Vidu Q4 Reference To Video API Documentation

Vidu Q4 Reference To Video API Documentation

Playground

Vidu Q4 Reference-to-Video generates subject-consistent AI videos from text prompts, 1-15 reference images, and up to 3 reference audio clips, supporting character consistency, product videos, social content, brand assets, and reference-guided storytelling workflows. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

Vidu Q4 Reference-to-Video generates a video of 3 to 16 seconds, up to 4K, from a text prompt and up to 15 reference images, so people, products and scenes stay consistent. Add up to 3 voice clips to give characters their own voice.

Why Choose This?

• Consistent subjects — Characters, products and settings from your references stay recognizable across shots.
• Up to 15 reference images — Cover several subjects and an environment in a single request.
• Voice references — Up to 3 MP3 clips guide how the speakers sound.
• Sound included — Dialogue, effects and ambience are generated with the picture, and can be switched off.
• Five aspect ratios and five resolutions — From 540p to 4K for widescreen, vertical or square layouts.

Parameters

images and audios are arrays even when you send a single item.

How to Use

1. Collect references — Choose clear images of the characters, products or places that belong in the scene.
2. Write the prompt — Say what each subject does. Refer to the images by their order in the array: @1 is the first image, @2 the second, and so on.
3. Add voices (optional) — Provide up to 3 short MP3 clips in audios.
4. Choose the format — Set the aspect ratio, resolution and duration.
5. Run — Submit the request and download the video when it finishes.

Pricing

Pricing is based on the selected resolution and video duration.

The total is the per-second rate for the selected resolution multiplied by duration (whole seconds, 3 to 16). The default of 720p and 5 seconds costs $0.475. Audio generation and the other settings do not change the price.

Best Use Cases

• Character-led content — Keep a recurring character or mascot consistent across clips.
• Product videos — Place a product from several angles into a promotional scene.
• Brand storytelling — Combine a brand’s visual assets with a written script.
• Multi-subject scenes — Bring several people and a location together in one shot.
• Social content — Render the same idea as widescreen, vertical or square.

Pro Tips

• Use clear references that belong in the scene, and avoid contradictory ones unless the contrast is intended.
• Say how each important subject should look and behave, and tie it to its image with @1, @2.
• Keep the action small enough to finish within the chosen duration.
• Treat the two audio controls separately: audios gives the model voices to follow, and generate_audio decides whether the video has sound at all.
• Explore at 540p or 720p, then render the final clip at a higher resolution.
• A fixed seed makes variations easier to compare, but identical output is not guaranteed.

Related Models

• Vidu Q4 Image-to-Video - Animate a single image into a video, with optional prompt rewriting.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "images": [
    "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg"
  ],
  "aspect_ratio": "16:9",
  "resolution": "720p",
  "duration": 5,
  "generate_audio": true
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/vidu/q4/reference-to-video" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Response Parameters

Result Request Parameters

Result Response Parameters