pouët.net

Mystran information 37 glöps

• general:
• level: user
• personal:
• first name: Pihlaja
• last name: Voipio

• 4k Windows Final Territory by Prismbeings
• The rendering is really nice and the music is really nice... but somewhere around half-way through it did start to feel a bit stretched and since it was fairly obvious that it was going to conclude the marathon of small intros, it was bit of a let down not to have a strong "hook" at the end after the repetitive blackouts, especially as the beginning was really cool... still was kinda "wtf" when it only came second, but then again the fun part of Assembly is that you never know about the crowd.
• rulezadded on the 2018-08-06 02:30:27
• 4k Windows flow
• When I started doing this, I realized from the very beginning that I probably wasn't going to end up with much of a crowd-pleaser this time (especially as I wanted to go for a fairly minimal and laid-back style), but I really wanted to do "one more" to sort of bring my series of particle intros to some sort of a closure that would tied together the essentially chaotic particle movements with some actual controlled animation... so that next time I make an intro I can just do something completely different. :)
• isokadded on the 2018-08-05 23:48:08
• 64k Windows Hardwood by Saksalainen Laatu
• I really liked the part with the sine-cut boards forming a wavy, moving surface that still felt like you could just build it out of physical wood if you wanted to. :)
• rulezadded on the 2018-08-05 07:11:42
• 4k Windows humppatunneli
• The main goal (and a sort of minor political statement) here was to make something that would compress really badly on Youtube and I think I a half-decent job with that. Apparently it killed the twitch stream pretty hard as well, haha. :P
• isokadded on the 2017-08-07 18:32:51
• demo Windows Sokia by Carillon & Cyberiad [web] & Fairlight [web]
• Watching it on bigscreen felt like watching a garbage quality stream. There's probably a top-notch demo buried somewhere and music is nice, but honestly on party I just kept wishing it would end already.
• isokadded on the 2017-08-07 16:02:55
• demo MS-Dos GoBIT by Adapt [web]
• Even though the framerate was very poor in many parts (sorry to hear it was an emulation issue with development), it was fairly obvious that it would have been really impressive with just a slightly faster machine.
• rulezadded on the 2017-08-07 15:46:15
• game Windows MacOSX Intel Crisis Wing - Caravan Edition by Pieslice Productions
• By default it'd shoot and bomb, without moving, I just unplugged my pad to make it work (maybe should have plugged in a stick though)... but oh my it plays well, nice stage design with all the subtle traps and it definitely feels like there's some depth built into the scoring and power up systems too. Only played it enough to barely beat the default score, but will definitely have to play more seriously later. :)
• rulezadded on the 2017-08-06 17:07:37
• 4k Windows mansikkajäätelön iltalento
• Music is custom synth that's not terrible capable, because I was totally out of size by the time I had to add a soundtrack. :)
• isokadded on the 2017-01-08 07:58:47
• 4k Windows mansikkajäätelön iltalento
• I just realised I don't remember seeing the particular GPU sorting algorithm ever mentioned anywhere, so might just as well drop a note about it here (even though it might not be novel), since it's extremely simple (ie. take very little code to implement).
  
  It's essentially a variation of bitonic merge sort, it's incremental and it's based on the idea of simply pretending we have a bitonic sequence (even if we don't, which curiously enough is not very important):
  
  1. take array of 2^k=N elements and pair-wise sort first and last, second and second to last and so on in parallel ie. foreach i..N/2, sort(e[i],e[N-i-1]).
  
  2. split the array in half from the middle and repeat for each half recursively (in parallel, obviously), takes k=O(log N) passes until all the sub-arrays have size 1. The only difference between passes is the sub-division size, essentially just add varying modulo to the indexing calculation.
  
  3. it turns out (informally, but it works) that if the above process is repeated k=O(log N) times, we get a fully sorted result (giving a total complexity of O(log^2 N) parallel passes), however...
  
  ...in practice, the outer loop (step 3) doesn't need to run to completion, it can be done a few times per frame (every round improves the order more or less consistently, although some elements here and there might get a bit worse sometimes), the positions can be updated / rendered, and then keep sorting from whatever the order was previous frame. The intro does 10 or 15 or something like that outer passes per frame, but it's not very critical at all, even just a couple of passes will keep everything sorted "reasonably well."
  
  Obviously this might not be the best sorting algorithm if the actual implementation complexity doesn't matter (IIRC it needs about twice the passes of a proper bitonic sort), but it's ridiculously small and simple and still not terrible inefficient.
  
  So.. yeah.. I'd be surprised if I was the first to think of this algorithm, but at least now it's documented somewhere. :)
• isokadded on the 2017-01-07 04:55:24
• 32k Amiga AGA Amiga OCS/ECS I would rather by Saksalainen Laatu
• Just put 64kB limit on demo compo.. but this demo was great either way. :)
• rulezadded on the 2016-08-08 13:18:07