Lens sort, a masked pixel sort glitch effect | Hacker News Reader

Lens sort, a masked pixel sort glitch effect

github.com

123 pointspyinstallwoes20 comments

I didn't expect an old project of mine I made in college would show up here :). I remember just using what I knew at the time to achieve some effects that I wanted. I'm sure there are better methods to go about the same thing now

There actually isn't, which is why I posted/had to find it by necessity. Thanks, it still works.

It still has the coolest way to go about it. The face detection algorithm can be updated I bet with the open source SOTA.

What else have you worked on since?

This is 5 years old and the Dockerfile no longer works.

If this was nix it would still work.

Have you tried using a 5 year old docker?

I'll get my 5 year old to look into it.

It looks fairly basic, creating a python venv and installing the requirements.txt should be all you need

Yeah, worked for me.

It's fun!

It would be much better if there was a requirements lockfile, of course.

I think the reason this was posted is not because of miscellaneous infrastructure, but that it’s an implementation of an interesting algorithm. If you care enough to leave a comment like this, update it yourself.

> If you care enough to leave a comment like this, update it yourself.

This attitude is exactly why I rarely contribute to open source. You're not entitled to my work, just like I'm not entitled to yours. I'll do something myself if I want to.

It's unfortunate that some loud and rude people have put you off of contributing to free software. On behalf of the open-source community, I am sorry.

If it's any consolation, the prevailing attitude in open source is that nobody is entitled to any work from others. We are not very good at calling people out on this bad attitude, so an occasional vocal exemplar can give the whole movement an unwelcoming outside impression. We should work on that.

> It's unfortunate that some loud and rude people have put you off of contributing to free software.

I've been told this directly by project maintainers when filing bug reports (usually along the lines of "PRs welcome"), so it's not just that seeing this stuff on the internet puts me off from contributing in general. It's that people tend to respond directly with it whenever I come near their project.

I do still contribute to free software from time to time.

I say "PRs welcome" sometime. It's not a demand that you submit a PR; it's an invitation and an acknowledgement that I'm unable to put the time into fixing all reported issues promptly on my own. I try to make this clear but may not do so reliably.

All the people claiming Docker would be better at preserving backward compatibility than Java look pretty dumb now huh.

I love where you're going with this, followed on github :) Looking forward to seeing progress here. Edit: Oh it's 5 years old, welp

This is super cool. I might end up rewriting in JS or a language with a WASM compilation toolchain.

The repos are 876 and 513 lines of code (lenssort + pixelsort).

I was going to share that information to encourage you.

However, those ~1400 lines depend on a facial recognition library (face-recognition, ~4500 lines), which depends on an ancient machine learning library (dlib, 481,573 lines), which depends on scipy (809,127 lines of code) and numpy (619,808 lines of code).

So... I suggest just swapping out the facial recognition part. Maybe face-api.js (24,016 lines of code), which depends on tfjs (321,310 lines, but hey, at least it runs in the browser! Also I think the tfjs-core part is smaller... at least, it used to be its own repo which was 4x smaller...)

(On another note, now I know why my Python imports take so long...)

Source: this utility to count the LoC on GitHub repos: https://codetabs.com/count-loc/count-loc-online.html

Might not be as good as dlib, but it seems like there are WASM implementations for facial recognition.

https://github.com/noahlevenson/wasmface

Y'all should check out https://old.reddit.com/r/pixelsorting/top/

Reply on news.ycombinator.com