7 ms·
The Automated Fake Album Project
- mahler 15y agoThis should really be choosing between two albums generated from the same (artist name, title, cover art), rather than completely different configurations. I'm finding it difficult to judge fairly when the art, title and album name are different.
- mikeinterviewst 15y agoThat was how we originally built it, but we found it less fun to rate in testing compared to this version. Hopefully the increased votes-per-person outweigh the loss of fidelity.
- BlackJack 15y agoI think you guys should've just shown one picture at a time and allowed people to upvote or downvote it.
- tylermenezes 15y agoThose sort of contests usually aren't as statistically significant.
- scottkrager 15y agoNice URL naming..../facemash
- dokidoki 15y agoI made this generator a couple of years ago... http://www.makemyalbumcover.com/ http://www.makemyalbumcover.com/
- twelvechairs 15y agoThe biggest problem is font colours that are unreadable on the background. Obviously that must be an algorithm-too-far for most of the competitors.
- sudowork 15y agoUnfortunately, for our team, we had to revert to a version of our generator that was about 5 hours old due to a issue with one of the library dependencies. We didn't catch this until last minute, so basically all of our low-level pixel analysis got thrown out the window.
- millzlane 15y agoReminds me of the Album cover generator. http://albums.demozoo.org/ http://albums.demozoo.org/
- haclifford 15y agoThere's bias introduced based on the visual appearance, some of the fonts used are terrible - and (as mentioned) the contrast.
- Groxx 15y agoCan't... choose...! http://cl.ly/3L0d0F1A0r0E3z231R0X http://cl.ly/3L0d0F1A0r0E3z231R0X Fun contest - I'd love to see the winners when this is all over :)
- unhappyhippie 15y agoI saw this "generate album art 'algorithm' " on a 4chan thread three years ago when I used to frequent there. Brings back memories.
- ashovlin 15y agoIt appears possible to compare two album covers that were created by the same generator (some of them have pretty distinct styles). Not sure if this was considered, or how it factors into scoring.
- tylermenezes 15y agoI was thinking the same thing - I got a lot of them in a row. If they're using standard techniques for ranking win/loss competitions, they're artificially decreasing the sigma (not entirely correct but, think of it as a margin of error) for no reason. Funny, a website based on giving out coding problems didn't realize that coding bug. Then again, I've experienced tons of bugs when I was playing with their main software, so I'm not surprised.
- azylman 15y agoAs someone who competed in this contest, that wouldn't be the only bug. There was a lot of bugs in the submission/grading process - so much so that my team got frustrated enough to leave early.
- rvivek 15y agoWe had glitches agreed because the kind of problems we introduced were the first for us - Game simulation. We have been doing ACM style programming contests which works brilliantly, this was an experiment. We know our mistakes and this will rock next time. Thanks for participating.
- 140dbs 15y agoThis is my favorite so far... http://imgur.com/XY6Uj http://imgur.com/XY6Uj
- meatsock 15y agoas someone who finds it easier to produce sounds than to give them useful and meaningful names after the fact, i love this. if i found a version that can deterministically generate a full ID3 tag of generated info given any mp3 input, i'd use it daily to fight the scourge of files named lkajf_39_q.mp3 in my output bin.