Just released Fragment v0.2.10 Among small fixes, I updated the canvas recorder to use mediabunny.dev by @vanilagy.bsky.social so you can now export mp4, mkv or mov files with the codec of your choice! github.com/raphaelameau...

What are good physical implementations of FIFO queues? I'm annoyed with my laundry basket because the clothes I threw in the longest time ago get washed last, so it's LIFO. I guess a "FIFO basket" would need to have two openings for it to work, which sounds awkward.

Mediabunny 1.25.0 ships with read/write support for track disposition information! 🚀 This information allows you to inspect which tracks should be preferred by a player, which are meant for hearing- or visually-impaired people, and more. Niche, but important for some uses.

Just filed 3 Chromium bug reports in 30 minutes. Practically half a Chromium dev at this point

Probably the funniest error message I've ever seen the browser throw at me. I apologize for not using 𗜀, 🐀, 𧄀, or 𮸀 as my bitrate

Mediabunny's number of Safari-specific workarounds is growing. Super dirty, but someone's gotta do it!

Had an absolutely terrific two days at Demuxed in London! Met tons of great people, and I have proudly lost my "conference speaker virginity" now 👀 Excited to share this talk once the VOD drops! Huge thanks to the Demuxed organizers for letting me have a 20-min slot. 💘

Had a blast at Demuxed day 1 yesterday! This is me (second on the left) with the Remotion folks and Konsti from Diffusion Studio!!

At the airport heading to London for two days of Demuxed! Really excited to give my talk about Mediabunny and meet a bunch of cool devs who I will to convince to use it 😊

Theo using FFmpeg to build SoraMarker is akin to using create-react-app for a new project in 2025. On the web, it's slow, unnecessarily bulky, and makes very little use of the platform. I rebuilt it using Mediabunny, and it runs 5x as fast while using 0.8% the bundle size⚡️

Mediabunny v1.24.0 ships with a highly-requested feature: custom video & audio transformations in the Conversion API! 🚀 This means you can now use Mediabunny to add watermarks and overlays, perform color correction, or apply any other video transformation; right in the browser.

Discord should leave tombstones for deleted messages. All other messengers do this already, and I think it's needed when you have deletion, so I'm not sure why Discord hasn't added it.

Open-source project management is quite challenging. I have to balance the roadmap, bug reports, feature requests, etc., and even on my days off, I find myself being dragged into the project because I feel responsible. So for me, this takes conscious effort and improvement.

It's wild how much more clarity you have about your problem when you rewrite your code multiple times. You start seeing patterns that in hindsight seem to obvious, you wonder why you didn't solve it like that the first time around.

Big shoutout to @rikschennink.com, the creator of FilePond and Pintura, for becoming Mediabunny's first silver sponsor! Dankjewel! 🥈💘

Mediabunny v1.21.0 ships with transparency support in the Conversion API, hugely improved error messages if conversions aren't possible (due to missing codec support), and a couple of fixes as usual! 🚀

Mediabunny v1.20.0 is another big milestone for the project: it can now read & write transparent videos! 🚀 All directly in the browser, super fast, no server needed. This was more challenging that it seems on the surface (took over 1000 LOC). A blog post about this will follow!

Am I wrong for spelling it Mediabunny? Lots of people misspell it as MediaBunny. Mediabunny felt cleaner to me. Might be my inner German speaking! Won't change the name, just wondering.

Mediabunny v1.19.0 comes with ID3 metadata support in .wav files, better support for alternate environments like React Native, and fancy new Symbol.dispose functionality for Inputs! 🚀 Always cool to work with cutting-edge JS features.

When's the last time your IDE made a sound?

When a new JavaScript feature comes out, I always think, "Who the hell needs that?" Async generators? WeakRef?? FinalizationRegistry??? Symbol.dispose???? And then I always end up finding a use for them a few years later. What's your favorite "niche" JavaScript feature?

Super excited to announce that I'll be speaking at Demuxed in London this October! 🚀 I'll share Mediabunny's origin story and some of the decisions that shaped its design. My first real conference talk, so I'm excited. Big thanks for the Demuxed team for giving me a slot! 💘

You guys are insane. 2.5 months after its release, Mediabunny has already hit more than 100k monthly downloads. Thank you! 🐰📈

Another big one: Mediabunny v1.17.0 now ships with full read & write support for .flac files! 🚀 Huge thanks to Jonny Burger for taking care of this one! ▶️

Neat small blog post I found today, for anybody trying to build code to run both in the browser and in Node. Learned about the "browser" package.json field! It allows you to remap imports if the code is bundled for a browser environment. zirkelc.dev/posts/bundli...

