The Bridge

« first day (4534 days earlier) ← previous day next day → last day (801 days later) »

Wipqozn

01:22

@Wipqozn Went ahead and just added in some support for Dijkstra too. Not feature complete yet, since I need to update nodes to have edges/distances... but will worry about that once I have an actual daily problem needing it.

Just nice to start building out this toolkit that I can keep using for future advent of code

parz

Hello.

2 hours later…

Unionhawk

02:56

Ah yes

Good work runelite, Ahrim did drop a broad bolt

when I shot him with it and he fell over

SaintWacko

03:41

@Unionhawk Haha, I love when it does that

Unionhawk

03:52

Can I please have a second drop

Or at least add like "another barrows chest, for nothing" to the C Engineer plugin

6 hours later…

Wipqozn

10:05

Getting part 1 in one attempt with no mistakes makes me worried...

ACtually part 2 doens't seem too bad... but first I gotta get ready for work

Ronan

10:41

Yeah it's a bit tricky just to make sure all the logic is there, but not that difficult, parsing the input in javascript using eval feels like cheating though

I don't particularly want to write my own parser though

Wipqozn

and part 2 done. That was actualy easy. PYthon just made these ones easy. I could just run each line through eval() and it created the list automatically. For compare function, I just needed to convert my comparison function from part 1 to align with customer key sorts in python.

@Ronan yeah fuck that, ha. Tedious as all hell.

WAy I look at it is that some languages will have easier times with certain solutions, just by virtue of their toolkits.

There's definitely been a few problems where I wish I had access to LINQ

3 hours later…

MBraedley

13:29

@Wipqozn I took a look at it this morning before I had to head in, and the parsing/data structure in C++ does not look fun.

Elva

14:09

Instead of advent of code, I am programming a christmas tree :D

Like this: youtube.com/watch?v=TvlpIojusBE

I wired my tree with 500 LED lights and calculated their 3D coordinates.

►

Wipqozn

14:54

@MBraedley Honestly I thought about you as soon as I saw it. Especially since you're limiting yourself to stdlib only, right?

@Elva That's nifty

Elva

Hmmhm! It's a shame he didn't go into the actual bit where he calculates the 3d coordinates XD

That's the bit I'm gonna do next and... no clue yet how

Well some clues...

cdn.discordapp.com/attachments/471201837064781824/…

At least the default built in effects of WLED are very pretty :D

Ronan

@Elva I thought he did, maybe in a follow up?

Or maybe not as much detail as you need

Elva

Thaaaat's a post processing step where he takes out points that weren't calculated correctly

So he's got:

How to get 2d coordinates
????
How to fix up 3d coordinates

Ronan

Oh, I guess you'll need an inverse projection matrix, or you just assume that your camera is isometric and hope it doesn't have any issues

Elva

Yeahh, I think I'm going to assume it's relatively isometric, find a point that's been read without much noise in all the scans and do ??? to reverse engineer something plausible

I tried looking up least squares but that sure is complicated :D

Ronan

15:05

It doesn't sound like it should be tricky, but I'm not the one doing it

Elva

Mmmmm

So I've got 4 noisy scans of 2d coordinates and want to turn those into 3d coordinates

Ronan

The scans are from the 4 directions? And the lights are indexed in the scans? Or just here are all the lights in a random order?

Elva

The lights are indexed

So I know light 400 is the same light in all cases

MBraedley

@Wipqozn not necessarily, vcpkg packages would probably be fine too

Elva

I might do some gradient decent thing...

Per scan I've got like, angle, and a vector to the tree so that's 16 variables... And in addition to that I've got two varaibles that are the same all the time that I'd throw rays too if I don't go isometric

Got a decent error measure too so

Mm vector might just be X/Y not X/Y/Z actually so 9 variables

MBraedley

15:11

but there are header-only json parsers too. Any header only lib is perfectly fine. Might do that.

only issue is it's not valid json right from the get-go

murgatroid99

I wrote my own parser and it wasn't that bad, after I fixed an off-by-one error: github.com/murgatroid99/adventofcode/blob/master/2022/…

MBraedley

@Elva It might be a video on the second channel

There's at least 3 videos for that project

murgatroid99

For I think the first time I managed to anticipate part 2: I wrote my comparison function for part 1 with with the API that goes into Array.sort, so it just worked for part 2.

Elva

Mmhm,

Intro where he explains it and shows how to do 2d
Video of him running untested random code
Video where he explans how to automatically fix out of bounds points

Video 2: youtube.com/watch?v=v7eHTNm1YtU&t=2s

Video 3: youtube.com/watch?v=WuMRJf6B5Q4

Also I don't know why it took 2 hours for him to scan XD I have more LEDs and it took 40 minutes

Including the tree rotating! It was half a second per led

Ronan

15:27

Yeah I was remembering the bit a few minutes in to video 3, with the tape measure

Elva

Mmhm! Definitly going to use that as an error checking thing

I could use the logic of:

* assume perfect 90 degree angles between scan
* Get all the "good" matches for that led
* Calculate it as if the camera doesn't have perspective

And that'd probably be good enough. I will probably try that for attempt 1

(I'm gonna probably put all of this on github, including the scans output)

Ronan

You are making me want to try it

Nzall

Okay, so High on Life is literally unplayable for me

Elva

@Ronan I'll share you the data if you want :P

Nzall

It doesn't allow me to change my movement keys from WASD to ZQSD, which is what I use with my AZERTY keyboard

Ronan

15:34

@Elva The raw video/images?

Elva

I don't have raw video, but I have coordinates

I used opencv to do thresholding of bright spots, and then uuh find connected components and calculate minimum enclosing circles, storing their coordinates and radius's per scan

Sepia Lazers

16:26

Q: Screenshot of the Week #80 - Happy Holidays 2022!

Hello and welcome to the 80th Screenshot of the Week! To start with, congratulations to the winner of the previous contest! DLosc's picture from dorfromantik won with 16 upvotes! To submit a screenshot, simply post it as an answer to this question, but please take note of the following: One sc...

3 hours later…

Wipqozn

19:27

@murgatroid99 I think I did my own parser for a similar problem last year... The one where things explode into each other or something? Part of why I didn't want to bother

« first day (4534 days earlier) ← previous day next day → last day (801 days later) »

all rooms