Openjutsu Object Swap API Documentation

Openjutsu Object Swap API Documentation

Playground

Try it on WaveSpeedAI!

Genjutsu-style object swap: change hair, clothes or the whole background in a video by naming it, keeping the rest of the shot and the original audio.

Features

Genjutsu Object Swap (OpenJutsu) changes one thing in a video by naming it: hair, a jacket, a car, or the whole background. Only that area is regenerated; the rest of the shot, the camera movement, the performance and the original audio are kept.


Why Choose This?

  • Name it, change it Say what to change (hair, jacket, car) and what it becomes (“short bright pink hair”). Everything else in the shot stays as filmed.

  • Copy a look from a picture Give a picture of an outfit, a hairstyle, a product or a place and the video takes on its look. The shape and movement of the original are followed.

  • New background, same person With background mode the person stays and the setting around them changes: a beach, a studio, a rainy street.

  • Original audio kept The soundtrack of the source video is carried over to the result.

  • Up to 15 seconds The result follows the length of your video, up to 15 seconds. Longer videos use their first 15 seconds.


Parameters

ParameterRequiredDescription
videoYesURL of the source video, at least 2 seconds. Use a single continuous shot.
targetYesWhat to change, in a word or two as it looks in the video: hair, jacket, car. Separate several with commas (hair, top). With background mode, what stays (person).
modeNochange (default): the target becomes what you describe or show. background: the target stays and everything around it changes.
change_toNoWhat it becomes, for example short bright pink hair. With background mode, the new place. Required unless image is given.
imageNoURL of a picture to copy the look from. Adds a flat fee.
resolutionNoOutput resolution: 480p, 540p, 768p or 1080p. Default: 768p.
seedNoRandom seed for generation. A negative value means a random seed will be used.

Tips

  • Name the target as it looks in the video, in a word or two. Every match changes, so jacket changes every jacket in the shot.
  • Several things can be changed together: hair, top.
  • change keeps the outline of what it replaces, so a tank top turned into a flannel shirt stays sleeveless. The shape and movement come from the original.
  • A picture only needs to show the thing itself: the outfit, the hairstyle, the place.
  • Use a video with one continuous shot. Mirrors and reflections of the target are not supported.
  • If nothing matching the target is found, the request fails with an error message. If the name is vague it may match something else, so use a specific word.

Pricing

> Billing duration: The length of your input video is rounded up to whole seconds. The minimum billed duration is 3 seconds; shorter videos are billed as 3 seconds. Videos longer than 15 seconds are billed as 15 seconds, and only their first 15 seconds are used.

The price depends on the billed length and the resolution. A reference picture adds a flat fee.

ResolutionPrice per secondMax billed length
1080p0.40 USD15 s
768p0.20 USD15 s
540p0.17 USD15 s
480p0.15 USD15 s

A flat 0.02 USD is added when you give a reference picture in image.

Billing rules

  • Billed duration = input video length rounded up to whole seconds, clamped to 3-15 seconds.
  • 1080p: 0.40 USD per billed second.
  • 768p: 0.20 USD per billed second.
  • 540p: 0.17 USD per billed second.
  • 480p: 0.15 USD per billed second. Lower resolutions are not proportionally cheaper: finding the people takes the same time at any size.
  • Total = billed seconds x rate per second + 0.02 USD if image is given.

So, for example:

  • 10 s video at 768p: 10 x 0.20 = 2.00 USD
  • 10 s video at 540p: 10 x 0.17 = 1.70 USD
  • 10 s video at 480p: 10 x 0.15 = 1.50 USD
  • 10 s video at 1080p: 10 x 0.40 = 4.00 USD
  • 10 s video at 768p with a reference picture: 10 x 0.20 + 0.02 = 2.02 USD
  • 15 s video at 768p: 15 x 0.20 = 3.00 USD
  • 2 s video at 768p (billed as 3 s): 3 x 0.20 = 0.60 USD
  • 30 s video at 768p (billed as 15 s): 3.00 USD

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'
{
  "video": "https://interactive-examples.mdn.mozilla.net/media/cc0-videos/flower.mp4",
  "target": "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg",
  "mode": "change",
  "resolution": "768p"
}
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/wavespeed-ai/openjutsu/object-swap" \
  -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

Retry temporary errors only while polling, never the submit request. Errors, retries and timeouts

Parameters

Task Submission Parameters

Request Parameters

ParameterTypeRequiredDefaultRangeDescription
videostringYes-URL of the source video. Use a single shot. The output follows the video's length, up to 15 seconds; the scene, camera, motion and original audio are kept.
targetstringYes--What to change, in a word or two as it looks in the video: hair, jacket, car. Several things can be separated by commas (hair, top). With mode background, what stays (person).
modestringNochangechange, backgroundchange: the named thing becomes what you describe or show. background: the target stays and everything around it changes.
change_tostringNo--What it becomes, for example short bright pink hair; with mode background, the new place. Required unless image is given.
imagestringNo-Optional URL of a picture to copy the look from: an outfit, a hairstyle, a product or a place. Adds a flat fee.
resolutionstringNo768p480p, 540p, 768p, 1080pOutput resolution. The price per second rises with it.
seedintegerNo--The random seed to use for the generation. A negative value means a random seed will be used.

Response Parameters

ParameterTypeDescription
codeintegerHTTP status code (e.g., 200 for success)
messagestringStatus message (e.g., “success”)
data.idstringUnique identifier for the prediction, Task Id
data.modelstringModel ID used for the prediction
data.outputsarrayOutput values, usually URL strings; some models return text strings or structured result objects (empty when status is not completed)
data.urlsobjectObject containing related API endpoints
data.statusstringTask status. completed is successful; failed, cancelled, timeout, and deleted are failure terminal statuses.
data.created_atstringISO timestamp of when the request was created (e.g., “2023-04-01T12:34:56.789Z”)
data.errorstringError message (empty if no error occurred)
data.timingsobjectObject containing timing details
data.timings.inferenceintegerInference time in milliseconds

Result Request Parameters

ParameterTypeRequiredDefaultDescription
idstringYes-Task ID

Result Response Parameters

ParameterTypeDescription
codeintegerHTTP status code (e.g., 200 for success)
messagestringStatus message (e.g., “success”)
dataobjectThe prediction data object containing all details
data.idstringUnique identifier for the prediction
data.modelstringModel ID used for the prediction
data.outputsarray<string | object>Array of generated outputs (empty when status is not completed). Items are usually URL strings, but may be text strings or structured result objects, depending on the model.
data.urlsobjectObject containing related API endpoints
data.statusstringStatus: completed is successful; failed, cancelled, timeout, and deleted are failure terminal statuses
data.created_atstringISO timestamp of when the request was created
data.errorstringError message (empty if no error occurred)
data.timingsobjectObject containing timing details
data.timings.inferenceintegerInference time in milliseconds
© 2026 WaveSpeedAI. All rights reserved.