opendunes
Video
Copy page ▾

Picture to video

Make a clip that opens on a picture you give, or moves from one picture to another.

  1. Pick a model that takes your frames.
  2. Send the job with a first frame.
  3. Add a last frame for a transition.
  4. Wait and fetch the clip.

1. Pick a model that takes your frames

A model's supported_frame_images on the video models list names the frames it takes: first_frame, last_frame, or both. A model takes only the frames it lists. Which models take which, read 2026-09-29: Choose a video model.

You havePick a model listing
one picture the clip opens onfirst_frame
two pictures, the clip moving from one to the otherfirst_frame and last_frame

2. Send the job with a first frame

frame_images carries the picture and where it goes. The sample opens a 5-second 720p clip in 16:9 on a picture of sand in golden light, on alibaba/wan-3.0, which takes a first frame. The prompt says what happens after that frame.

import json
import os
import urllib.request

API_KEY = os.environ["OPENDUNES_API_KEY"]
ADDRESS = "https://opendunes.com/v1/videos"
MODEL = "alibaba/wan-3.0"
FIRST_FRAME = "https://opendunes.com/hero-dunes.webp"
PROMPT = (
    "The golden haze settles into sand dunes at dusk, wind lifting fine "
    "sand off the crests, slow pan right"
)

request = urllib.request.Request(
    ADDRESS,
    data=json.dumps(
        {
            "model": MODEL,
            "prompt": PROMPT,
            "duration": 5,
            "resolution": "720p",
            "aspect_ratio": "16:9",
            "frame_images": [
                {
                    "type": "image_url",
                    "image_url": {"url": FIRST_FRAME},
                    "frame_type": "first_frame",
                },
            ],
        }
    ).encode(),
    headers={
        "Authorization": "Bearer " + API_KEY,
        "Content-Type": "application/json",
    },
)
with urllib.request.urlopen(request) as response:
    job = json.load(response)

print(job["id"], job["status"])
In each frame_images entryHolds
typeimage_url for a picture at an address, file for a picture you stored
image_url.urlthe picture's address: public HTTPS, answering with the picture itself
file.file_idthe ID of a picture stored in Files and storage
frame_typefirst_frame or last_frame: where the picture goes

The answer is 202, with the job's id and status pending. prompt may be left out when the picture carries the whole request. Every field: Clips.

Watch out

A frame takes the picture only from a model that lists that frame. Check supported_frame_images before you send a last_frame: alibaba/wan-3.0 lists first_frame alone.

3. Add a last frame for a transition

With a first and a last frame, the model makes the motion between them. Send two entries to a model that lists both, such as bytedance/seedance-2.0-fast: the second can be a picture you stored, by its file_id. The body, with your two pictures in place of …:

JSON
{
  "model": "bytedance/seedance-2.0-fast",
  "prompt": "The shop's shutter rolls up on a morning in Oran",
  "duration": 5,
  "resolution": "720p",
  "aspect_ratio": "16:9",
  "frame_images": [
    {
      "type": "image_url",
      "image_url": {"url": "https://…/shutter-down.jpg"},
      "frame_type": "first_frame"
    },
    {
      "type": "file",
      "file": {"file_id": "file_…"},
      "frame_type": "last_frame"
    }
  ]
}

At most two entries: one first frame, one last. A file_id that is not yours, has expired, or is not a picture gets a 422, and no job is made.

Good to know

A frame fixes the picture exactly. To guide the look of a clip without fixing a frame, send pictures in input_references: Reference pictures. When both are sent, frame_images wins.

4. Wait and fetch the clip

Ask GET /v1/videos/{id} every ten seconds until the job ends, then fetch the clip from /v1/videos/{id}/content?index=0 with your API keyAPI keyThe key you create and send with every request that calls a model.More on API keys →. The whole loop, in four languages: Text to video. To have the end posted to you instead: Video webhooks.

The clip is charged in DADAThe Algerian dinar, the one currency the docs show.More on Balance and charges → once, when the job reports completed. On some models a clip from a picture has its own rate, and on others each picture sent in is charged: the model's page in Models lists what it charges for. A job that ends failed, cancelled or expired charges nothing.

Last modified on