cho.sh
2,244 notes
Liked on XI made a mistake. When Jack and I decided to organize this $100K Hackathon, my biggest concern was cheating. Unlike a regular hackathon, everyone participates from home. And because the winner is whoever makes the most PROFIT (not revenue), there are so many ways to manipulate the numbers. So I added a $99 entry fee. One thing I’ve learned from building businesses is that people tend to be more committed when they pay for something. I thought it would filter out spam, attract serious builders, and make the competition fairer for everyone. But I completely missed the other side. 3 years ago, I was a living with my apps. Today, I’m financially free. My life changed so much that somewhere along the way, I lost perspective on what $99 means to someone else. For someone in India, someone struggling to pay their bills, or someone trying to make their first dollar online, $99 can be a lot of money. And those are exactly the people I want to help! I’ve spent years sharing what I’ve learned, reviewing random people’s landing pages, and organizing free meetups around the world. Not because there’s money to make, but because I genuinely love seeing people build things and succeed. And somehow, I was so focused on keeping cheaters out that I didn’t realize I was also keeping genuine builders out. That’s my mistake. So I’ve decided to make the $100K Hackathon completely FREE :) Everyone who already paid got a full refund. I’m still worried about cheating. That problem hasn’t disappeared, and Jack and I will have to figure out better ways to verify profits and keep the leaderboard fair. But that’s our problem to solve. Thank you to everyone who shared their thoughts. Sometimes you need people to remind you of things you’ve forgotten. I genuinely hope this hackathon changes someone’s life. ❤️Liked on XIt’s time to bring back Cafe CursorPosted on Xyo @pendev i can't live share edit a .pen file? it's the most needed featurePosted on XI'm coining a word "Agentsplaining" Agents condescendingly assuming the user don't know fuck and repeating the obviousReplied on X@jayhxmo @raycast Hard to explain. It's the small detailsPosted on X@raycast iOS keyboard is insanely good. Might resub Raycast pro just cuz of thisReplied on X@ishuagra02 I saw a theory that F5.2 is a post train of 5.1, but O5.5 is a distillation of F5.5(unreleased new pretrain) and thats why O5.5 > F5.2Liked on X@anaclumos don't know what is actually improved 👀Posted on Xam i the only one underimpressed with 5.2 fableReplied on X@bvvst great minds think alike https://t.co/YRFWhb1mcOLiked on XClaude Fable 5.2 is out and it's the perfect time to run this god-level prompt on your codebase: "I want you to audit our codebase, is anything here unnecessary, overly complicated in our current setup, or based on weak assumptions? Challenge them. - What can be deleted entirely? - What can be simplified now that unnecessary pieces are gone? Then make the changes. Prefer deleting over simplifying, simplifying over optimizing, and optimizing over automating. This product is being used by thousands of customers, do not cut any corners." This prompt is one I run often and leads to the best results.Replied on X@akshayadinesh19 @bot can we please BYOM for botsLiked on XTo be fair, the Chinese models have a much lower refusal rateLiked on XF*CK IT, im creating a library of AI slop indicators. my feed is full of taste larpers. stay tuned https://t.co/5Muv8nQRm3Liked on XJENSEN! THATS NOT WHAT WE MEANT BY CHINESE MODELS!!!!!!!Liked on Xmy parents had to declare bankruptcy when I was a kid, so they had every reason to pressure me towards a more stable career path instead, they were so patient with me, even as i failed startup after startup for years so getting to tell my mom that everything turned out okay was a really special moment :) for those who can't read Korean, her first message roughly translates to: "Really!!! Congratulations my son. So much :') Thank you God!" i know I've been posting about my exit for like a week now, but it really is a big deal for me and my family! i'll share more about how this all happened and some tactical advice later this week, but for now i'm just riding out all the emotions to quote my mom, thank you God!!!!Liked on X@thefarmers_Dog it stands for Egg GzampleLiked on X🦍 Introducing Memorilla: memory as a modality for LLMs Memorilla turns everything a frozen LLM needs to remember into just 16 memory embeddings ⚡ It uses up to 4,212× fewer tokens and is 27× faster than full context inference while more accurate 🎯 Pushes the Pareto frontier over RAG and agentic memory in personalization, long-document and knowledge QA A thread 🧵:Liked on XThere's a huge product opportunity downstream of Tailscale etc too. At @tldraw we've started designing for "high trust networks" and it's been a beautiful thing. 1. Features that would be a nightmare on the internet are okay on my computer and mostly okay on private networks. Websites that execute user-generated scripts, control production systems, allow for users to send desktop notifications, spend tokens, or directly prompt my agent. Crazy stuff that would only make sense if everyone was in a committed polycule. 2. When everyone on the server can be trusted to have _external_ reasons for good behavior (eg keeping their job, or maintaining friends, family, church, mafia org), the software can be designed in a default social way. Most apps default to isolation, low horizontal visibility, highly restricted collaboration. Makes sense if the social contracts are inside of the app, but if you can trust that they live outside of the app then you can get wild. Example being small group chat names as an emergent comedy phenomenon because anyone in the group can change the name. 3. This can work for commercial products, too, not just internal stuff. Security matters more than ever but the way we've seen that manifest is that more companies are more confident in their own solutions than anything we could provide, to the point where owning data and access is a better value proposition than having it managed. Always has been true for big companies but now even small companies can get a good secure network set up pretty easily, and the hooks we need to build are easy enough. tldraw offline got us started down this road (and shout out to @pendev for first introducing me to the idea of desktop + mcp). We've been bringing that product to private networks and it's been exciting to see how much we can get away with. Keep an eye outReplied on X@saradu birds eye viewLiked on XApplications close in 2 days!! If you have a cool bio idea, and $5-10k will go a long way for you, please apply!!! https://t.co/NCvmrNC53cLiked on XWe've been early adopters of Rust since 2018 and a lot of our infra, voice tech, and even some of the current client is already Rust, the holy grail was getting the whole client there. That means Electron goes away, and so does React Native, mobile keeps native UI but shares app code with desktop via Rust. We hope to get it into testing with users soon™. (And it's not Tauri, it's GPUI, so good guess everyone.) Alright, back to staring at the WoW Forever login screen.Reposted on XOh boy. https://t.co/X7SSskGuGvLiked on XRIP .si domain squatters
How Video Compression Works

