Sign in

Gianni Rosato

@giannirosato.com
168 followers 183 following 76 posts

Multimedia Compression Engineer (mainly image encoders) Mastodon: @gianni@disobey.net Web: giannirosato.com Work: halide.cx

PostsRepliesMedia
Reposted by Gianni Rosato
WebPerformance Report @webperfreport.bsky.social · 20/09/2026
⚡Featured posts this week: ✨ New ad metrics in CrUX: @alextcone.bsky.social ✨ The case against JPEG XL: @giannirosato.com ✨ Why Spotify Is Not Using Bayesian A/B Testing: Mattias Frånberg, Mårten Schultzberg ✨ International SEO: @polemicdigital.com 🙏 Special thanks to the authors! #webperf
WebPerformance Report performance email preview highlighting optimization scores, Perf Links, and Core Web Vitals metrics on a colorful gradient background.
131
Reposted by Gianni Rosato
Joan León @nucliweb.net · 20/09/2026
The case against JPEG XL Investigating JPEG XL's place as a Web image codec, by @giannirosato.com, via @webperfreport.bsky.social giannirosato.com/blog/post/ca... #JPEGXL #JXL #ImagePerf #WebPerf
A dark-themed line chart titled “SSIMULACRA2,” showing BD-Rate (%) versus encoding speed in milliseconds for several image codecs. Colored lines compare codecs including AOM, JPEG XL, aperture-alpha, Iris-WebP, libjxl, SVT-AV1, and libwebp, with labeled data points. Most curves show higher BD-Rate at slower speeds, while some reach negative BD-Rate at faster speeds.
032
Gianni Rosato @giannirosato.com · 15/09/2026
Author here – I completely agree! New image formats need very strong justification, otherwise they aren’t worth including. JXL’s justification has been “nice to have” while AVIF saves the most bandwidth for the Web’s #1 (by far) kind of image (lossy)
130
Gianni Rosato @giannirosato.com · 15/09/2026
Read the article – AVIF supports progressive decode, and “backwards compatibility” isn’t correct. You can re-encode JPEGs to JXL losslessly to save space at the expense of decode time.
110
Reposted by Gianni Rosato
Jake Archibald @jakearchibald.com · 14/09/2026
This is a great article. I've always been sus about the hype around JPEG XL, as it never stacked up when I actually tested it with web use-cases. giannirosato.com/blog/post/ca...
giannirosato.com
The case against JPEG XL | Gianni Rosato
Investigating JPEG XL's place as a Web image codec.
25111
Gianni Rosato @giannirosato.com · 14/09/2026
HN thread: news.ycombinator.com/item?id=4969... lobsters thread: lobste.rs/s/e1lcnf/cas...
news.ycombinator.com
The case against JPEG XL | Hacker News
010
Gianni Rosato @giannirosato.com · 14/09/2026
Might be one of my more controversial posts: "The case against JPEG XL" Link: giannirosato.com/blog/post/ca... 🫡 #compression #jpegxl #av1
giannirosato.com
The case against JPEG XL | Gianni Rosato
Investigating JPEG XL's place as a Web image codec.
164
Gianni Rosato @giannirosato.com · 07/09/2026
New Halide blog post on some of the weaknesses of perceptual metrics: halide.cx/blog/metric-... These lessons are helping us develop our next encoder, Aperture. #compression
lnkd.in
LinkedIn
This link will take you to a page that’s not on LinkedIn
010
Gianni Rosato @giannirosato.com · 04/09/2026
Maybe more relevant link for the origins of tune=iq is svt-av1-psy.com/avif/ Shoot me an email any time if you have any questions!
svt-av1-psy.com
SVT-AV1-PSY
120
Reposted by Gianni Rosato
Jake Archibald @jakearchibald.com · 25/08/2026
It's really weird that the response to JPEG XL shipping is "yesss fuck WebP!" Like, we've had AVIF for years, and it outperforms JPEG XL in almost all web use-cases.
261048
Gianni Rosato @giannirosato.com · 10/07/2026
Hey everyone! fmetrics, a fast open-source perceptual metrics toolkit that runs on the CPU, is here: github.com/halidecx/fme... fmetrics supports IW-SSIM, MS-SSIM, CVVDP, SSIMULACRA2, & Butteraugli. See the blog post (halide.cx/blog/fmetrics) for more info. All code is Apache 2.0. Enjoy!
github.com
GitHub - halidecx/fmetrics: Fast image & video fidelity metrics in C, Zig
Fast image & video fidelity metrics in C, Zig. Contribute to halidecx/fmetrics development by creating an account on GitHub.
191
Gianni Rosato @giannirosato.com · 18/06/2026
Making Software: Image Compression www.makingsoftware.com/chapters/ima... Cool to see WebP addressed here too!
makingsoftware.com
Making Software: Image compression.
