Skip to content
  1. Developers
  2. Models
  3. Video generation
Built by VidmoatAvailable

vidmoat-image-live-1

Our own image-to-video model. A router sends camera moves and region cinemagraphs to fast depth-parallax and flow pipelines, and faces, bodies, vehicles and anything described in words to a generative tier we run on our own GPU. Every clip is checked so nothing appears that the still does not show.

id vidmoat-image-live-1

What it does well

  • Moves only what the picture already shows: no invented people, hands, animals or vehicles.
  • Real depth parallax on push, pull, drift, orbit and pan.
  • Cinemagraphs where one region moves and everything else holds still, loopable.
  • quoteOnly tells you the tier and exact price before anything is charged.

Limits

  • Up to 10 seconds (5 on the generative tier), 720p on the short side, no audio.
  • Things appearing, pouring, breaking or transforming are refused: use a video model for those.
  • The generative tier takes about 2 to 5 minutes and has a monthly GPU allowance.
  • One living image at a time per account.

Where to call it

REST APIPOST /v1/ai/live-imagesMCPgenerate_live_image
Editor agentgenerateMedia kind "live"

Try it

curl
curl -X POST https://api.vidmoat.com/v1/ai/live-images \
  -H "Authorization: Bearer $VIDMOAT_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image":{"url":"https://example.com/lake.jpg"},"prompt":"slow push-in, the water ripples","durationSec":5}'

A test key answers with a sample and spends nothing. Every parameter and error is in the API reference.

Read more