Generates dynamic videos from pure text, covering high-quality default mode (4–15s, all ratios, 480p/720p, audio default on) and high-dynamic fast mode (3–15s, 16:9/1:1/9:16 only, 720p, audio default off). prompt required; video_duration, sound, aspect_ratio, resolution as needed. Prohibited: image-driven video generation, audio-driven generation, video editing, applying effects/style changes to video.

meitu text-to-video

Usage Examples

# Scenario 1: Basic text-to-video
meitu text-to-video \
  --prompt "Starry sky timelapse, milky way slowly moving" \
  --json

# Scenario 2: Specify duration and aspect ratio
meitu text-to-video \
  --prompt "City night traffic" \
  --video_duration 8 \
  --aspect_ratio 16:9 \
  --sound on \
  --json

# Scenario 3: Full parameters + download results
meitu text-to-video \
  --prompt "Waves crashing on rocks, slow motion" \
  --video_duration 10 \
  --aspect_ratio 16:9 \
  --resolution 720p \
  --sound on \
  --json \
  --download-dir ./output

Parameters

Parameter	Required	Description
`--prompt`	Yes	Type: string; video content description
`--video_duration`	No	Type: number; 4–15s in high-quality mode, 3–15s in fast mode
`--sound`	No	Type: string; default on in high-quality mode, off in fast mode; options: on / off
`--aspect_ratio`	No	Type: string; all ratios in high-quality mode, 16:9 / 1:1 / 9:16 only in fast mode
`--resolution`	No	Type: string; 480p / 720p in high-quality mode, 720p only in fast mode
`--download-dir`	No	Type: string; download result files to the specified local directory
`--output`	No	Type: string[]; specify output file paths, mapped to data.result.urls order
`--skill-name`	No	Type: string; optional invoking skill name for server-side attribution
`--os-type`	No	Type: string; override mtcc os_type (macos/linux/windows/ios/android/web/pc)
`--json`	No	Print JSON output for easier parsing by scripts or agents