Tencent Cloud Video Effects

Problem

When a user provides an image containing a person and wants a short video generated from a Tencent Cloud video-effects template, manually calling TemplateToVideo usually involves API keys, job submission, async polling, and result parsing. This skill wraps those steps in main.py so an agent can execute the script and return the generated video effect.

How It Works

• Input: image URL or local path, up to 10MB, supporting png, jpg, jpeg, webp, bmp, and tiff.
• Core capability: detects the full human outline in the image, performs cutout, and generates video from the selected effects template.
• Common parameters: template name (e.g. hug), --resolution for output quality, --bgm to add background music, --logo-add for the AI-generated label, and --poll-interval / --max-poll-time for async polling.
• Return mode: by default it submits the job asynchronously and polls until a result is available or the timeout is reached; --no-poll returns only the job ID.
• Execution boundary: if the call fails, the agent should return a clear error instead of inventing recognized content.

Notes

It suits automated workflows where Tencent Cloud Video Effects is enabled and API credentials are configured. It is not a local offline renderer, nor is it suited for images without a recognizable human outline, unmatched templates, or pure image editing. Results depend on the template, resolution, timeout, and asynchronous job status.