vizb chart subcommands

Chart subcommands (vizb bar, vizb line, vizb scatter, vizb pie, vizb heatmap, vizb radar) generate one chart at a time. They produce the same output as the root command with --charts limited to that type, but accept only the flags valid for the chart — unsupported flags fail fast.

Use a subcommand for a single chart; use the root command with --charts for multiple chart types from one dataset.

Usage

vizb <chart> [target] [flags]

<chart> is one of bar, line, scatter, pie, heatmap, radar. Output is HTML by default, or JSON when -o ends in .json (same rules as the root command).

vizb bar data.csv -g impl,size -p n,x -o bar.html
vizb pie data.csv -g impl -o pie.html
go test -bench . | vizb line -p n/y -o line.html

Shared flags

Every chart subcommand accepts the common data, grouping, and metadata flags:

Flag	Short	Default	Description
--output	-o	(stdout)	Output file path. .json → JSON, else → HTML
--parser	-P	auto	Parser: auto (detect), go, js:tinybench, js:vitest, rs:criterion, rs:divan, csv, json
--name	-n	Comparisons	Dataset name
--description	-d	""	Dataset description
--tag	-t	""	Tag identifier for release tracking
--id		""	Stable dataset id for ?id= deep links in the HTML UI
--group-pattern	-p	x	Pattern-based grouping (n/x/y/z with separators matching -g for CSV/JSON)
--group-regex	-r	""	Regex-based grouping (named captures)
--group	-g	""	Names each dimension in --group-pattern order; csv/json column names must use matching separators in -p
--select		""	csv/json only: select value columns; optional rename with {label}
--filter	-f	""	Regex to include only matching rows (CSV/JSON: --group label) or benchmark names
--sort	-s	""	Sort order: asc or desc
--swap		""	Axis permutation override, e.g. yx, yxn
--show-labels	-l	false	Show value labels on the chart
--mem-unit	-M	B	Memory unit: b, B, KB, MB, GB
--time-unit	-T	ns	Time unit: ns, us, ms, s
--number-unit	-N	""	Number unit: K, M, B, T

Chart-specific flags

Flag	bar	line	scatter	pie	heatmap	radar	Description
--scale (-S)	✅	✅	✅	—	—	—	Value scale: linear or log
--3d	✅	✅	✅	—	—	—	Pseudo-3D for grouped x+y data (y → depth, metric → height)
--3d-visualmap	✅	✅	✅	—	—	—	Color 3D geometry by metric value
--visualmap	—	—	✅	—	—	—	Color 2D scatter points by metric (off by default)
--3d-rotate	✅	✅	✅	—	—	—	Auto-rotate the 3D scene (only meaningful with 3D data)
bar, line, and scatter render in 3D when the data has a z dimension (-p n/x/y/z) or auto-value detects 3+ numeric columns. Pie, heatmap, and radar are 2D-only for those flags — passing an unsupported flag is an error:

vizb pie data.csv --scale log   # Error: unknown flag: --scale

Examples

# 3D bar with log scale and auto-rotation
vizb bar data.csv -g category,metric,group -p n,x,y,z --scale log --3d-rotate -o bar3d.html

# Pie with value labels, ascending sort
vizb pie data.csv -g impl --sort asc --show-labels -o pie.html

# Line from Go benchmarks via stdin, swapping axes
go test -bench . | vizb line -p n/y --swap yn -o line.html

# Scatter from all-numeric columns (auto-value)
vizb scatter data.csv -o scatter.html

# JSON output for later merging / re-rendering
vizb radar data.csv -g impl -o radar.json

Note

For multiple chart types from one dataset, or per-chart overrides via --chart, use the root command. Subcommands focus on a single chart with a flag set tailored to it.