Runway Gen-4 Turbo
Gen-4 Turbo from Runway animates one still image into a 5 or 10 second video at 24 fps. Your image becomes the first frame, and the prompt only has to describe the motion.
Overview
Gen-4 Turbo is the lighter variant of Runway's Gen-4 video model, and it always starts from a picture. The image you upload becomes the first frame of a 5 or 10 second clip at 24 fps, and the text prompt describes what happens after it.
The image already sets the look of the shot, so Runway recommends spending the prompt almost entirely on motion: how the subject moves and what the camera does.
What you can build
- Product clips from one packshot, such as a slow push-in on a bottle.
- Vertical 720x1280 clips for stories, or square 960x960 clips for feeds.
- Animatics that put motion into storyboard frames before a shoot.
- Landing page backgrounds where fog or water drifts through an otherwise still scene.
- 1584x672 widescreen shots for title cards and banners.
Inputs
- First Image (required): the opening frame, in JPEG, PNG or WebP (Runway does not accept GIF). Its width divided by its height must be between 0.5 and 2.358. If it does not match the chosen ratio, it is cropped from the center.
- Prompt (required): up to 1000 characters describing the motion.
- Last Image (optional): Gen-4 Turbo takes a first frame only, so an image here does not decide how the clip ends.
- Ratio: the output size in pixels. 1280:720 (16:9, default), 720:1280 (9:16), 1104:832 (4:3), 832:1104 (3:4), 960:960 (1:1) or 1584:672 (21:9).
- Duration: 5 seconds (default) or 10 seconds.
- Public Figure Threshold: auto (default) or low.
- Seed: an integer from 0 to 4294967295. Runway says the same seed on an identical request gives similar results.
Outputs
One MP4 video at the chosen pixel size, 24 fps, 5 or 10 seconds long.
Recommended settings
- Begin with a short prompt covering only the main motion, then add subject motion, camera motion, scene motion or style descriptors one at a time.
- Write what should happen. Runway says negative phrasing is not supported and can give unpredictable or even opposite results, so ask for a "locked camera" in place of "no camera movement".
- Call people and objects "the subject" or "the man on the left" and describe physical actions. Re-describing details already in the image can reduce motion.
- Use 5 seconds for one simple action and 10 seconds when the prompt asks for several movements.
- Pick the ratio closest to your image to keep the crop small, and use a source without visible artifacts.
Limitations
- Each clip works best as a single scene. Runway says prompts that script several scene changes or style shifts tend to give unintended results.
- Output stops at the six sizes above, with no 1080p or 4K option.
- Runway now lists Gen-4 and Gen-4 Turbo as an older generation of its models, with Gen-4.5 as its current image-to-video model.
Safety & compliance
Runway moderates both the image and the prompt of every request, and a moderated request fails without returning a video. Setting Public Figure Threshold to low makes that check less strict about recognizable public figures. If your first image shows a real person's face, make sure you have that person's consent.
Example prompts
Great starting points for image-to-video-gen4-turbo.
API quick start
Run image-to-video-gen4-turbo with a single API call.
{
"prompt": "Your prompt here",
"inputImage": "https://your-cdn.com/input.png",
"ratio": "1280:720",
"inputImageLast": "https://your-cdn.com/input.png"
}curl -X POST "https://api.wiro.ai/v1/Run/runway/image-to-video-gen4-turbo" \
-H "Content-Type: application/json" \
-H "x-api-key: YOUR_WIRO_API_KEY" \
--data-binary @- <<'JSON'
{
"prompt": "Your prompt here",
"inputImage": "https://your-cdn.com/input.png",
"ratio": "1280:720",
"inputImageLast": "https://your-cdn.com/input.png"
}
JSONOther models you might like
Explore similar video generation models.