falVideo to videoComing soon

fal-ai/ltx-2.3-quality/extend-video/lora

Name: fal-ai/ltx-2.3-quality/extend-video/lora
Brand: fal

fal-ai/ltx-2.3-quality/extend-video/lora is a video to video tool from fal that runs in the background and returns video. Coming soon: listed with its price, not yet open to runs.

This tool is listed but not yet available to run.

Price pending

Pricing

Price pending.

Limits

Runs: In the background
Returns: video
Longest queue wait: 3600s
Run deadline: 1800s
Largest inline input: 1024 KB
Largest input file: 500 MB
Largest output: 4768 MB
Outputs per run: 16

Input

Field	Kind	Default	Description
`prompt` required	text	—	The prompt describing how the video should continue.
`video_url` required	URL or file	—	The URL of the video to extend.
`extend_direction`	one of forward, backward	forward	Direction to extend the video. 'forward' continues from the end of the video, 'backward' generates a new beginning.
`num_frames`	number (whole) 9 to 481	121	The number of frames to generate for the extension segment, including num_context_frames of overlap with the source video. The new content adds approximately (num_frames - num_context_frames) / frames_per_second seconds. The stable generation budget caps the resolution x frames volume: at 24 fps roughly 20s at 480p-class, 11s at 720p-class and 5s at 1080p-class; a request above the budget returns a 422 (lower num_frames or use a smaller resolution).
`num_context_frames`	number (whole) 9 to 121	25	The number of source frames used as context/overlap for the extension. Snapped down to the LTX temporal grid (8k+1) and clamped to the frames available in the source video.
`end_image_url`	URL or file	—	Optional keyframe image for the far end of the extension: the final frame for forward extension, or the new first frame for backward extension.
`end_image_strength`	number 0 to 1	1	Conditioning strength of the optional end keyframe. 1.0 = exact match, lower = more freedom for the model.
`resolution`	JSON	auto	The size of the generated video. 'auto' follows the source video size/aspect up to the LTX limits; the source is scaled to the final size in the stitched output. Sizes are 32px-aligned (e.g. 1920x1080 is delivered as 1888x1056). Higher resolutions reduce the maximum stable extension length; lower resolutions can run longer.
`match_input_fps`	true or false	true	When enabled, the output FPS matches the source video's FPS (rounded to an integer and clamped to the supported range) instead of frames_per_second.
`frames_per_second`	number 1 to 60	24	Frames per second of the generated video.
`video_strength`	number 0 to 1	1	Conditioning strength of the source context frames. 1.0 keeps the overlap locked to the source for a seamless transition; lower values give the model more freedom.
`num_inference_steps`	number (whole) 8 to 30	15	Number of inference steps. Defaults to 15 and can be increased up to 30.
`guidance_scale`	number 1 to 20	1	Classifier-free guidance scale. The default is tuned for fast, high-quality generation.
`generate_audio`	true or false	true	Whether to generate new audio for the extension segment (crossfaded with the source audio at the seam). When disabled, the source video's own audio track is preserved and the extension stays silent.
`negative_prompt`	text	color distortion, overexposure, static, blurry details, subtitles, style, artwork, painting, frame, still, dim overall tone, worst quality, low quality, JPEG compression artifacts, ugly, mutilated, extra fingers, poorly drawn hands, poorly drawn face, deformed, disfigured, malformed limbs, fused fingers, motionless frame, cluttered background, three legs, crowded background, walking backwards	The negative prompt to steer generation away from.
`seed`	number (whole)	—	Random seed for reproducibility. If None, a random seed is chosen.
`enable_prompt_expansion`	true or false	true	Whether to enable prompt expansion.
`video_quality`	one of low, medium, high, maximum	high	The quality preset of the generated video.
`video_write_mode`	one of fast, balanced, small	balanced	The write mode of the generated video.
`loras` required	JSON	—	Up to 3 LoRAs to apply on top of LTX-2.3. Each path is downloaded through the registry SSRF-safe downloader before it is loaded. Max size: 3 GB per LoRA.

Same request in code

curl https://api.spiral.ai/v1/runs \
  -H "Authorization: Bearer $SPIRAL_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{
    "provider": "fal",
    "model": "fal-ai/ltx-2.3-quality/extend-video/lora",
    "input": {
      "prompt": "Continue the scene naturally, maintaining the same style and motion.",
      "video_url": "https://v3b.fal.media/files/b/0a8824b1/sdm0KfmenrlywesfzY1Y1_if6euPp1.mp4",
      "loras": {}
    }
  }'