Veo 3.1 Fast Reference-to-Video API

veo3.1-fast/reference-to-video

Generate an 8-second video from exactly three reference images. Use the prompt to describe how their visual guidance should appear in the resulting shot.

Input

Prompt

238/10,000

Reference image 1

Drag and drop files, paste from clipboard, or provide a URL.

Reference image 2

Drag and drop files, paste from clipboard, or provide a URL.

Reference image 3

Drag and drop files, paste from clipboard, or provide a URL.

Duration

Fixed by this endpoint

Resolution

Aspect ratio

Output

Idle

Generated video appears here

Configure the input and run the task. Progress and the finished video will stay in this panel.

Examples

One continuous low tracking shot in the referenced wet misty forest. The referenced compact moss-covered brass-and-ceramic robot walks four careful steps along the puddled path, pauses, tilts its rounded head toward a falling droplet, and its single amber eye brightens slightly. Preserve the robot's exact silhouette, one eye, chest hatch, materials, moss coverage and scale; use the forest and cool dawn lighting from the references. No text, no logos, no brands, no products, no advertising, no watermark.

One continuous wide underwater shot using the three references. The transparent hand-blown glass jellyfish drifts slowly through the cobalt cavern, its eight curled glass tentacles lagging with gentle water resistance while caustic light refracts through trapped bubbles. Preserve the rounded bell, aqua rim, clear material and artisan bubbles; use the referenced cavern and lighting. The camera tracks beside it without cuts. No animals, no text, no logos, no brands, no products, no advertising, no watermark.

One continuous vertical shot in the referenced rainy conservatory. The referenced small ceramic mechanical bird stands on the stone ledge; its brass winding key turns once, cobalt wing plates open in a measured clockwork sequence, and the bird makes one short balanced hop before settling. Preserve the white crackle glaze, brass beak and legs, cobalt hinges, key and hand-sized scale. Use the referenced pale rainy light. No text, no logos, no brands, no products, no advertising, no watermark.

REST API

Quick Start

Authenticate, submit the exact model contract, then poll the returned task ID. The complete field reference remains in the API documentation.

Authenticate

Send your Vidgo API key as a Bearer token. Keep the key on the server and never expose it in browser code.

Endpoint: POST https://api.vidgo.ai/api/generate/submit
Authentication: Authorization: Bearer VIDGO_API_KEY

Submit the generation task

Use the exact model ID veo3.1-fast/reference-to-video. The submit endpoint returns an asynchronous task ID.

Submit request

curl --request POST \
  --url "https://api.vidgo.ai/api/generate/submit" \
  --header "Authorization: Bearer $VIDGO_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "veo3.1-fast/reference-to-video",
  "input": {
    "prompt": "A single continuous shot of a paper boat drifting through a rain-lit city canal at dusk.",
    "duration": 8,
    "resolution": "720p",
    "aspect_ratio": "16:9",
    "image_urls": [
      "https://example.com/reference-1.jpg",
      "https://example.com/reference-2.jpg",
      "https://example.com/reference-3.jpg"
    ]
  }
}'

Submit response

{
  "code": 200,
  "data": {
    "task_id": "task-unified-...",
    "status": "not_started"
  }
}

Poll for the result

Request task status with task_id until the task reaches finished or failed.

GET https://api.vidgo.ai/api/generate/status/{task_id}

not_started → running → finished | failed

Full documentation: Veo 3.1 Fast Reference-to-Video API reference

Veo 3.1 Fast Reference-to-Video

Veo 3.1 Fast Reference-to-Video uses exactly three reference images and a prompt to guide an 8-second generated video. It is available on the Fast endpoint. Fast prioritizes a lower credit cost for iteration.

Why Choose This?

Guide the shot with three referencesProvide a fixed three-image reference set when text alone is not enough to communicate the intended visual direction.
Use an explicit endpoint contractThe model ID locks this page to exactly 3 reference image urls, so the form and JSON request stay aligned.
Choose the delivery resolutionUse 720p, 1080p, or 4K output with landscape or portrait framing.
Budget before submissionThe Playground shows the required credits for the selected resolution before the task is sent.

Parameters

Parameter	Requirement	Description
prompt	Required	String. Describe visible content, action, camera direction, lighting, and composition; maximum 10,000 characters.
image_urls	Required	Array of exactly 3 public HTTP(S) reference image URLs.
duration	Optional	Integer. Defaults to 8; this endpoint accepts no other duration. Default`8`
resolution	Optional	String. Select the output resolution; defaults to 720p. Default`720p1080p4k`
aspect_ratio	Optional	String. Select landscape or portrait framing. `16:99:16`

How to Use

Prepare three referencesUpload exactly three public images that clearly communicate the visual guidance needed for the shot.
Direct the motionDescribe concrete subject movement and camera movement in separate clauses so each instruction has a clear role.
Set the outputKeep duration at 8 seconds, then select resolution and aspect ratio.
Run and track the taskSubmit the request, retain task_id, and poll status until the task is finished or failed.

Pricing

Billing is per generated video and depends on the selected resolution.

Usage	Rate	Details
720p	36 credits	Per successful 8-second video request.
1080p	36 credits	Per successful 8-second video request.
4k	60 credits	Per successful 8-second video request.

Best Use Cases

Visual developmentExplore a motion treatment from a compact set of approved visual references.
Campaign consistencyGuide repeated generations with the same three-image visual reference set.
Style explorationCombine a written shot direction with concrete visual cues.

Pro Tips

Write one coherent shot rather than several unrelated scenes.
Separate subject action, camera movement, and visual treatment into clear clauses.
Use three references with clear, complementary visual roles.

Notes

Generation is asynchronous; keep task_id and stop polling when status reaches finished or failed.
The public model ID for this page is veo3.1-fast/reference-to-video.