How Video Compression Works

Warning

Over a year old. May be outdated

Note

I wrote this post in Korean first, as .

Example of Video Ghosting

How video compression works, and why that glitch happens.

Videos are too big

A video is a bunch of photos. Store it as full photos back to back and it would get ridiculously big. Take the 1920 x 1080 60FPS videos we watch on YouTube all the time, uncompressed close to 7GB a minute. On YouTube it's at most about 40MB a minute. Almost 200 times smaller and we barely notice. What happened?

So we encode

Raw video is too big, so almost every video gets compressed to some degree. That's video encoding, and the algorithms are amazingly sophisticated and beautiful.

Encoding saves space by finding redundancy. Picture a singer standing still, singing. Only the mouth moves. Background and body don't. Why resend the black background pixels and the singer's body every frame? You don't. They repeat.

Video repeats in space and in time. Cutting repetition in space is intra-frame coding (inside one frame). Cutting it in time is inter-frame coding (across frames). Under the hood you get the Discrete Cosine Transform for shrinking neighboring pixel data, motion-vector prediction, in-loop filtering, and more.

Intra-frame coding

Make each photo smaller!

A photo is a bunch of pixels. If we can cut the information about repeating pixels inside one image, we cut spatial redundancy. One of the simplest ways is averaging. Leave a pixel empty, keep the ones around it, and at playback fill the gap with the average of its adjacent pixels.

Fun part, adjacent doesn't mean up, down, left, right. Pixels are stored left to right, top to bottom. Averaging all four sides means waiting for the right and bottom pixels, then coming back. Too slow for playback. So intra-frame coding keeps the upper-left, upper, upper-right, and left pixels and averages from those when it hits a gap.

