Vidu Reference To Image Q2

Vidu Reference To Image Q2

Playground

Vidu Reference-to-Image Q2 generates high-quality images from 1–7 reference images plus a text prompt, preserving style and composition while allowing controlled changes to subjects, backgrounds, and fine details. Ready-to-use REST inference API, best performance, no cold starts, affordable pricing.

Features

vidu/reference-to-image-q2 is the reference-guided sibling of vidu’s text-to-image model. It takes one or more reference images (up to 7) plus a prompt, and generates new, high-resolution images that keep the subject and composition while adjusting style, lighting, or scene details.

What it’s good for

• Keeping product, character, or actor identity consistent across many shots
• Creating new scenes from a small set of reference stills or keyframes
• Generating campaign variations while locking in pose, outfit, or layout
• Up-res, clean re-renders of storyboard / concept frames with cinematic quality

Key features

• Up to 7 reference images

Upload 1–7 images in images to steer identity, pose, outfit, or composition. The model blends information across them while following your text prompt.

• Cinematic aspect ratios

aspect_ratio supports:

• 1:1, 4:3, 3:4, 2:3, 3:2 – square and classic photo ratios
• 16:9, 21:9 – widescreen and banner formats
• 9:16 – vertical / mobile content
• auto – let the model choose a ratio that best matches the references + prompt

• High resolutions (1080p → 4K)

resolution lets you pick:

• 1080p – fast preview / web use
• 2K – more detail and better crop flexibility
• 4K – maximum sharpness for key visuals and print-adjacent work

• Prompt-driven control

Combine references with a rich prompt (“dramatic studio lighting, cinematic close-up, 85mm lens, shallow depth of field”) to re-style while keeping the same subject.

• Seed-based reproducibility

seed set to -1 gives random variation; using a fixed integer lets you rerun the same combination of prompt + references for consistent outputs.

How to use (Playground)

1. prompt* – Describe what you want to change or keep: style, lighting, mood, background, camera angle, etc.
2. images* – Click “Add Item” and upload 1–7 reference images (subject, pose, layout, or mood).
3. aspect_ratio – Choose a ratio, or leave as auto and let the model decide.
4. resolution – Select 1080p, 2K, or 4K depending on detail vs. speed needs.
5. seed – Use -1 for randomness or a fixed integer for reproducible results.
6. Run the job, inspect the result, then iterate on prompt / references as needed.

Pricing

Pricing depends on resolution and how many reference images you use. Base rate is $0.04 per 1k compute units, applied via the internal formula:

Up to 3 reference images (1–3 refs)

4–7 reference images

Tips for best quality

• Use clean, well-lit reference images; avoid heavy motion blur or extreme compression.
• Keep references stylistically consistent when possible (similar lighting / medium).
• In the prompt, clearly state both what must stay the same (“same person and outfit”) and what should change (“different background, golden-hour lighting”).
• For hero shots, generate at 2K or 4K, then downscale slightly for extra sharpness.

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": "auto",
  "resolution": "1080p",
  "seed": -1
}
JSON
)

# 1. Submit the prediction.
SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  -X POST "https://api.wavespeed.ai/api/v3/vidu/reference-to-image-q2" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}")

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 "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id
' >&2
  exit 1
fi
RESULT_URL=$(printf '%s' "${TASK}" | jq -r '.urls.get // empty')
if [ -z "${RESULT_URL}" ]; then RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"; fi

# 2. Poll until the prediction finishes.
while true; do
  RESPONSE=$(curl --silent --show-error --fail-with-body \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}")
  RESULT=$(printf '%s' "${RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout) printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    created|processing) sleep 2 ;;
    *) printf 'Unexpected status: %s
' "${STATUS}" >&2; exit 1 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Response Parameters

Result Request Parameters

Result Response Parameters