/dev/stdout
5.31K subscribers
25.7K photos
3.68K videos
106 files
4.03K links
cat /dev/random
Download Telegram
They've gone insane.
🔥83🍓2
>XYZ COM LLC
.xyz is literally a tld for scammers and grifters.
👎4
13
Forwarded from 0xff00cc2f
Сегодня на ужин у нас пожилая сега дримкаст. С ушатанным в щи недоброжелательными вандалами линзой и корпусом лазерной головки. А также паяным кодером, который преобразует замультиплексированные 12битами 24 бит цифровое видео в аналог. И этот кодер мертв, даже смеет садить шину данных почти на землю. Но наши руки не для скуки, так что при помощи первой потужной попытки осилить фпга получилось как-то преобразовать параллельные 12 бит данных в hdmi. Остается только разобраться с выравниванием слов, тк в зависимости от рандома два слова, формирующие пиксель, могут флапнуться и попортить цвета.
Ну и да, тут нужен экран и хорошая земля. 54МГц сигнал так-то бежит по мгтф проводочкам(SI инженеры плачьте)
(бтв видео там 640x480, завернутое в 720x480(в режиме вга прогрессив))
10❤3😨1
Forwarded from HN Best Comments
Re: Photopea creator weighs in on Photosuite project

Hey guys, I am the creator of Photopea, I just woke up! I did not expect that it will blow up like this.

Thank you all for your support! Many repositories like this one exist on GitHub, I tried to report in in the past, but I was completely ignored by Github / Microsoft. I feel pretty powerless here :( I tried to call them out publicly - https://x.com/photopeacom/status/2106877355738140875

This "PhotoSuite" thing got me a bit angry, since the author presented it as an original work, when I clearly see that such a project can be made from my "pp.js" file in two hours using Claude. Also, I believe that their README.md did not mention Photopea before I wrote my first comment there.

Thousands of people shared "PhotoSuite" on social media and nobody minded the fact that they don't know who the creator is. The creator is still offering it on their website as their original work: https://www.photosuite.app/

We live in interesting times. AI can (or will be able to) de-obfuscate every bit of code that was ever published, convert binaries to readable C++ and so on. So, in theory, AI models will train on the source code that we consider "closed-source" now, and will be able to recreate this software if asked. But it wouldn't be possible if a real person did not spend thousands of hours of their life to write that software years sooner.

I have always been excided to see people make AI create something new, what didn't exist before. But so far, I mostly saw only "approximations" of expensive software, when AI tries to recreate each feature and each button according to what it thinks it should do, or "forks", when AI takes the full code of something and tries to "bend" it to hide the similarity with the original work.

IvanK_net, 2 days ago
10❤1
🍓4👍1🥰1
😭33❤1🥰1
Friendly reminder
Forwarded from /dev/stdout
Instead of creating and destroying opengl textures for each window/pixmap each time, create textures and re-use them; never destroying them.
This is more efficient and leads to lower memory usage (nvidia leaks memory on texture create/destroy, on purpose).

https://git.dec05eba.com/phoenix/tree/TODO
🐳4🙏1
/dev/stdout
Photo
Не жрать, это мое
2
/dev/stdout
Photo
Me when I work on an Imperishable Friday Night
🍓1