● tokenstore

Pinning First and Last Frames for Controlled NSFW Clips

Pin the opening or closing composition of an NSFW clip with role first_frame / last_frame. Frame-conditioned renders ignore reference images — use a frame OR refs, not both.

3 min read

Auto mode picks its own opening and closing compositions. When you need the clip to start or end on an exact image — a specific pose, a framing, or a frame carried over from another clip — pin it. Nureta lets you nail down the first frame, the last frame, or both.

Frame pinning is the most direct control you have over composition, and it comes with one hard constraint worth understanding before you use it.

What frame pinning does

On any image_url item, set role to first_frame or last_frame. The clip then opens (or closes) on that exact still, and the render is frame-conditioned — the motion is generated to flow out of, or into, the composition you pinned. Pin both ends of the same clip to control the whole arc from opening pose to closing pose.

Pinning the opening

role: "first_frame" makes your image the literal first frame; everything after it animates forward. Use it to start on a precise setup — a specific position, or a framing you already rendered as a still. This is the image-to-video path when you want the clip to begin exactly where your picture does; see image-to-video.

Pinning the close

role: "last_frame" fixes where the clip lands. The motion is generated to arrive at that composition on the final frame. Handy when you know the pose you want to end on — a climax framing, a specific resting position — and want the model to build toward it rather than wander.

A frame ignores reference images

A pinned frame is mutually exclusive with reference images. Send a frame and the request renders frame-conditioned: role-less reference images and preset assets are ignored. The pinned shot's identity comes from the frame image itself — so pin a frame that already shows your character. Send a frame OR reference images, never both.

The practical consequence: if identity matters and you are pinning a frame, the frame has to carry that identity, because your reference images will not. Either pin a frame that already shows the right person, or drop the frame and let reference images hold identity instead.

Chaining clips together

Frame pinning is how you get continuity across separate clips. Take the last frame of clip A and pin it as the first_frame of clip B, and the two cut together seamlessly — B picks up exactly where A left off. Repeat to build a longer sequence by hand, clip by clip. For an automatic multi-shot scene that stays consistent without hand-chaining, use a stitched scene instead.

Example

Open the clip on a pinned still and animate forward. Note there are no role-less reference images here — the frame is the sole identity source.

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": 8,
    "content": [
      { "type": "text", "text": "she rises off the pillows and moves toward the camera, nude, slow" },
      { "type": "image_url", "image_url": { "url": "https://.../opening.jpg" }, "role": "first_frame" }
    ]
  }'

Tips

  • Pin a frame that already shows your character — the frame is the only identity source for that render.
  • Pinning both first and last frame gives the tightest control over the clip's arc.
  • To chain clips, export the final frame of the previous clip and reuse it as the next first_frame — that is the join.
  • Need identity across many shots without hand-chaining? A stitched scene handles it for you.

The role field and the frame-versus-reference rule are documented in the generation API docs.

Open the playground →
Pinning First and Last Frames for Controlled NSFW Clips — Nureta Guides