Veo 3.1 image to video: how it works on imaginary.app

How Veo 3.1 image-to-video works

In image-to-video, your picture becomes the opening frame and Veo 3.1 generates what happens next: camera movement, subject motion, changing light and, if audio is on, sound. The model keeps the look of your image (composition, colours, characters) far more closely than a text prompt alone can, which is why people use it for product shots, brand characters and illustrations.

The prompt now does a different job. Your image already describes the scene, so the prompt should describe change: what moves, how the camera moves, and what you hear. "The camera slowly orbits the bottle as condensation runs down the glass; soft café ambience" does more than re-describing the bottle.

First and last frame

Upload a second image as the Last Frame Image and Veo 3.1 generates the motion that connects the two. This gives you control over where a shot ends: a before-and-after reveal, a product turning to face the camera, a character crossing the room, or a transition that lands on your logo. It works with Lite, Fast and Quality on imaginary.app.

Step by step in the generator

1. 1

   Open the generator and sign in

   Go to the generator and sign in with Google. No credit card is needed to create an account.
2. 2

   Click the Image-to-Video tab

   It sits next to Text-to-Video at the top of the form. The upload fields appear below it.
3. 3

   Upload the First Frame Image

   Click the upload box and choose a PNG or JPEG under 20 MB. A preview with the file size appears once it uploads.
4. 4

   Optionally upload a Last Frame Image

   Add a second PNG or JPEG (also up to 20 MB) if you want the clip to end on a specific frame. Leave it empty for a normal image-to-video clip.
5. 5

   Describe the motion (optional)

   The field is labelled "Prompt (Optional)". Describe movement, camera and sound rather than the image itself. Up to 2,000 characters; Enhance can expand a short prompt.
6. 6

   Choose model, resolution, aspect ratio, length and audio

   Pick Fast, Quality or Lite, then 720p, 1080p or 4K (no 4K on Lite), 16:9 or 9:16, 4s/6s/8s, and the With Audio toggle. The credit cost shows next to Generate Video.
7. 7

   Generate

   Click Generate Video. The clip renders on our servers and lands on the page and in My Creations. If the generation fails, your credits are returned.

Input rules at a glance

What image-to-video costs

Image-to-video uses the same credit prices as text-to-video, so the numbers below are simply each model's range. Full tables are on each model page; the cost calculator prices a whole batch, and pricing lists the credit packs.

Tips for better results from a still image

• Match the aspect ratio to your image. Choose 9:16 for a portrait photo and 16:9 for a landscape one, so the model isn't forced to reframe it.
• Prompt the change, not the content. One camera move and one or two actions per clip; 8 seconds goes quickly.
• Keep first and last frames compatible. Similar lighting, lens and framing make the in-between motion believable. Two unrelated images tend to produce a morph.
• Add sound in its own line. With audio on, an Audio: line ("Audio: rain on the window, distant traffic") tells the model what to hear.
• Test on Lite. Try motion ideas on Veo 3.1 Lite, then re-run the keeper on Fast or Quality.
• Chain clips for longer scenes. Use the last frame of one clip as the first frame of the next; our guide to going past 8 seconds walks through it.

For prompt structure in general, see the Veo 3.1 prompting guide.

Why there are no image-to-video examples here

Our public gallery only accepts text-to-video clips: the "Share to gallery" switch is not shown on the Image-to-Video tab, because the result contains the user's own image. We won't show image-to-video results we can't attribute and verify, so instead of examples, this page gives you the exact steps and costs. The gallery is still useful for motion and sound wording that transfers to image prompts; the model pages for Quality, Fast and Lite each show real clips.

Image-to-video through the API

The developer API takes the same inputs as base64 strings. A prompt is required on the API, and lastFrame is optional. Poll GET /api/status/:taskId for the result. Base URL, authentication options and response formats are in the API documentation.

POST /api/generate
Authorization: Bearer veo_your_api_key_here
Content-Type: application/json

{
  "model": "veo3-fast",
  "modelVersion": "3.1",
  "prompt": "The camera slowly pushes in as steam rises from the cup",
  "audio": true,
  "image": {
    "bytesBase64Encoded": "<base64 first frame>",
    "mimeType": "image/jpeg"
  },
  "lastFrame": {
    "bytesBase64Encoded": "<base64 last frame>",
    "mimeType": "image/jpeg"
  },
  "options": { "aspectRatio": "16:9", "resolution": "1080p" }
}