● tokenstore

AI Film Production via API: Multi-Shot Scenes, Frames, and Continuity

The Nureta video API collapses the film stack into parameters — multi-shot scenes, per-shot first/last frames, automatic character continuity — and it's adult-native.

2 min read

Film production is a logistics problem: cast, crew, camera, continuity, edit, sound. The Nureta video API turns each of those into a parameter. You stop hiring the pipeline and start passing it as JSON — and unlike every mainstream film-gen API, this one renders adult scenes too.

Casting is a reference image

Your lead is a reference image in the content array (up to nine images per task). The API holds that identity across every shot of a stitched scene, so your performer does not turn into a different person at the cut. Setup is in reference images for consistent faces.

A scene is a duration

Ask for any integer from 16 to 180 seconds and the API renders a stitched multi-shot scene instead of a single clip — continuity handled for you. That is the jump from clip to cinema, on one field. The mechanics are in multi-shot adult scenes up to 3 minutes.

Direct the cut with first/last frame

Pin how a shot opens or closes with role: "first_frame" / "last_frame" on an image item. A pinned frame renders that shot frame-conditioned — the shot's identity comes from the frame itself. It is the API equivalent of setting your camera's first and last mark; details in first and last frame control.

curl -s -X POST "$TOKENSTORE_URL/api/v3/contents/generations/tasks" \
  -H "Authorization: Bearer sk-..." -H "Content-Type: application/json" \
  -d '{
    "model": "seahorse-720p", "duration": 60, "ratio": "9:16",
    "content": [
      { "type": "text", "text": "a slow bedroom scene, one continuous character" },
      { "type": "image_url", "image_url": { "url": "https://.../lead.jpg" } }
    ]
  }'

Per-shot control when you want it

Drop into segments and you are the shot supervisor: per-clip script, duration_seconds, frames, and the explicit flag. Or hand it to the director in mode: "auto" and only intervene where you disagree. The whole console is in the video API docs.

Sound arrives with the frame

generate_audio defaults on, so shots come back with sound instead of a separate foley pass — see adding audio to NSFW video. Post-production shrinks to selection and trimming.

The crew's-eye difference: it's adult-native

Adult film has always adopted production tech early — and it is exactly what mainstream film-gen APIs refuse. Nureta's explicit pipeline renders sex acts as the point of the shot, not a blur, so the full production collapse applies to the content other APIs lock out. See rendering explicit sex acts on video.

Production economics, rewritten

  • Billed per second of output in USDC — no crew day, no location fee.
  • Failed renders refund automatically, so a bad take is free.
  • Shoot a dozen variations before lunch; marginal cost of another take is near zero.

The crew did not get cheaper — it got absorbed into the request body. What is left is the only job that was ever the point: deciding what the audience sees.

Direct a scene in the playground →
AI Film Production via API: Multi-Shot Scenes, Frames, and Continuity — Nureta Guides