Exploiting the quirks of human vision to make images smaller, without looking awful.
040
Reposted by Gianni Rosato
Rachel Barker @rachelplusplus.me.uk · 12/06/2026
I've been improving my image compression project on and off for the last six months or so. Today I got it to basically where I want it feature-wise, so I've put out an alpha release and written a quick little blog post about it: www.rachelplusplus.me.uk/blog/2026/06...
rachelplusplus.me.uk
Tinyavif v2.0 alpha - Rachel++
I've put out a preview of the next version of tinyavif. It's basically feature-complete to what I want v2.0 to be, and I've run some quick correctness tests, so it's ready for experimental use.
021
Gianni Rosato @giannirosato.com · 28/05/2026
The owner of the x266.mov domain lost access, and we subsequently lost the wiki subdomain. Now we're back up under a new domain – there are a couple of mirrors, but codecs.wiki is the only official source I endorse at this time.
codecs.wiki
Codec Wiki
Your Guide to All Things Encoding
010
Gianni Rosato @giannirosato.com · 28/05/2026
The Codec Wiki has moved from wiki.x266.mov to codecs.wiki – apologies for the down-time!
wiki.x266.mov
141
Gianni Rosato @giannirosato.com · 16/03/2026
Some visual comparisons are here: halide.cx/iris/
halide.cx
Iris | Halide Compression
011
Gianni Rosato @giannirosato.com · 16/03/2026
If you're interested in a much better WebP encoder, let me know ... mail [at] halide [dot] cx Iris-WebP is up to 50% better than JPEG, and even outperforms AVIF encoders like SVT-AV1 in perceptual image metrics. WebP FTW :)
Iris-WebP encoder performance on a BD-rate to speed graph. Encoders closer to the top left of the graph are better. Iris-WebP is at the Pareto front for image compression in SSIMULACRA2, a perceptual image metric with high correlation with subjective quality.
121
Gianni Rosato @giannirosato.com · 12/03/2026
fcvvdp 0.2.0! github.com/halidecx/fcv... Highlights: ~9-12% speed boost for single-threaded, task threading (220% faster than the reference implementation with 8 threads on my 13700k), FFmpeg support via patch. Check it out and lmk what you think!
010
Reposted by Gianni Rosato
Ignacio Castano @icastano.bsky.social · 10/02/2026
I've put together an updated version of the Sponza scene with uncompressed PNG and compressed AVIF textures. I wrote about the process and compared the results against KTX. www.ludicon.com/castano/blog... #webgpu #web3d #sparkjs
ludicon.com
An Updated Sponza glTF – Ignacio Castaño
3132
Gianni Rosato @giannirosato.com · 05/02/2026
New Halide blog post on post-processing for chroma as it is relevant to compression efficiency: halide.cx/blog/chroma-... Seems like you can get a roughly 2% efficiency improvement with 4:2:0 decoder output just with smarter post-processing for chroma. Pretty neat! #avif #webp #jpeg
halide.cx
Same Image, Different Score? | Halide Compression
Metrics don't just measure encoder performance; decoding matters too, and even after decoding, not all post-processing is created equal.
040
Reposted by Gianni Rosato
damarberlari.bsky.social @damarberlari.bsky.social · 22/01/2026
the 2nd part of my dithering visual article is finally out! 🔗 visualrambling.space/dithering-pa... this one mainly explores the threshold map and how it generates those unique visual patterns hope you enjoy this as much as I enjoyed making it! made with #threejs & #animejs
714129
Gianni Rosato @giannirosato.com · 24/01/2026
SVT-AV1 4.0.0 is here! Changelog: gitlab.com/AOMediaCodec... SVT-AV1-PSY's porting efforts are fully complete, and this release also features huge AVIF improvements. Maybe the most exciting release to date by my standards – SVT-AV1 is finally a real perceptual encoder! #av1 #avif #compression
gitlab.com
v4.0.0 · Alliance for Open Media / SVT-AV1 · GitLab
[4.0.0] - 2026-1-23 API updates Major release with...
081
Gianni Rosato @giannirosato.com · 24/12/2025
Happy Holidays! Introducing fcvvdp, a faster CVVDP implementation: github.com/halidecx/fcv... Compared to the reference implementation running on the CPU, fcvvdp is faster (~15%) despite being single-threaded. It sees stronger wins with user time (up to ~15x) and memory usage (>90% less memory).
github.com
GitHub - halidecx/fcvvdp: Fast CVVDP implementation in C
Fast CVVDP implementation in C. Contribute to halidecx/fcvvdp development by creating an account on GitHub.
