Iris
Iris-WebP is a WebP encoder tuned for how people see images. At the same encode time as libwebp's default setting, it produces files up to 27% smaller at equal quality. The output is standard WebP and decodes everywhere WebP already does.
A Better WebP Encoder
WebP is supported by every major browser, and large sites serve billions of WebP images every day. Iris-WebP keeps the format and replaces the encoder. Switching means smaller files at the quality you already ship, or higher quality at the file sizes you already pay for, with no change to decoders, CDNs, or markup.
Speed
Iris-WebP has seven speed settings. Its slowest encodes a Kodak image in 72 ms against 118 ms for libwebp's slowest, and its fastest runs in 13 ms.
Efficiency
At matched speed, Iris-WebP beats libwebp on every perceptual metric we track: CVVDP, SSIMULACRA2, MS-SSIM, IW-SSIM, and even VMAF.
Tradeoffs
Iris-WebP's default perceptual tune is not as performant on legacy metrics (SSIM, PSNR).
See the difference
Each image is encoded to the same file size. The captions list the size and metric scores. Higher is better for SSIMULACRA2 and CVVDP, and lower is better for Butteraugli.
Preserving fine detail
Switch between encoders with the buttons. libwebp smooths away the texture in the trees, while Iris-WebP and libjpegli keep it.
Encoded Feb 2026
Avoiding distortion
The opposite happens here. libjpegli distorts flat regions badly, and both WebP encoders hold up. This image shows why we do not tune for one metric or one image.
Encoded Feb 2026
Metrics
The numbers below were aggregated using the Kodak Lossless True Color Image Suite at medium to high fidelity.
Measured Aug 2026
Using Iris-WebP
Iris-WebP is licensed commercially. A license includes the encoder, new releases, maintenance fixes, and direct support from us.
If you want to test it on your own images before deciding, email us with a sample and the encoder settings you use today. We will send back a size and metric comparison.
Test Iris-WebP on your images
Send us a few representative images and your current encoder settings. We will reply with file sizes and metric scores from Iris-WebP.