pouët.net

Blueberry information 1403 glöps

• general:
• level: user
• personal:
• first name: Aske Simon
• last name: Christensen
• cdcs:
• cdc #1: Nexus-7 by Andromeda [web]
• cdc #2: Tint by The Black Lotus [web]
• cdc #3: Gift by Potion
• cdc #4: STS-02: Electric Kool-Aid by Synesthetics
• cdc #5: TBC Realtime Experience by Too Bloody Cheesy [web]

• 64k cracktro Amiga OCS/ECS Twelve Jokers In A Deck by Loonies [web] & The Twitch Elite
• That's kind of what this does, just simpler. It sets 33 $182 colors back to back, and the other 31 colors with the CPU.
  
  The remaining time is spent fetching texels into registers for the color writes, plus some moves directly from texture to color, plus some of the copper colors.
  
  
  Quote:

  2. On that, draw 5 bitplanes of 0000111100001111 etc and make that show columns in 31 colors with 4 pixels and 4 pixel background.

  
  5 bitplanes won't work, since the bitplanes will steal some cycles from the copper. The 7 bitplane trick can be used, since it has only 4 bitplanes of DMA load.
  
  
  Quote:

  The copper may only allow 55 colors to be set in a row..

  
  56.5 in fact. ;)
  
  
  Quote:

  But then lat the copper set the 32 $0180 and then the next 16 colors $0182, $0184, $0186 etc.
  Thats 48 colors.
  
  The last 16 colors can be set by CPU simultaneously
  

  
  1 + 16 + 16 = 33. Oops. :)
• isokadded on the 2025-08-01 13:12:44
• demo Commodore PET Ephemeral Cathodic Relay by Genesis Project [web] & Distant Future [web]
• tl;dr. But cool image display technique.
• rulezadded on the 2025-07-31 15:15:41
• intro Amiga OCS/ECS Still Snacking by Batman Group & Capsule
• Quote:

  an old Amiga hardware trick that even Photon knows about

  
  Yeah, I decided against having a bitplane-free section and using color 0 as the on-the-fly color, since I wanted the border stripes to match the center column and to line up properly with the chunks on both sides. This resulted in a very clean solution which was good enough to reach the 57 rows I wanted.
  
  Of course, if I had noticed the size of the first hidden part, I would have gone for 58, which would have been doable with a bit more optimization and by cutting some of h0ffman's generous 7 scanline budget for music and scroller. ;)
  
  On that note, I realize I didn't quite express myself clearly in the scroller (and readme). When I wrote "with music", what I had in mind was "being able to tolerate unpredictable DMA activity on all four audio channels", since I foresaw this being an issue when pressing the cycle exact approach further.
  
  You still hold that record, since I only matched it. But you did not break it here.
  
  When writing the comment above about Hannibal still being in the lead, I then somehow managed to confuse myself to think that this also invalidated your width record. Of course it doesn't. Congratulations on your 85 columns! :-D
  
  And don't get me wrong: I do agree that making sacrifices (such as realtime capability and audio channels in this case) is an essential and even commendable part of record breaking. For one thing, it inspired me to contemplate better uses of disabled audio channels than silence.
  
  In conclusion, I feel the urge to repeat what I said to Photon nine years ago:
  
  Quote:

  Nice proof of concept. Makes me want to beat it some day. ;)

  
  The game is definitely not over.
• isokadded on the 2025-07-29 11:04:36
• demo Amiga AGA Stonenautica by Stonehead [web]
• Delicious!
• rulezadded on the 2025-07-27 10:01:42
• intro Amiga OCS/ECS Still Snacking by Batman Group & Capsule
• Of course. ;)
  
  I mainly care about breaking assumptions in ways that inspire other people to explore new areas, as has happened successfully twice now.
  
  I don't hope this entry into the fight inspires people to make demos without sound. But there are plenty of opportunities for making sound without DMA. That would be a fresh take. :)
• isokadded on the 2025-07-26 16:28:49
• intro Amiga OCS/ECS Still Snacking by Batman Group & Capsule
• Quote:

  Me neither, but even though it may not seem like it, it only has two channels and there are no pattern-effects. I don't know why Rhino asked me to make the music like that. Ask him.

  
  Because the audio DMA interferes with cycle-exact CPU code if you don't place your instructions properly.
  
  Instead of meeting that challenge and beating previous records, Rhino decided to tackle a different, easier problem.
  
  Hannibal still in the lead. :)
• isokadded on the 2025-07-26 14:32:53
• demotool Amiga AGA Amiga OCS/ECS Cranker by The Electronic Knights
• Ah yes, if fast crunching on Amiga is also a goal, then ZX0 is not suitable, at least with the currently available implementations. I do think it would be possible to make a fast cruncher for ZX0 that would still compress significantly better than LZO, but that is obviously a lot more work.
• isokadded on the 2025-07-25 11:36:39
• demotool Amiga AGA Amiga OCS/ECS Cranker by The Electronic Knights
• Have you considered using the ZX0 compression algorithm?
  
  It compresses the raw data of Retro Dentro down to 193301 bytes, which is more than 10% smaller than Cranker (even taking the 1100 bytes decrunch header into account), and it has a small (80 bytes) decruncher which can decompress that data in just 5 seconds on 68000, which should be plenty fast to keep up with the floppy loading.
  
  There's some work involved in adding chunking support to the decruncher, of course, but given its simplicity, I'd guess that's manageable. :)
• isokadded on the 2025-07-25 10:28:51
• intro Amiga OCS/ECS Still Snacking by Batman Group & Capsule
• I wonder why it's repeating the same movement over and over...
  
  
  Quote:

  Are we allowing disabling some/all audio dma now though?

  
  No. ;)
• rulezadded on the 2025-07-24 16:11:34
• 64k cracktro Amiga OCS/ECS Twelve Jokers In A Deck by Loonies [web] & The Twitch Elite
• Quote:

  Is any of these attempts using the technique that Flash Production did in
  "Danish Knowhow" from 1990?
  https://www.pouet.net/prod.php?which=27125
  
  There is a copper splitter on each 4 pixels.
  It uses 5 bpl with %00001111 masks to set all different bitplane 31 colors, and then it sets 32 background colors to see between the 4 px bitplane gaps.
  That makes at least 63 colors that way.

  
  Interesting. I count 52 columns on the effect.
  
  It's a pure copper display, running in 4 bitplanes and using DMA-sourced sprites for the upper colors. Since it's 4x1, not 4x4, there is no copper looping involved, so it just needs one WAIT per scanline on top of the color changes, leaving plenty of time to set 52 colors. The effect itself is drawn with the blitter.
• isokadded on the 2025-07-22 10:59:51