030
Gianni Rosato @giannirosato.com · 11/12/2025
Here’s the talk: youtube.com/watch?v=NtLHRX4uFhw
010
Reposted by Gianni Rosato
Julio Barba @juliobbv.bsky.social · 05/12/2025
SVT-AV1-HDR 3.1.3 "Cyclonus" is out! Including all-new features from SVT-AV1, plus code cleanups and stability fixes. Get the newest version of this all-purpose AV1 video encoder! ffmpeg and HandBrake builds are available by clicking the link below: github.com/juliobbv-p/s...
github.com
GitHub - juliobbv-p/svt-av1-hdr: An all-purpose AV1 video encoder designed for visually-optimal SDR and HDR image and video encoding.
An all-purpose AV1 video encoder designed for visually-optimal SDR and HDR image and video encoding. - juliobbv-p/svt-av1-hdr
041
Gianni Rosato @giannirosato.com · 22/11/2025
Enhancing HDR on Instagram for iOS With Dolby Vision: engineering.fb.com/2025/11/17/ios/e… #av1 #vp9 #hdr
021
Gianni Rosato @giannirosato.com · 05/11/2025
Oh that's awesome to hear! Considering we also get user-defined QMs in AV2 (that's the one I'm most excited for, thanks Julio) I'm expecting AV2F (or whatever it will be called, if it happens) to be very good
120
Gianni Rosato @giannirosato.com · 04/11/2025
I spoke with @juliobbv.bsky.social at VideoLAN Dev Days in London! Here are our slides, discussing SVT-AV1-PSY: giannirosato.com/files/svt-av... I'll find access to the video & post it soon!
giannirosato.com
242
Gianni Rosato @giannirosato.com · 14/10/2025
Thanks @ericportis.com – glad you’re enjoying my work :)
010
Gianni Rosato @giannirosato.com · 12/10/2025
Introducing oavif: faster target quality image compression giannirosato.com/blog/post/oa... oavif can be up to 63% faster than traditional target quality encoders. Learn how in the attached blog post! #av1 #avif #compression
giannirosato.com
Introducing oavif | Gianni Rosato
oavif is a new approach to target quality encoding in image compression, designed around smarter convergence strategies and quicker scoring to be as fast as possible.
030
Gianni Rosato @giannirosato.com · 03/10/2025
fssimu2 has its first release! github.com/gianni-rosat... fssimu2 is a fast SSIMULACRA2 implementation in Zig, with support for distortion map output. It is ~23% faster than the reference implementation at ~40% less memory usage, while being 99.97% accurate. Give it a look if you're interested!
github.com
Release 0.1.0 · gianni-rosato/fssimu2
First release! 🎉 Check README for more information. What's Changed generate LUT at compile time by @damster101 in #1 Zig 0.15.1 by @gianni-rosato in #3 feat: remove all simd in favor of scalar imp...
030
Gianni Rosato @giannirosato.com · 15/09/2025
New Halide blog post! halide.cx/blog/consist... Quality and speed define an image encoder's compression performance. Consistency is a close third, and easily overlooked in image encoder design. What value does it provide, and how can we measure it? #avif #webp #jxl #jpegxl #compression
halide.cx
Measuring Image Encoder Consistency | Halide Compression
Quality and speed define an image encoder's compression performance. Consistency is a close third, and easily overlooked in image encoder design. What value does it provide, and how can we measure it?
120
Gianni Rosato @giannirosato.com · 15/09/2025
The Halide Compression website (halide.cx) now has an Atom feed – just paste the URL into your favorite RSS reader and you should be good to go!
halide.cx
Measuring Image Encoder Consistency | Halide Compression
Quality and speed define an image encoder's compression performance. Consistency is a close third, and easily overlooked in image encoder design. What value does it provide, and how can we measure it?
020
Reposted by Gianni Rosato
Ignacio Castano @icastano.bsky.social · 09/09/2025
New blog post about my experiences using spark.js with three.js: www.ludicon.com/castano/blog... Re-posts and feedback appreciated!
ludicon.com
three.js + spark.js – Ignacio Castaño
082
Gianni Rosato @giannirosato.com · 09/09/2025
Is the source code for your site available anywhere?
100
Gianni Rosato @giannirosato.com · 30/08/2025
Anyone using rss/atom readers can now type `giannirosato.com` into their reader to get my website's atom feed
020
Gianni Rosato @giannirosato.com · 29/08/2025
I got to interview Julio Barba from Google for the Halide blog: halide.cx/blog/julio-b... Go check it out if you're interested in what the next decade of multimedia tech might look like!
halide.cx
An Interview With Julio Barba | Halide Blog
010
Gianni Rosato @giannirosato.com · 29/08/2025
Completely redesigned my website with Zola (www.getzola.org) – check it out at giannirosato.com Let me know if any links are broken if you link to my site – it should be fully backwards compatible with my old link hierarchy
getzola.org
Zola
Everything you need to make a static site engine in one binary.