Inter-frame coding

Don't resend what you already sent. Recycle it!

Remember school award ceremonies before the holidays? Same award, 30 people. If the principal read every certificate out loud, how long and boring and painful would that be? They don't. They say "same as above" and move on. Saying "same as the last person" is all it takes, and we get our holiday afternoon back. The principal just did inter-frame compression.

Video is the same. Many videos have similar frames back to back, so a frame can be described by the frames before and after it, or skipped altogether. That cuts temporal redundancy.

Who's better? Principal announcing for 2 hours or 2 minutes?

#1. I-Frame, the reference

An I-Frame (Intra-coded picture) is basically a photo. Everything in an I-Frame is new information, and it's the reference for describing the frames around it.

#2. P-Frame, only the change

A P-Frame (Predicted Picture) sits between I-Frames and holds only the change from the previous frame. Anything shared gets pulled from the frame before. The picture shows it better.

Copyright: Blender Foundation 2006. Netherlands Media Art Institute. www.elephantsdream.org

The arrows are motion vectors, which show the change. A P-Frame also carries transform values to correct the prediction, and sometimes brand-new image data. It's about half an I-Frame's size. In real encoding, of course, the encoder doesn't compare every pixel. It splits the frame into blocks called macroblocks, or coding tree units in HEVC, the latest video codec, and compares those.

#3. B-Frame, saving data

A B-Frame (Bidirectionally Predicted Picture) sits between I-Frames and P-Frames and builds its picture from the I-Frames or P-Frames both before and after it. Sounds like a P-Frame, so why bother? Size. Because a B-Frame uses data from both sides, it can leave out that much more. A B-Frame is only about 25% the size of a P-Frame.

Like a P-Frame, a B-Frame uses motion vectors and transform values for prediction correction. B-Frames refer to I-Frames and P-Frames, but in recent codecs like HEVC and VVC, a B-Frame can also refer to other B-Frames.

Copyright: Cmglee, CC BY-SA 4.0.

Why ghosting happens

Trouble starts when the packet with an I-Frame gets lost. The reference the surrounding P-Frames and B-Frames depend on is gone. A good streaming app uses all sorts of algorithms to detect packet loss early and ask for the packets again. But if the server is unstable or the app is poorly made, it won't notice the lost I-Frame.

Then, with the I-Frame gone and only the following P-Frames and B-Frames arriving, their changes get applied to the wrong I-Frame.

A picture is worth a thousand words

Still not clicking? Easier to see it. With a video library like FFmpeg you can deliberately corrupt a video file's frame data. This kind of art is called datamoshing.

I used Python and FFmpeg to damage the I-Frames in a music video and force the ghosting.

  • Every I-Frame got overwritten with the frame before it, probably a P-Frame or B-Frame. So no new information ever comes from an I-Frame. The scene stays put but the people still move, because the changes (P-Frames and B-Frames) were applied to the wrong reference (I-Frame).
  • Sometimes part of the screen looks clean for a moment. A P-Frame can carry new data too. But since the I-Frames, which are all new data, were deleted, the whole screen never comes back clean, even when parts briefly do.
  • It breaks into big, obvious squares, not fine sand. Compression runs per macroblock (coding tree unit), a bundle of pixels, not per pixel. When this happens during a broadcast, people usually call it a "pixelated video".

Watch it knowing there are no I-Frames and the relationship between I-Frames, P-Frames, and B-Frames gets much clearer. I'll cover how to break a video with FFmpeg some other time.

  • If you find an error in this post, please send it to mail@chosunghyun.com.
  • An earlier version said "However, to prevent the error from getting bigger, B-Frames do not refer to other B-Frames. They only refer to I-Frames or P-Frames." That was wrong. In codecs like HEVC and VVC, B-Frames can reference other B-Frames. Thank you so much for reporting it. Credit: (anonymous)
Backlinks1