October 07, 2026 · The Halide Team

Introducing Iris-AVIF

Our most efficient image encoder, and the best image encoder in the world.

Read More Iris page
21% Better than libaom at equal quality (CVVDP) with faster encode time
20% Better than Iris-WebP at equal quality and speed
29× Faster than libaom at 1% better quality (CVVDP)

Measured Oct 2026 on the Kodak Lossless True Color Image Suite.

We’re introducing Iris-AVIF (formerly codenamed Aperture), the world's best image encoder.

Over a year ago, our research began with Iris-WebP. Extending our methodology and the techniques we've discovered to a vastly larger set of coding tools brings us to Iris-AVIF, where we've achieved state of the art results and produced a world-class image encoder that represents the frontier of image compression efficiency.

A generational improvement

Competing with libwebp via Iris-WebP required discovering new psychovisual research and engaging in rigorous performance engineering. The best open-source AV1 encoders pushed our efforts even further; libwebp stopped improving a long time ago, while the best AV1 encoders continue to improve today. The two leading encoders (SVT-AV1 and libaom) are developed by massive companies with incredible resources, and lots of attention is put into AVIF.

To compete at the frontier of image compression with Iris-WebP as a base, we had to go beyond what we initially thought was possible and experiment liberally with AV1's broad set of intra-coding tools to produce the best AVIF images. AV1's intra-coding toolset is wildly versatile; we've seen and even worked on big improvements before.

Iris-AVIF is the culmination of our best work. It represents years of multi-disciplinary effort: perceptual fidelity work, new tools to measure improvements, new features WebP could never deliver, and setting a new pace for our research and engineering. The result is the world's most versatile, capable, and efficient image encoder.

AVIF

AVIF has been considered widely supported since January 2024. Support is guaranteed on >95% of all Web browsers, just 1.47% behind WebP and 1.67% behind JPEG. Decode speed is also fast, not far from WebP or JPEG.

Iris-AVIF is also a unique opportunity for us to focus beyond raw compression efficiency on new problems AVIF lets us solve: accurate film grain via synthetic reproduction and high-bitdepth HDR encoding stand out, and are not available in WebP. Because of these tools, Iris-AVIF is a uniquely intelligent encoder.

Automatic grain synthesis

Iris-AVIF intelligently detects whether an image contains a grain pattern it can reproduce synthetically, and then determines exactly how much. It can flexibly emulate anything from coarse film grain to fine digital chroma noise, without manual intervention.

HDR

HDR allows images to display a wider range of color and light, further extending an image's ability to saturate the infinite potential of the human eye. HDR isn't properly supported in JPEG and WebP, requiring inefficient layering tricks.

AVIF supports HDR natively, and we're excited to support it in Iris-AVIF. Iris-AVIF's performance is a highlight; notice below how Iris retains the lines on the bottom building better than SVT-AV1 while being 5 times faster.

Screen content encoding with Iris-AVIF offers up to 415% better compression than JPEG, and other areas like progressive decoding are yet to be explored by us. Whatever images need, Iris-AVIF aims to uniquely deliver via detection, adaptation, and efficient reproduction. While Iris-WebP is still versatile, Iris-AVIF is intelligent in its ability to understand and adapt to the content it encounters.

Perceptual alignment

Iris-AVIF is our most perceptually optimized encoder, and places the preferences of the human eye first. We ran into some of the limitations of perceptual metrics early in its development, and as Iris-WebP achieved perceptual improvements that metrics failed to reward, it became clear a strong perceptual tune should be available upon release. We took extra time to make sure it was up to our standards before this announcement.

Perceptual fidelity remains top of mind in our mission to produce a more beautiful Web, but we understand that subjective methodology has limitations. Iris-AVIF has impressed us, but we believe it can be even better. Our work continues to focus on this aspect of the encoder, and we'll always strive to keep improving.

Looking Forward

Iris-AVIF has already been under evaluation since before this announcement. Iris-WebP remains mature, available, and continually improving. Looking ahead, we'd like to strengthen our perceptual tunes in both encoders and implement strong progressive decoding in Iris-AVIF, among other things. For those who value WebP, expect Iris-WebP to improve in tandem with Iris-AVIF.

We're very proud of what Iris has become, and we're excited to see even more users take advantage of our most capable encoder. Iris-AVIF is our most ambitious bet to bring the world a more beautiful Internet, and we can't wait to see where it takes us all next.

Send us an email if you're interested, we'd love to hear from you.

Email Us