---
title: "Generate a video from 1-5 reference images and a text prompt (references-to-video). Unlike createVideo, which animates a single source image, this composes a new scene that borrows characters, objects, and style from the reference images. Each image can be a URL or base64. Synchronous until September 10, 2026: the call blocks until rendering finishes and returns the video URL and its actual duration in seconds. Choose the output shape with `aspect_ratio` (\"default\" lets the model decide). The chosen `model` and `duration` must be compatible (incompatible combinations return HTTP 400). Credits are charged only on success, based on the produced duration and never more than the duration you requested. Pass an optional `request_id` to tag the result so you can locate it later via `GET /assets/videos/results`. Related tools: `createVideo` for image-to-video, `editVideo` to modify a generated video. Requires an API key (user scope). Async is the intended integration pattern: pass `async: true` and the call returns 202 with a job id - poll `GET /assets/jobs/{id}` (`getApiJob` in MCP), honouring poll_after_ms or long-polling with wait=30, until status is succeeded, whose result field is exactly the response documented for this operation. Synchronous responses are deprecated: they remain the default until September 10, 2026, after which requests default to async and return 202 with a job id; set `async: false` explicitly to keep synchronous behavior during and after the transition. Synchronous calls run on the same queue: the response carries an X-Ludo-Job-Id header, and a job still running after 15 minutes comes back as 202 with the job instead of an error. Each account may have up to 50 generations queued or running at once via the API; a request beyond that returns 429 (code PENDING_JOBS_LIMIT)."
method: POST
path: "/assets/video/references"
tags: ["Videos"]
---

# Generate a video from 1-5 reference images and a text prompt (references-to-video). Unlike createVideo, which animates a single source image, this composes a new scene that borrows characters, objects, and style from the reference images. Each image can be a URL or base64. Synchronous until September 10, 2026: the call blocks until rendering finishes and returns the video URL and its actual duration in seconds. Choose the output shape with \`aspect_ratio\` ("default" lets the model decide). The chosen \`model\` and \`duration\` must be compatible (incompatible combinations return HTTP 400). Credits are charged only on success, based on the produced duration and never more than the duration you requested. Pass an optional \`request_id\` to tag the result so you can locate it later via \`GET /assets/videos/results\`. Related tools: \`createVideo\` for image-to-video, \`editVideo\` to modify a generated video. Requires an API key (user scope). Async is the intended integration pattern: pass \`async: true\` and the call returns 202 with a job id - poll \`GET /assets/jobs/{id}\` (\`getApiJob\` in MCP), honouring poll_after_ms or long-polling with wait=30, until status is succeeded, whose result field is exactly the response documented for this operation. Synchronous responses are deprecated: they remain the default until September 10, 2026, after which requests default to async and return 202 with a job id; set \`async: false\` explicitly to keep synchronous behavior during and after the transition. Synchronous calls run on the same queue: the response carries an X-Ludo-Job-Id header, and a job still running after 15 minutes comes back as 202 with the job instead of an error. Each account may have up to 50 generations queued or running at once via the API; a request beyond that returns 429 (code PENDING_JOBS_LIMIT).

`POST /assets/video/references`

## Request body

- ReferencesToVideoPayloadPublic — Payload for generating a video from 1-5 reference images and a text prompt.
  - `prompt` string, required — Text description of the video to generate.
  - `images` string[], required — Reference images (1 to 5), each a URL or base64. The generated video borrows characters, objects, and style from them.
  - `duration` number, float — Video length in seconds.
  - `model` 'eagle' | 'eagle-audio' | 'forge-pixel' — Video model to use. forge-pixel renders pixel-art video.
  - `aspect_ratio` 'default' | 'ar_1_1' | 'ar_16_9' | 'ar_9_16' | 'ar_4_3' | 'ar_3_4' | 'ar_21_9' — Output aspect ratio. "default" lets the model choose.
  - `request_id` string — Optional client-provided identifier, unique per request. Re-sending the same request_id returns the existing job instead of generating again. Also usable with the results endpoint.
  - `async` boolean — When true, return 202 immediately with a job id instead of waiting for the result. Poll GET /assets/jobs/{id}. Defaults to false (synchronous).

## Response `200`

Success

- VideoResult
  - `url` string — URL to the generated video file
  - `duration` number, float — Duration of the video in seconds
  - `has_audio` boolean
  - `request_id` string
  - `created_at` integer

## Other responses

- `202` — Accepted - the generation was queued (async: true, or a synchronous call still running after 15 minutes). Poll GET /assets/jobs/{id}.
- `400` — Error

## Changes

- **2026-08-30** `a8a0b85d1839` — 1 info
  - added the success response with the status `202`

[Change history](https://skmtc.dev/ludo/apis/ludo-ai-api/changes/assets/video/references/post.md)

---

[API](https://skmtc.dev/ludo/apis/ludo-ai-api.md) · [All operations](https://skmtc.dev/ludo/apis/ludo-ai-api/llms.txt) · [OpenAPI document](https://skmtc-service-production.skmtc.workers.dev/v1/apis/ludo/ludo-ai-api/revisions/8c2e600e07d4/schema)