SparkpondSparkpond · Image Tools
Reproducible benchmark
Measure real local encodes on this device without invented marketing numbers.
Run benchmark
- Corpus
- benchmark-corpus-v1 · CC0-1.0
- Engine
- 6.0.0
Versioned and licensed corpus
Benchmark corpus v1 contains three synthetic samples generated by Sparkpond: a photo-like gradient, a sharp interface screenshot, and a transparent illustration. The corpus recipe is versioned in the repository and released under CC0-1.0 so every run starts from the same public inputs.
Reproduce on this browser
Use the Run benchmark button below. Sparkpond creates the corpus locally, encodes it through the production Worker, verifies each output, and reports measured source size, result size, format, local duration, browser environment, engine version, and run date.
Interpretation
Results vary by browser, hardware, codec cache state, and system load. Technical similarity can rank candidates but never proves that a difference is invisible to every person. Compare runs only when their corpus, engine, and environment are recorded.