Methods to Generate New Videos

This guide covers key features and workflow parameters supported by LipDub API for flexible video generation, including looping, selective dubbing, frame indexing, quality options, and webhooks.

1. Loop Video

Description:
Create videos that repeat or loop segments seamlessly.
Parameter:

loop_count (integer) — Number of times to repeat the base video segment.
Example:
To loop the input clip 3 times:

{
  "input_video": "input.mp4",
  "loop_count": 3
}

2. Start Frame Index

Description:
Begin dubbing or processing from a specific frame in the input video, enabling precise synchronization or cropping.
Parameter:

start_frame (integer) — Index of the first frame to process.
Example:
To start dubbing from frame 150:

{
  "input_video": "input.mp4",
  "start_frame": 150
}

3. Selective Dubbing

Description:
Target specific audio segments or time ranges for dubbing, leaving other parts of the video unchanged.
Parameter:

segments (array) — List of time or frame intervals to dub.
Example:
To dub only the section from 10s to 30s:

{
  "input_video": "input.mp4",
  "segments": [
    {"start": 10, "end": 30}
  ]
}

4. Choose Output Video Quality

Description:
Set the output video resolution or encoding for optimal balance of speed, file size, and visual quality.
Parameters:

output_quality (enum) — Options like '480p', '720p', '1080p', 'custom'
Optional: bitrate, codec
Example:
To generate a 720p output:

{
  "input_video": "input.mp4",
  "output_quality": "720p"
}

5. Callback URL

Description:
Receive asynchronous status updates and download links when video processing is complete.
Parameter:

callback_url (string) — Your webhook endpoint.
Example:
To specify a callback:

{
  "input_video": "input.mp4",
  "callback_url": "https://yourdomain.com/api/lipdub/callback"
}

Example Full Request

{
  "input_video": "input.mp4",
  "loop_count": 2,
  "start_frame": 100,
  "segments": [{"start": 20, "end": 40}],
  "output_quality": "1080p",
  "callback_url": "https://yourdomain.com/lipdub/notify"
}

6. LipDub2 (One-Call Render)

Description: LipDub2 (LTX2) is a generation-only model — no training or preprocessing. Create a render in a single POST /v1/renders call, providing each side as either an already-uploaded id or a source URL.

Parameter(s):

model_type (string, required) — must be "lipdub2"
output_filename (string, required)
video_id (int) or video_url (string) — exactly one (the reference video)
audio_id (string) or audio_url (string) — exactly one (the target audio)
callback_url (string, optional) — HTTPS URL notified on completion

Example:

{
  "model_type": "lipdub2",
  "output_filename": "my_render.mp4",
  "video_id": 4056,
  "audio_id": "dadddd0a-2d89-4749-876f-21a483af6965"
}

Notes & Tips

Combine multiple features for advanced workflows.
Ensure frame indices and time segments align with the input video’s actual length.
Use callback URLs for scalable, asynchronous integrations.
LipDub2: both sides as ids → 201 { "data": { "generate_id", "credit" } }; either side as a URL → 202 { "data": { "tracking_id" } }, then poll GET /v1/renders/track/{tracking_id}.
LipDub2: video_id is the shot id from /v1/video's success callback; audio_id is the upload id from /v1/audio.

See API reference for detailed parameter definitions and supported values.