Skip to contents

artbenchmark

artbenchmark computes image-derived metrics for artwork analysis within the Artalytics benchmark framework. It extracts color, coverage, and complexity metrics from artwork images to support portfolio-relative benchmarking across Time & Effort, Skill & Artistry, and Complexity & Detail categories.

Installation 

# Install from GitHub (requires GITHUB_PAT)
pak::pkg_install("artalytics/artbenchmark")

Environment Variables

artbenchmark requires no environment variables. It is a pure computation package with no database or API dependencies.

The package uses only standard R dependencies (data.table, magick, stats, stringr) and operates on local image files.

Quick Example 

library(artbenchmark)

# Compute metrics from an artwork image
metrics <- calc_art_metrics(
    img_path = "/path/to/artwork/main.png",
    file_sz = 1500000,    # bytes (from file.info)
    draw_mins = 120,      # minutes (from Procreate stats)
    strokes = 45000       # count (from Procreate stats)
)

# Returns data.table with metric_key and value columns
print(metrics)
#>          metric_key        value
#>  1: canvas_coverage    87.234567
#>  2: n_unique_colors  8234.000000
#>  3: share_of_spectrum   0.000491
#>  4:  ave_blend_rate      0.817234
#>  5: ave_colors_pstroke  0.183011
#>  6:  q75_color_freq   1234.000000
#>  ... (extended metrics)

# Add identifiers and persist to database
metrics[, `:=`(
    art_uuid = "99a61148-1d3b-4340-8cf6-92ad26046b0f",
    artist_uuid = "746b8207-72f5-4ab6-8d19-a91d03daec3d",
    created_utc = Sys.time()
)]

What artbenchmark Does

artbenchmark is a specialized computation package that:

  • Extracts color palettes from artwork images with background filtering
  • Computes 6 core metrics used in the benchmark framework
  • Computes 8 extended metrics for future analytics
  • Provides predictions for missing metrics when Procreate stats are unavailable
  • Defines the metric framework with authoritative names and categories

Core Metrics Computed

Metric Category Measures
Canvas Coverage Time & Effort % of canvas with color
Unique Colors Skill & Artistry Palette diversity
Spectrum Coverage Skill & Artistry RGB color space utilization
Blend Rate Skill & Artistry Color blending quality
Colors per Stroke Complexity & Detail Brushing precision
Color Variability Complexity & Detail Color frequency distribution

Package Hierarchy 

appPlatform / mod* packages (application layer)
        |
    artutils (data access + benchmarking)
        |
    artbenchmark (metric computation)  ← This package

artbenchmark has no dependencies on other Artalytics packages. It is a pure computation layer used by artutils::update_artist_benchmarks() for portfolio-relative percentile calculations.

Documentation

Key Features

Automatic Background Filtering

Raster extraction filters out non-content pixels: - Transparent pixels - Near-white colors (#f[fedc8a]*) - Near-black colors (#0[012]*)

This ensures metrics reflect actual artwork content, not canvas background.

Prediction Models

When Procreate stats aren’t available (e.g., photos of physical art), estimate metrics from color count:

# Extract color count
rast_data <- get_image_rast("/path/to/artwork.png")
n_colors <- nrow(rast_data$colors)

# Predict missing metrics
predict_strokes(n_colors)
predict_minutes(n_colors)

Predictions use ML models trained on verified artwork data. Actual Procreate stats are always preferred when available.

Metric Framework Reference

Access the authoritative 16-metric benchmark framework:

# Get all metric definitions
get_metric_definitions()

# Check which metrics artbenchmark computes
get_artbenchmark_metrics()

# Lookup display names for UI
get_metric_name("canvas_coverage")  # "Canvas Coverage"

⚠️ Migration Notice

calc_percentile() Deprecated (v0.7.0)

calc_percentile() has been moved to the artcore package to maintain proper package dependency hierarchy.

The function in this package now throws an error with migration instructions. See ?artcore::calc_percentile for documentation.

Why this change? To respect the dependency hierarchy: artcore ← artutils ← artbenchmark

Development

For AI agent instructions and coding standards, see AGENTS.md.

devtools::test()
devtools::check()
pkgdown::build_site()

Proprietary - Do Not Distribute