Kwaivgi Kling V2.5 Turbo Pro Image To Video API Documentation

Kwaivgi Kling V2.5 Turbo Pro Image To Video API Documentation

Playground

Kling 2.5 Turbo Pro converts images to cinematic videos with fluid motion, dynamic effects, and precise prompt-driven motion for seamless transitions. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

Kling 2.5 Turbo Pro turns a single image and a text prompt into cinematic video with fluid motion and accurate intent. A new text-timing engine, improved dynamics, and faster inference enable high-speed action and complex camera moves with stable frames, while refined conditioning preserves palette, lighting, and mood.

This version additionally supports first–last frame control: you can specify both a starting image and an ending image, and the model will animate a smooth transformation between them.

What makes it stand out?

• Better prompt understanding Precisely parses multi-step, causal instructions and turns a single image and prompt into coherent, well-paced shots that stay true to your creative idea.

• More realistic look and greater stability Improved dynamics and balanced training data closely mimic real-world motion, even at high speeds and with complex camera moves. Playback is smooth with fewer jitters, tears, and dropped details.

• Detail and style consistency Refined image conditioning maintains color, lighting, brushwork, and mood, keeping frames visually unified even during aggressive motion or transitions.

• First–last frame animation When you provide both an initial image and a last_image, Kling 2.5 Turbo Pro treats them as keyframes and generates a video that naturally evolves from the first to the last frame.

Inputs

• image (required) The starting frame of your video. Composition, style, and subject are primarily taken from this image.

• last_image (optional) An optional target frame. If provided, the model interpolates between image and last_image, creating a smooth visual evolution from start to end.

• prompt (required) Text description of the scene, actions, camera movement, and style.

• negative_prompt (optional) Things you want the model to avoid (for example, blur, text overlays, distortions).

• guidance_scale Controls how strongly the model follows the prompt versus being more free-form. Lower values = more creative variation; higher values = stricter adherence to the prompt.

• duration Length of the generated video:

• 5 seconds

• 10 seconds

Output: a single video clip of the chosen duration, animated from the initial image (and optionally toward the last_image) according to your prompt.

Designed For

• Marketing and brand teams – Consistent, on-brand motion spots, feature demos, and campaign assets.
• Creators / YouTubers / Shorts teams – Strong narrative motion that boosts watch-through and engagement.
• Film / animation studios – Previz, style tests, and technique exploration with reliable dynamics.
• Education and training – Turn static diagrams or slides into clear, animated explainers.

Pricing

How to Use

1. Upload or paste the URL of your image as the starting frame.
2. (Optional) Upload a last_image if you want the video to end on a specific frame or design.
3. Write your prompt, specifying subject, scene, motion, and style.
4. (Optional) Add a negative_prompt to filter out unwanted artifacts or styles.
5. Adjust guidance_scale to balance between strict prompt following and looser creativity.
6. Choose the duration (5 s or 10 s).
7. Run the model, preview the result, then iterate by tweaking the prompt, images, or guidance_scale until you reach the desired look.

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",
  "image": "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg",
  "guidance_scale": 0.5,
  "duration": 5
}
JSON
)

# 1. Submit the prediction.
SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  -X POST "https://api.wavespeed.ai/api/v3/kwaivgi/kling-v2.5-turbo-pro/image-to-video" \
  -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