DreamLake
Browse docs

Visualize signals and recordings

@dreamlake/viz is a React library for time series, recorded media, and dataset viewers. Use it to inspect measurements, compare signals, or view a recording's cameras and annotations at the same time.

  • Plot arrays. Pass timestamps and values to a chart; add series, uncertainty bands, markers, and zoom controls as needed.
  • Compare synchronized views. Share a time cursor across charts, video, image frames, and labeled intervals.
  • Inspect datasets. Use a .dreamrc to map fields from a supported recording format to views. For individual files, use the file-preview components.

Your first chart

This chart contains four force measurements. Move the pointer to read a sample; expand the code to see the complete React component.

(N)
86420-2
0s0.5s1s1.5s2s2.5s3s(s)
Code ForcePlot.tsx

x contains time in seconds; y contains force in newtons. Both arrays have one entry per sample. duration={3} sets the time extent, and the containing div gives the chart its height. The parent stores the pointer time and sends it back through time to display the readout.

The readout uses a recorded sample. The line connects samples; it does not create additional measurements between them. Next, add series and interactions to the same chart.

For the original discussion of data queries, aggregation, and presentation, read the Visual Grammar notes on Vuer-Viz.

Put views on the same clock

Keep time in the parent component and pass it to each view. A pointer can preview a time while a selected playhead stays fixed. The synchronization example connects charts, a timeline, and video using this state.

Use timelines for labeled intervals, video or image frames for cameras, and overlays for boxes and keypoints. Parallel coordinates compare records across several metrics instead of time.

Open your data

InputStart here
Arrays and media URLs already in your appCharts and media
A supported LeRobot, UMI, MCAP, or folder datasetView a dataset
A single table, image, recording, or JSON filePreview a file

Dataset readers support specific layouts and encodings. Check the format requirements before adapting a recording.

Install the package

Install the public npm release into a React application:

bash
npm install --save-exact @dreamlake/viz@0.32.3 @dreamlake/uikit@0.2.11

The package includes ESM and TypeScript declarations. React, React DOM and UIKit (^0.2.11) are peer dependencies. Parallel coordinates use UIKit's shared list selection and virtualization through its focused /list entry. Components use Tailwind v4 utilities. In a stylesheet at src/styles.css, import the Viz styles and register its published source:

css
@import "tailwindcss";
@import "@dreamlake/viz/styles.css";
@source "../node_modules/@dreamlake/viz/src";

Adjust the source path relative to your stylesheet. Set data-theme="dark" on the document element for dark mode. Import charts from @dreamlake/viz or @dreamlake/viz/episode-line-chart, dataset readers from @dreamlake/viz/dataset-viz, and file previews from @dreamlake/viz/file-preview.

For a small embedded chart, use size="compact" (150px total height) or set an explicit height. Charts and timelines have transparent backgrounds by default; background="subtle" opts into a panel tint. The compact example combines four charts with shared time and viewport state beside a video.

Run the examples

The viz-workspace repository uses local TypeScript source exports for development. From a checkout:

bash
pnpm install
pnpm --filter docs dev

Maintainers can follow the repository's release and consumer integration guide for npm tarball verification and documentation deployment.

Existing sources / panels integrations use the separate Schema API. Markdown exports make these docs available to coding tools.