---

name: media-and-custom-charts description: "Log SwanLab text, HTML, rich media, 3D objects, molecules, and custom pyecharts charts safely." disable-model-invocation: true metadata: disco-role: operating license: Apache 2.0

Media and custom charts

Use this sub-skill when the request is about SwanLab rich media or chart objects rather than scalar metrics.

Start here

1. Read Media reference for constructor signatures, accepted inputs, output files, and log_* wrappers.
2. Read Custom charts for ECharts, swanlab.echarts.Table, and swanlab.plot.
3. Read Troubleshooting when a file path, dependency, or chart object fails.
4. Run the lightweight smoke script for a safe check that skips missing optional media extras.

What this sub-skill owns

• Text and HTML logging.
• Image, audio, and GIF video logging.
• pyecharts-backed custom charts, text tables, and diagnostic chart helpers.
• 3D object and molecule media objects.
• Distinguishing format errors, path errors, type mismatches, and missing optional dependencies.

What it does not own

• Scalar metrics, configs, or define_scalar → experiment-tracking.
• Framework-generated media from callbacks or plugins → integrations-and-plugins.
• Login, mode, host, or environment setup → settings-and-modes.
• Sync, conversion, or API browsing → the other sub-skills in the parent skill tree.

Safe defaults

• Treat swanlab[media] as optional and skip rich media checks when numpy, Pillow, soundfile, moviepy, or rdkit are missing.
• Prefer tiny fixtures for smoke checks; these transformers buffer the full payload in memory.
• Remember that ECharts accepts any object with dump_options().
• Remember that current Video support is GIF-only.
• Prefer swanlab.Text and swanlab.Html for the lightest successful media smoke.