Generate images, videos, and digital humans via Aliyun Bailian DashScope APIs.
npx skills add https://github.com/openakita/openakita --skill happyhorse-video
Please help me install this Agent Skill. Skill: happyhorse-video Source: https://github.com/openakita/openakita/tree/main/plugins/happyhorse-video Command: npx skills add https://github.com/openakita/openakita --skill happyhorse-video
Producing AI-generated images, videos, and digital-human content on Aliyun Bailian requires juggling many model endpoints, protocols, and billing rules; this Skill unifies them into one studio with 22 LLM-callable tools and a storyboard-driven long-video pipeline.
Ask the agent to generate a 5-second 720P text-to-video clip of a sunrise over the sea using the happyhorse-video tools after configuring your DashScope API key and OSS settings.
High-intent search queries and answers about installing and using this skill.
Call the hh_t2v tool with a prompt, duration, and resolution; it submits a HappyHorse 1.0 or Wan 2.6 job to DashScope and returns a task_id plus video_url when done. Use hh_cost_preview first to estimate the per-second price before submitting.
Use hh_storyboard_decompose to split a story into structured segments with prompts and transitions, then pass them to hh_long_video_create for serial or parallel generation, and finally merge downloaded clips with hh_video_concat using ffmpeg crossfade.
Five modes are available: photo_speak (wan2.2-s2v), video_relip (videoretalk), video_reface (wan2.2-animate-mix), pose_drive (wan2.2-animate-move), and avatar_compose (wan2.7-image followed by s2v). Voice audio comes from CosyVoice-v2 or the free Edge-TTS engine.
DashScope enforces a concurrency cap of one async task per API key, so submissions are serialized through a semaphore and queue up. Also note task_ids expire after 24 hours, and HappyHorse 1.0 rejects legacy parameters like with_audio, size, quality, fps, and audio.
Yes. Every hh_image_* tool publishes results as asset_ids on the Asset Bus, and video tools accept from_asset_ids which map to first frames, reference images, or source videos depending on the mode, enabling image-to-video pipelines without re-uploading.
smilebank7Transcribes audio files into timestamped, human-readable text.
gaoqiongxieGenerate animated MP4 and WebM videos using React and Remotion.
bytedanceCoordinate web-search, image-generate, and video-generate tools to create storyboard images and videos.
ShowTimeWalkerGenerate images and videos via Doubao Seedance and save outputs locally.
