Jump to content

Sheddy

Members
  • Content Count

    803
  • Joined

  • Days Won

    1

Everything posted by Sheddy

  1. Much more basic than RMT though, certainly nothing a musician could do anything fabulous with. Just a slightly extended sfx player.
  2. This thread reminds me, I wanted to see why songs that end by just looping on a blank line suck so much CPU. All the songs I had for my project do that, but is there a better way of a song ending itself?
  3. Yes, and there's always the danger of "institutional" (unintended) bias even when attempting to be objective (which is kind of unlikely in said machines' forums anyway, if only for little jokey digs to the crowd). Only someone who loves both and is knowledgeable would give best balance (if they wanted)... though god knows what kind of cognitive dissonance they would be living with if they had to judge them against each other seriously!
  4. 👍Jimi looks fantastic, but the tiny man is always going to be a tough ask at a8 resolution
  5. Sorry if this is obvious, but you are decoding some rle bmp from Windows, converting to a8 and then doing rle suitable for a8? The rle in Windows will be very different than a8 so make sure picture is decoded properly from bmp
  6. So young. Very sad. He was very smart and ingenious and was a regular here about 10 years ago until a terrible cancer. Although he somehow got over that, it sounds like maybe it came back? ☹️
  7. I once knocked up a quick script where I assumed I could brute force best possible arrangement of what was normally 6 or 7 items, but could occasionally be up to 15. The time wasted trying to debug was a bit embarrassing after I went back to basics of permutations...
  8. Emkay Nice filter effects! But I don't know what you mean about 3 channels. 2 channel with drums on bass channel? Lot of songs share channel like that, or am I missing something?
  9. Not too surprisingly, TIA also distorts http://atariage.com/forums/topic/271920-tia-sound-abnormalities/?do=findComment&comment=3936913 Odd that 2600 emulators have only just started emulating this. Coders have to do "in tune" music with samples anyway, so maybe it isn't noticed much.
  10. Thanks! I was too impatient to wait till I got home to try it out myself.
  11. ARJ is really nice tebe👍 I assume that classic ARJ pic run on rapidus has to have rapidus changed to standard clock speed?
  12. The source pic and adjustments are the perfect balance between not too much fine detail and not too many colours bunched up. Excellent!
  13. I guess jedit is expecting text. It's a binary file, ie raw data. Most editors don't crash due to that though, which is totally down to the editor. You need something like a hex editor if you want to poke around the raw data
  14. Rastaconverter only does register writes on 2 cycle boundaries from looking at the code it generates. Not sure if that completely explains why it doesn't suffer glitches - It might play things even safer internally?
  15. Sadly, I lack the c++ understanding and skills too, but I believe the idea of the late acceptance hill climbing algorithm with its multiple solutions, is that the solutions with the worst mutations don't win out long term. But the single solution algorithm is unlikely to undo what turns out to be a dead end mutation, as mutating it back gives a worse fitness score, and would rely on multiple simultaneous mutations to do so. Yeah, you could probably fine tune things somehow by applying lots of rules and special cases on top, but the beauty of these algorithms is that with the right selection pressure and mutations, it isn't necessary, and they could even be detrimental to some case you hadn't foreseen. Late acceptance Hill climbing isn't the pinnacle (no pun intended) of such algorithms, so others could be worth looking at too. Not sure if it's already in there, but having an added selection pressure for fewer register changes when picture distance is about the same could be interesting
  16. Race condition seems unlikely to me, but then again, this is Windows we're talking about. Certainly worth making sure.
  17. If you're maxing out all your cores with Rastaconverter, that's not really Altirra's fault! It needs more CPU than atari800 as it is doing a much more accurate emulation. You could try running it at a higher priority than Rastaconverter so it gets more of a look in on the CPU.
  18. The problem that stands out is that the low/high setting of addresses at the start are round the wrong way. should have lda #<address for sta register, and lda #>address for sta register+1.
  19. Referring back to the RasterConverter picture thread, there is a small amount of interest in a slideshow program for RastaConverter pictures. Although people have found it is just about possible with some DOSes to chain load the xex files without them crashing on exit, as suggested, it would be nice to be able to rely on an already running VBI to pick up the real time clock and to do other slideshow things. I would personally want them to work as stand alone pictures as well, which I think ought to be possible. I am working on a decompiler/optimiser script that should help make existing xex pictures more amenable to tweaks of the default build program. Working on that, I only recently realised it would ideally need to be able to decompile the more optimised program kernel too (Doh!). That is a bit harder to parse. At that stage a better way would need a proper format for RastaConverter, at least for the program kernel. Maybe a bytecode that gets compiled at run time? Maybe that isn't needed if just a general compressor is used. The bitmap and sprites could certainly benefit from that. Will probably post a new topic asking about that some more.
  20. Thanks to pps for pointing me here. Not sure how I missed it before. I'm not sure how beneficial it would be, but there should be enough memory to leave DOS on in a 48K machine with a bit of reorganizing of the no_name memory space: I suppose the biggest Rasterconverter picture could stuff registers on every line in 54 of the 57 available cycles. 9x5* bytes=45 + the 2 filler bytes for the 3 remaining cycles = 47. *Corrected* For full 240 height: program: 240x47=11280 (ouch) bitmap: 240x40=9600 plus some alignment bytes players: 1024 DLIST: 720 +Viewer code total: <23k Still doable with DOS
  21. Glad that is possible to create slideshows with the existing xex files. What would be the most useful change in the default viewer program to support loading inside a proper slide program? Just asking as I'm taking a detour from doing my title screen to making a script that optimizes the raster program better and/or use a smaller DLIST. One of the goals is to convert the xex back if that's all you have. There's a small change in the viewer program to allow different DLISTS so making a small change to be more friendly to slideshow programs, should they be written, can be added. I don't know what DOS can feed back after the xex is returned from or whether a page 6 location or equivalent would be most useful?
  22. Lol Optimising compilers are complicated I don't suppose GCC supports 6502 though
×
×
  • Create New...