Notices by Gianni Rosato (gianni@disobey.net), page 4

New blog post is up: QOI Encoding in Zig!

https://giannirosato.com/blog/post/qoi-zig/

#programming #codecs #images

Aviator v0.6.0! https://github.com/gianni-rosato/aviator/releases/tag/0.6.0

> Aviator has been migrated to use newer libadwaita widgets, 5.1(side) audio encoding is fixed, and variance boost updates and optimizations in SVT-AV1-PSY improve encoding efficiency, eliminate artifacts, and provide significant speed boosts across various presets.

Download on Flathub with the link below!

https://flathub.org/apps/net.natesales.Aviator

I should clarify - I don't think Zig is easy, I think falling in love with it is. Zig is hard, imo, lol

Hey, just wanted to shamelessly plug a recent project of mine, a freestanding QOI encoder in Zig: https://github.com/gianni-rosato/qoi-enc-zig

I must say, I really like the QOI image codec despite its shortcomings for coding efficiency and flexibility - it is dead simple, easy to implement, and super effective for what it is! I'd be happy in a world where we used QOI instead of PNG for a lot of stuff. And I've already fallen in love with Zig, which I think is very easy :)

#codec #programming #zig

@eb @samlavigne ah, thanks! Followed!

FFmpeg Explorer: A tool to help you explore FFmpeg filters

https://ffmpeg.lav.io/

This is SUUUPER cool!!! I'm very impressed, good work Sam Lavigne! (not sure if they have a mastodon or not, didn't see one linked on their site)

#ffmpeg #encoding #video

@Firepal3D 🔥🔥🔥

JPEG-XL and the Pareto Front

A closer look at libjxl 0.10

https://cloudinary.com/blog/jpeg-xl-and-the-pareto-front

#photography #codec #jpegxl #jxl

I don't know where to contact the author, but if you want more information, let me know!

https://www.msn.com/en-us/news/technology/google-tensor-g3-supports-av1-encoding-but-you-can-t-use-it-yet/ar-BB1iUn76

@MishaalRahman I'm aware, but I'm really interested in the specific experiential differences between the two. For example, the first gen watch got around 10-14 days of battery life in my experience during the first couple of months of use

@MishaalRahman I have the first-gen OnePlus Watch, and it honestly kind of soured my impression of OnePlus's ability to make watches. Is this one a significantly better experience?

@MishaalRahman I definitely *tried* to try it out, haha. I was doing visual quality testing across all of the hardware encoders present in Tensor G3 when I came upon the Google AV1 encoding block which is distinct from the Exynos blocks. The fact that it doesn't write any container metadata makes it appear impossible to use, though; I haven't found a way around that.

@MishaalRahman It is exposed via FFmpeg, but the resulting bitstream lacks any metadata regardless of the specified container format so the files are unplayable.

https://ericportis.com/posts/2024/okay-color-spaces/

I've had a bit of trouble wrapping my head around color spaces since writing about XYB JPEG and discovering OKLAB and the world of perceptual color - these 3-D models do help a lot, though! Highly recommend reading!

About 6 months ago I got a bee in my bonnet about the quality of current codecs. I ended up doing more than 1000 encodes, comparing various codecs, presets and quality settings. A little over a month ago I wrote something up about it.
https://colinmckellar.com/2024/01/11/video-encoder-comparison/
#AV1 #ffmpeg #x264 #x265

@kremzli @colin

> "I understand VMAF is a popular algorithmic video encoder comparison system, but all algorithmic systems are flawed, and cannot completely replicate the human visual system. I do not have the resources or inclination to do extensive human testing of comparative video quality."

This is greatly appreciated, and something a lot of people miss! Well done, very thorough analysis! One question - why did you calculate the harmonic mean of the VMAF score instead of the arithmetic mean? Just curious, as I was under the impression that arithmetic mean was the way to go. Well done overall though!

You define a whole category of Linux user on your own. If people can start saying “I use Fedora btw”, it will only be awesome to hear because of you.

To: @archlinux
From: Fedora

#ILoveFS #Arch #ArchLinux #FedoraLoveFS

https://github.com/microsoft/sudo

> Sudo for Windows allows users to run elevated commands direcly from unelevated terminal windows.

How innovative! 😂

Hey everyone - wanted to announce a nice Aviator update that happens to coincide with an SVT-AV1-PSY update that we pushed just yesterday.

Aviator 0.5.1 has a couple new features, but chief among them is a new Subjective Tuning toggle for enabling encoder changes that improve visual quality at the expense of metrics. This is powered by SVT-AV1-PSY's new Tune 3 authored by Clybius; it adds subjectively motivated optimization to the existing Tune 2's great performance for an incredible balance of retaining fine detail in videos while still preventing blocking.

Here's a comparison encoded with Aviator (both 10-second videos were 7.5MB, down from the lossless source's 732.8 MB): https://slow.pics/c/tdrO5dY4

(the source is Netflix's `foodmarket` at 1080p)

If you prefer the smoother-looking, more appeal-oriented shot here, that is OK too - Subjective Tuning is toggleable!

Aviator 0.5.1 should be available on Flathub pretty soon. Enjoy!

#av1 #transcoding #Flatpak

@Jain Implementing too many codecs at once can get messy, which is why a new video-based codec every couple of years is an unfortunate sight to see, in my opinion. Especially considering WebP's failure (except the lossless part).

The AVIF spec was actually submitted as a proposal for JXL, but was rejected. That should have been the sign to relinquish AVIF development, if implementing things "together in a random order" is bad for the Web since it was known it would overlap with JXL which was always supposed to be a superset of AVIF or something better.

Regarding interest from professional use cases, I've seen JPEG XL discussed for these use cases in the JXL Discord. The JPEG XL project itself touts features like wide dynamic range support, layers, excellent lossless compression, and an incredible number of possible channels. Aside from HDR, the above are missing from AVIF. So, while adoption is still early, there is excitement about having a modern codec that can handle specialized needs beyond what JPEG and PNG currently offer. More info on JXL: https://wiki.x266.mov/docs/images/JXL

HEIC is not supported pretty much at all on the Web due to licensing restrictions, which make it very difficult to ship HEIC images. I would say AVIF has the most momentum now, even moreso than HEIC, but JPEG-XL and other future formats could gain traction once native browser support spreads.