Generate By Image And Description

Gen-4 image-to-video is now ready!
Just change the model from "gen3" to "gen4" and you’re good to go.

Please note: the source image should be either 1280x720 or 720x1280 (in this case, set "flip" to true).

Example body:

{
  "text_prompt": "ocean",
  "img_prompt": "https://files.aivideoapi.com/example/ocean.jpg",
  "model": "gen3",
  "image_as_end_frame": false,
  "flip": false,
  "motion": 5,
  "seed": 0,
  "callback_url": "",
  "time": 5
}

Guidance

text_prompt (required, string): "masterpiece, cinematic, man smoking cigarette looking outside window, moving around"
- If you see the error about The img_prompt should be a direct, unprotected image link, you may need to check content_typeto be image/png, image/jpeg, image/webp or image/jpg
img_prompt (required, string): img url or base64 image
model (optional, string): default to be gen3. Image to Video model will be using Gen3 Alpha Turbo if model is gen3
image_as_end_frame (optional, bool): default to be false. If true you'll assign the input image as the last frame
flip (optional, bool): default to be false. The Gen3 Alpha Turbo model has a fixed video resolution of 1280x768 or 768*1280. If you wanna use the portrait mode just set the flip to be true
motion (optional, int): default to be 5; Increase or decrease the intensity of motion in your video. Higher values result in more motion. Gen3 Alpha won't use this parameters.
seed (optional, int): default to be 0, means random value.
callback_url (optional, string): The callback_url is where a GET request will be sent once the job finishes, whether it’s successful or not. It will include the UUID in this format: <your-callback-url>?uuid=<uuid>
time (optional, string): default 5s for gen3 model, and 4s for gen2 model. You can use either 5 or 10 for gen3 models

Guidance

RunwayML Gen3 Model Now Officially Support Img2vid with portrait mode!