I feel like most developers nowadays don't even notice complexity. I was sent a repro for a Mediabunny bug in the form of a Next.js app. What could've been a 50-line HTML file is instead a 0.5 GB (!!) behemoth. It's wild to me that this doesn't make people shudder.

You know it's over when you start dreaming about audio codecs.

The world would be better with a little less robustness principle

Mediabunny v1.15.0 ships with cropping functionality in the Conversion API! 🚀 You can now convert videos and only keep a specific rectangular region of the original input video. It also works nicely in conjunction with the rotation and resizing options.

The world is full of signal and full of noise. Which do you choose to contribute?

An interesting phenomenon is how people (even those with no media background) are quick to write off Mediabunny as an FFmpeg wrapper - if not directly, then surely indirectly through browser APIs. I attribute this to FFmpeg's strong image. Happy to expand yalls world view, tho🫰

If you want to compress videos before uploading them, make sure to use Mediabunny's stream-upload feature, where the video is compressed and uploaded at the same time. If the network is slow, the compression will automatically adjust its pace as well. Super cool!

If writing software is building a tree of features, there are two approaches: - Depth-first: Build out every feature to 100% completion before moving on - Breadth-first: Work on all features at the same time, adding detail and polish later on Which approach do you practice more?

You can play with Mediabunny's new metadata tag extraction feature in the metadata extraction example on the website. Here's it extracting title, artist, cover art and more from a song!

My lovely girlfriend takes me out for all-you-can-eat sushi for every GitHub star milestone of Mediabunny. This one was for 1k stars. She said the next one happens at 10k. Wonder when that will be 🤔

Mediabunny v1.14.0 is out and it's a huge milestone for the project! 🚀 Mediabunny can now read and write metadata tags (like title, artist, album, cover art) for all formats, including ID3v1/v2, iTunes, QuickTime, Matroska, Vorbis, and RIFF-style metadata.

Spotify's "multiple devices, one state" approach is so utterly genius. Instead of a "you can't use Spotify on multiple devices" error, they turned it into a feature. Now I can just control the playback on my PC from my phone. Love it

Mediabunny's website now has API docs! 🚀 Complementing the guide, these additional 148 pages document every class, method, property and type in Mediabunny's public API. They're pretty sexy, so, go check them out! mediabunny.dev/api/

Ah yes

Mediabunny v1.13.0 adds ReadableStreamSource, offering a way to pipe append-only byte streams into Mediabunny. This enables many new use cases, such as real-time processing of MediaRecorder output while the recording is taking place. 🚀 See the docs: shorturl.at/Yo2KF

Who even needs <video>? I certainly don't 👀

The upcoming version of Mediabunny will ship with a built-in way to read directly from the file system, for server-side use. Works in Node, Bun and Deno! (interestingly, Deno comes out on top performance-wise)

Mediabunny has now already overtaken mp4-muxer, my previously most-starred repo which I released only 2 years ago. Thank so you much for all the support and kind words, it has really meant a lot. Super excited for where this goes over the next year! 🚀

Mediabunny can generate videos at ludicrous speeds. This 720p video is rendered & encoded at around 500 FPS, fully client-side:

Do you guys prefer API docs that live in a single, giant document (optimized for Ctrl+F), or do you want them to be organized into multiple pages?

Mediabunny v1.9.0 adds support for reading & writing ADTS (.aac) files! 🚀 It also fixes some nasty race conditions in the MP3 demuxer.

Used @vanilagy.bsky.social 's Mediabunny in supermosh.github.io for rendering. Perf gain for most devices (6s -> 0.3s for this video), but most importantly since I don't rely on the live <canvas> recording anymore, the output is deterministic & the same for all devices. Many ppl asked for this :)

I've heard Big Buck Bunny in the most cursed ways due to audio rendering bugs. This is the first bug that acutally sounds quite cool?? Besides the bird getting smacked at the end there

Mediabunny v1.8.0 adds better chunked encoding for long AudioBuffers, and now supports way more sample rates for @mediabunny/mp3-encoder! (8000, 11025, 12000, 16000, 22050, 24000 are the new ones)

I'd love an LLM-backed "no-outdated-comment" ESLint rule that revalidates comment validity for comments around the latest diff. Would be crazy. I'm like 99% good at catching outdated comments but somehow, one always slips through.