AI Agent Hub
Back to skills
Jimeng OmniHuman 1.5 Digital Human Video icon

Jimeng OmniHuman 1.5 Digital Human Video

Design & Media Updated 2026.08.29

Paste the following prompt into your AI chat to install this skill:

Please install @user_f95f21b8/dlazy-jimeng-omnihuman-1-5 following https://skillhub.cn/install/skillhub.md.

About this skill

Problem

Digital human broadcast videos usually require connecting portrait images, audio or text, lip-sync, and output video into one generation pipeline. This skill wraps Jimeng OmniHuman 1.5 behind the dlazy jimeng-omnihuman-1.5 command, reducing fragmented work around prompts, media uploads, and task polling. It fits engineering and content workflows that need quick digital-human clips, virtual-host previews, or media experiments.

How It Works & Limits

The skill is a thin client over dLazy’s hosted API:
- Inputs: portrait image, audio or text, prompt, and other parameters are sent to api.dlazy.com through the dlazy CLI.
- Media handling: local file paths are uploaded to files.dlazy.com, and generated outputs are returned as hosted URLs; this is not local inference.
- Task mode: by default it waits for results; with --no-wait it returns a task object and generation ID, requiring status polling.
- Credentials: requests require a dLazy API Key, supplied via CLI config or DLAZY_API_KEY; errors such as 401, insufficient balance, missing parameters, and local file read failures should be handled by returned codes.

Note: the skill does not add new video capabilities to Jimeng itself. It mainly standardizes invocation, error reporting, and async task flow. For real-person portraits, commercial broadcasting, or public release, follow platform terms, asset licenses, and safety policies.

Use Cases

  • E-commerce content: generate product-explainer broadcast clips from audio and host portraits for detail pages or live-stream warm-ups.
  • Training content: synthesize instructor portraits with narration audio into digital-human lesson clips, reducing repeated live recordings.
  • Marketing tests: swap prompts and audio to produce virtual-host teasers, comparing lip-sync and presentation quality.
  • Remote demos: prepare local portraits, script audio, and prompts, then submit async CLI tasks and poll result URLs for sharing.

Best For

  • Digital-human video engineers: integrating portraits, audio, and prompts into Jimeng OmniHuman 1.5 with async task handling.
  • E-commerce or content operators: producing weekly product-explainer clips from host portraits and scripts for ad testing.
  • Course producers: converting narration audio and instructor portraits into lesson clips without repeated live recording.
  • dLazy API developers: building digital-human generation pipelines around the dLazy CLI and error codes.