Smaller images, better quality
Halide Compression builds image encoders and the tools to measure them. Our first product, Iris-WebP, produces files 27% smaller than the reference WebP encoder at equal visual quality and equal speed.
Frontier Compression Engineering
We write encoders, we measure them with perceptual metrics, and we help teams put the results into production. Our tuning targets what the human eye sees.
Encoder engineering
Encoders written from scratch, and speed and efficiency work on existing codecs in C, Zig, and assembly.
Perceptual quality
We evaluate every change against the best percetpual metrics.
Consulting
On-site and remote consulting on encoder selection, tuning, and image & video pipelines at scale.
Iris-WebP
A WebP encoder tuned for the human visual system. Compared to libwebp, the reference encoder, Iris-WebP produces smaller files at the same visual quality and the same encode time. Full results are on the Iris page.
From the blog
Encoder write-ups, metric comparisons, and interviews with people who work on compression.
Limitations of Perceptual Metrics
As encoders mature, metrics become less valuable.
Read moreIntroducing fmetrics
Faster perceptual video and image fidelity metrics.
Read moreSame Image, Different Score?
Metrics don't just measure encoder performance; decoding matters too, and even after decoding, not all post-processing is created equal.
Read moreTalk to us
We license Iris-WebP and take on consulting work. Send us a note about your image pipeline and we will reply.
