opaque-product-photo
Deterministic opaque product-photo-like gradient, texture, shelf and product shape.
Tested: July 17, 2026
Reproducible image compression benchmark
This test uses deterministic image fixtures and the jSquash codec versions currently used by BatchMint. The values are real, reproducible file sizes—not a compression guarantee or a claim about every photograph.
Measured results
| Fixture | Dimensions | Source | Output setting | Final size | Size reduction |
|---|---|---|---|---|---|
| opaque-product-photo | 1600×1200 | JPEGQuality 95 · 565.1 KB | JPEGQuality 82 | 230.3 KB | 59.2% |
| opaque-product-photo | 1600×1200 | JPEGQuality 95 · 565.1 KB | WebPQuality 82 | 316.0 KB | 44.1% |
| transparent-product-card | 1200×1200 | PNGUnoptimised lossless PNG · 1505.5 KB | PNGOxiPNG level 2 | 65.9 KB | 95.6% |
| transparent-product-card | 1200×1200 | PNGUnoptimised lossless PNG · 1505.5 KB | WebPQuality 82 with alpha | 78.6 KB | 94.8% |
transparent-product-card
Deterministic transparent product-card fixture with a translucent ring and opaque badge.
How to reproduce it
Run the command below from the repository root. It regenerates the deterministic fixtures, every output file and the JSON result report.
corepack pnpm --filter @batchmint/local-engine benchmark:contentEncoders: @jsquash/jpeg 1.5.0 (MozJPEG); @jsquash/png 3.1.1 + @jsquash/oxipng 2.3.0; @jsquash/webp 1.5.0 (libwebp)
Download the machine-readable JSON resultsTest boundaries
- Fixtures are generated by a fixed algorithm, so every run starts from the same pixels.
- The table measures encoded bytes, not speed, perceptual quality, decoding, resizing or watermarking.
- Different content can produce very different results; WebP is not guaranteed to be smaller than JPEG.