

• Credit total updates with images

  The first five image references are included. A sixth and each following image reference adds 18 credits, so the reference to video price remains clear before you generate.

Why Use a Reference to Video Workflow?

• Start from approved visual direction

  Use an approved keyframe, product packshot, character board, environment image, or editorial still when reference to video needs a more intentional starting point than text alone.

• Separate what each asset controls

  An image can establish identity and art direction, a short video can suggest motion or pacing, and audio can communicate rhythm or atmosphere. Give each reference to video asset one clear purpose.

• Iterate without replacing the whole brief

  Keep strong references fixed and revise one prompt instruction, duration, framing choice, or single supporting asset at a time. This creates a practical review loop for reference to video tests.

• Prepare versions for different placements

  Create a vertical social concept, a 16:9 presentation shot, or an adaptive crop from the same source direction. Select the output shape before generation instead of relying on a later crop.

How to Create a Reference to Video Clip

1. 1

   Choose the assets that matter

   Add one to nine images, up to three video references, and up to three audio references. At least one image or video is required for reference to video; do not use audio by itself.

2. 2

   Check media limits

   Each video and audio reference must be 2–15 seconds. Video references together can total no more than 15 seconds, and audio references together can total no more than 15 seconds. Keep all reference files to twelve or fewer.

3. 3

   Write one connected shot direction

   Name the subject, describe its action, identify the setting and light, then state camera movement and pacing. Explain what from the reference to video assets must stay recognizable rather than listing unrelated ideas.

4. 4

   Set output and generate

   Choose 5–15 seconds, 768P, 2K, or 4K, plus your aspect ratio. Review subject consistency, motion, framing, and sound relationship, then refine the next reference to video attempt.

Reference to Video Use Cases

• Product launch motion

  Use packshots and material references to direct a product reveal, then use a short motion sample to describe a camera push, turntable move, or macro detail.

• Character and fashion tests

  Provide an approved look reference and a motion example to test one walk, turn, gesture, or lighting change before a larger production.

• Campaign social cutdowns

  Use reference to video to make a short vertical concept from brand imagery, approved colors, and an audio mood reference for a social review.

• Storyboard and pitch frames

  Turn selected visual-development frames into a moving scene test. Keep the prompt focused on one shot so reviewers can evaluate the idea quickly.

• Music and atmosphere concepts

  Pair image or video direction with a short audio reference when rhythm, ambience, or a sound-world cue should influence the concept. Audio supports reference to video but never replaces a visual reference.

Reference to Video Limits and Good Practice

• Visual input is mandatory: reference to video requires at least one image or video, and audio-only requests are blocked.
• Reference counts are fixed: the form allows 0–9 images, 0–3 videos, 0–3 audio files, and a maximum of 12 total reference files.
• Clip lengths are checked: every video or audio reference must last 2–15 seconds; each media type has a 15-second combined limit.
• Use assets you can use: only upload material you have permission to use and avoid protected characters, brands, or a person without necessary rights.

Continue Your Video Workflow

• Image to Video

  Animate a single image when one starting frame is all you need.

  Explore Image to Video

• Text to Video

  Generate a scene from a written direction without upload references.

  Explore Text to Video

• Story321 Pricing

  Review available credits before planning a set of reference to video tests.

  View Pricing

Frequently Asked Questions

Reference to Video Pricing

Reference to Video Specifications