020
Reposted by Gianni Rosato
Luc Trudeau @luctrudeau.bsky.social · 01/08/2025
The “AV1 is a disaster” myth Here’s what’s actually true: - SVT-AV1 is fast, fully multithreaded and packed with SIMD. - It scales beautifully on modern CPUs. - It beats x264, x265, and VP9 in bitrate vs runtime. AV1 on CPU isn’t a disaster. It’s quietly winning.
051
Gianni Rosato @giannirosato.com · 22/08/2025
Tune 3: Still Picture is coming to SVT-AV1: gitlab.com/AOMediaCodec/SVT-AV1/-/m… Formerly Tune 4 in SVT-AV1-PSY. Thank you @juliobbv.bsky.social !
032
Gianni Rosato @giannirosato.com · 09/08/2025
I have the same hope … let’s see if it pans out 🤞
000
Gianni Rosato @giannirosato.com · 08/08/2025
Yeah I think they do – JPEG XL is great, and I’m happy it’s seeing more adoption on the web! I remain worried about its status since Google removed support from Chrome, but since Apple added support systemwide I am relieved to see that driving adoption
110
Gianni Rosato @giannirosato.com · 07/08/2025
:( hope you and the computer both get better soon
110
Gianni Rosato @giannirosato.com · 30/07/2025
Exciting #avif news – libaom’s tune iq (derived from SVT-AV1-PSY’s Tune 4) is used across The Guardian’s website now. They say they serve mostly AVIF, with JPEG XL shipped when supported. Very exciting stuff! Thanks @juliobbv.bsky.social!
“[sorry for the spam] Just wanted to say that theguardian.com is on `tune=iq` for the last two weeks (we serve AVIF to non-JXL-supporting browsers) and we couldn’t be more happy. Encoder consistency is much, much higher (still not as good as JXL, but that’s just an impression, not a rigorous assessment): some highly compressible images where delicate detail was being obliterated grown even ~3× in filesize and look much better while unnecessarily weighty images shrank, mostly without any perceptible degradation. Interestingly, we kept the same quality values (which differ by dpr) and as much as we can say, the overall bandwidth stayed more or less the same (I can’t see any change in the graph), so it’s likely the massive reduction of rare weighty images offset small images growing.

Huge thanks to juliobbv for his work on `tune=iq` and to whomever else worked on it. And to friends at our image CDN (Fastly). And to everyone here as without this server, I wouldn’t even know about it…
As always, we welcome improvements to any and all encoders. So the idea of “splitting Julio” couldn’t sound more exciting 😉.”
360
Gianni Rosato @giannirosato.com · 29/07/2025
Post:
“This is one of the craziest ideas l've ever seen. He converted a drawing of a bird into a spectrogram (PNG -> Soundwave) then played it to a Starling who sung it back reproducing the PNG.”

Response:
“Ok fine. Sure. Whatever. We can save PNGs to birds. I give up. You win world”
020
Reposted by Gianni Rosato
Rachel Barker @rachelplusplus.me.uk · 19/07/2025
New post! After my previous post about image compression, a few people asked me to try different encoder settings. So I went ahead and tested a couple of them :) www.rachelplusplus.me.uk/blog/2025/07...
rachelplusplus.me.uk
A Better Image Compression Comparison - Rachel++
141
Gianni Rosato @giannirosato.com · 14/07/2025
Improving AVIF in Open Source via the Halide blog halide.cx/blog/improvi... I wanted to write this to explain some of what went into Tune 4/Tune IQ, & help outline some of my reasoning for starting Halide Compression in the first place. I hope you enjoy! #av1 #avif #encoding
halide.cx
Improving AVIF in Open Source | Halide Blog
012
Gianni Rosato @giannirosato.com · 09/07/2025
More info: - The graph uses libaom's tune SSIM (which is what most CDNs use). Tune iq is definitely more efficient. - I want to make Halide Compression into a real company that can sustain itself. If you know of anyone who would like to license Iris, let me know!
000
Gianni Rosato @giannirosato.com · 09/07/2025
Introducing Iris-WebP, a better WebP encoder: halide.cx/iris/index.h... Iris offers up to >20% better compression than libwebp, and trades blows with libaom for AVIF. My goal is to build an image-first encoder ecosystem for the modern Web, so that new web image codecs aren't generally superfluous.
Iris-WebP BD-rate performance, showing speed-to-quality efficiency roughly on par with standard libaom.
272
Reposted by Gianni Rosato
Christian Timmerer @timmerer.com · 03/07/2025
AV1 @ Scale: Film Grain Synthesis, The Awakening netflixtechblog.com/av1-scale-fi...
netflixtechblog.com
AV1 @ Scale: Film Grain Synthesis, The Awakening
Unleashing Film Grain Synthesis on Netflix and Enhancing Visuals for Millions
022