Skip to content

Image-to-Video Generation

Image-to-video models take a still frame and generate motion from it — turning a photo into a short animated clip. This is useful for animating product shots, hero images, b-roll stills, or any frame where you want to add movement without shooting new footage.

Click on a still image clip on your Premiere Pro timeline, or select an image in the Project Bin. modelBridge detects the selection and shows the clip information in the media card — including dimensions, file size, and media type.

Search for an image-to-video model using the model selector. Some popular options:

ModelStrengths
Kling v3 ProHigh quality motion, consistent with source
Wan 2.6Good balance of quality and speed
LTX 2.3Fast generation, lower cost
SeedanceStrong motion quality

Use the filter chips to show only image-to-video models.

The prompt describes what motion you want the AI to add to your still image. Focus on movement, not on describing the image itself — the model already sees the image.

Good prompt:

“Slow zoom in, hair blowing in the wind, subtle eye movement, cinematic”

Less effective prompt:

“A woman standing in a field” (describes the image, not the motion)

Adjust duration, resolution, and aspect ratio. The cost estimate updates live as you change these values.

  • Duration — longer clips cost more and take longer to generate
  • Aspect ratio — match your sequence settings (16:9, 9:16, 1:1)
  • Resolution — higher resolution increases quality and cost

The media card shows real-time validation status:

  • Green border — your image meets all model requirements
  • Red border — something is wrong (image too small, wrong format, etc.) with a specific error message

If validation fails, the Generate button is disabled and the error message tells you exactly what to fix.

Click Generate. When the result is ready:

  • Import to Timeline replaces the source image on the timeline at its exact position and scale
  • Save to Project Bin imports without timeline placement
  • Preview in Source Monitor lets you evaluate before committing

The original image stays in your Project Bin — nothing is destroyed.

modelBridge checks your selected image against the model’s requirements before any API call:

  • Minimum dimensions — e.g., “Image too small (640x480, min 1024x768)”
  • Maximum file size — e.g., “File too large (15 MB, max 10 MB)”
  • Aspect ratio — some models require specific ratios
  • File format — supported formats vary by model

These checks are instant and prevent wasted API credits. If a model rejects your image for a requirement not declared in its schema, the plugin learns that requirement and catches it automatically on future attempts.

Click the sparkle button below the prompt field to have AI rewrite your prompt for better results. This costs approximately $0.01 and is useful when you are not sure how to describe the motion you want.

Some models take more creative liberty than others. If the result diverges too much from your source, try a different model or add “maintain source appearance” to your prompt.

Image-to-video models with audio enabled or high resolution can take 5–15 minutes. The generation moves to the background automatically — check the Active Generations panel for progress.

Check the error message on the media card. Common causes: image too small, unsupported format, or file too large. The message tells you the exact requirement.