STEP-BY-STEP GUIDE

How to turn product photos into a video with LuluLab Skills

Give your AI agent a product photo and verified selling points. LuluLab Product Video plans the creative, estimates credits, and generates a video after you approve the configuration.

Get Product Video Skill ↗See all Skills

Before you start

An AI agent that supports local Skills and terminal commands, such as Codex or Claude Code.
Node.js 20 or later. The current Product Video Skill does not require Python or FFmpeg for new tasks.
A LuluLab account, an API key stored locally as LULULAB_API_KEY, available credits, and at least one product photo you have permission to use.

Get your key from the LuluLab Key center. Configure it on your own computer; never paste the key into an agent chat or a shared prompt. Downloading and installing the Skill does not consume LuluLab credits. Planning and generation can.

1. Install and check the Skill

Send this instruction to your compatible AI agent. It should read the repository instructions and install the complete Skill folder in its user Skills directory.

Install https://github.com/LuluLab-AI/LuluLab-Product-Video in my agent's user Skills directory. Read its README, copy the complete Skill folder with SKILL.md and companion files, check that Node.js 20 or later is available, then reload Skills and verify discovery. Do not upload media or submit a paid task yet.

After reloading the agent, ask it to check that LuluLab-Product-Video is available and that the API key is configured without displaying its value.

2. Add your photo and brief

Attach a clear product photo, replace the brackets with real details, and send the following request. The Skill can make 1–20 videos from one product brief; start with one while you learn the flow.

Use LuluLab-Product-Video to make one 15-second, 9:16 product video for a US audience from the product image I attach. Product: [name]. Verified selling points: [facts from the product or packaging]. Audience: [who it is for]. Spoken language: English. Show the complete proposed configuration and the current credit estimate before asking me to approve generation. Do not invent product claims.

Check the proposed audience, creative type, model, duration, output language, and credit estimate. The agent should ask you to approve that specific configuration and estimate before it submits a paid task. If you change the configuration, review the revised estimate.

3. Review and download the result

After approval, the Skill plans the script, creates the video task, checks its status, and downloads completed output. Review product identity, on-screen text, spoken claims, and audio before publishing.

Product video example

This is an existing LuluLab product video supplied for public display. It illustrates the type of output; it was not generated from the sample prompt above. Your result depends on your image, brief, settings, and model.

If something does not work

Skill not found: Check that the installed folder itself contains SKILL.md, then reload your agent.
API key missing: Create or copy a key in the Key center and configure LULULAB_API_KEY locally.
Estimate unavailable or insufficient credits: Resolve the issue before approving a paid task. Do not treat an unknown estimate as zero.
Task interrupted: Ask the agent to resume its saved task state. Do not submit the same paid request again until its status is known.

For command details and current limitations, read the official Skill README. You can also contact LuluLab support.