The 64 MB limit at BP07
category: general [glöplog]
Quote:
But what I need is a reason that can't just as well be used to argue that prerendered effects work better than realtime effects.
The last three-four demotunes I made needed more CPU-power than I had, so tracks needed to be rendered to audio and re-imported into Cubase so I could work on other parts. That system was an athlon xp 1800+ with a gig of ram. That's a bit ancient really, but consider that was running without any 3D-graphics at all. If I we're to run the project in realtime, without any rendered audio tracks, I'd probably need something about a 3000+ athlon or something.
The thing is that while "everyone" has pretty capable 3D-cards, very few has NI Powercores or similar DSP-hardware connected to their computers. So modern computers can have heavily hardware acceleration for graphics, but nothing which can be used for music. Some higher end gaming soundcards have hardware effects, but demomakers which rely on demosceners to have the latest and greatest Soundblaster models will most likely be flamed to the moon and back by their fellow sceners, in the same way those who have made demos for high-end graphic cards in the past have been. Anyway, the hardware effects on gaming soundcards aren't really all that useful for music anyway.
So there, happy now? :)
Quote:
Anyway, the hardware effects on gaming soundcards aren't really all that useful for music anyway.
http://209.85.135.104/search?q=cache:_TbPuinDkWQJ:arstechnica.com/news.ars/post/20060526-6932.html+GPU+%2Bdsp&hl=da&gl=dk&ct=clnk&cd=2
Sorry, I would have found a better URL if I had the time...
What I wanted to say, was: You could use a GPU as a DSP....
yes, except that the AGP transfer backwards would prolly take more time than simply running it through CPU instructions.
@Nutman:
First, I started making music in 1990 on an Atari ST, in Noisetracker. The sound quality was even poorer than on the A500, and didn't even have stereo sound. In the demoscene, I have done music for 64k intros (with and without custom softsynths, both on PC and Amiga), for GBA-demos (4chn .mod with almost no support for any controlcodes), for Amiga-demos (being a Spaceballs-member and all) and for PC-demos (both modules and rendered audio). So no need to suggest I have a go at restricting myself for challenges, but thank you.
Also, this is a discussion about the _PC_ demo compo. In the context of PC-demos, I welcome the opportunity to encode my soundtracks in as good quality as possible. There is no logical reason to restrict musicians to lower bitrates, since it doesn't say anything at all about neither the skills of the musician or the coder. It's a pointless limitation, which benefits none.
And oh.. 32mb equals to about 15 minutes of 320kbps CBR MP3-compressed audio, in 44khz 16 bit stereo. So why you think that the extra 32mb of filesize is because of musicians is beyond me. However, if you have read the thread, you'd probably noticed that Smash^FLT has some ideas on how to use the extra space. For me, I'd like a couple of megabytes more, so that I can encode in something which I find acceptable, ie 192 VBR MP3 or OGG at equivalent rates.
First, I started making music in 1990 on an Atari ST, in Noisetracker. The sound quality was even poorer than on the A500, and didn't even have stereo sound. In the demoscene, I have done music for 64k intros (with and without custom softsynths, both on PC and Amiga), for GBA-demos (4chn .mod with almost no support for any controlcodes), for Amiga-demos (being a Spaceballs-member and all) and for PC-demos (both modules and rendered audio). So no need to suggest I have a go at restricting myself for challenges, but thank you.
Also, this is a discussion about the _PC_ demo compo. In the context of PC-demos, I welcome the opportunity to encode my soundtracks in as good quality as possible. There is no logical reason to restrict musicians to lower bitrates, since it doesn't say anything at all about neither the skills of the musician or the coder. It's a pointless limitation, which benefits none.
And oh.. 32mb equals to about 15 minutes of 320kbps CBR MP3-compressed audio, in 44khz 16 bit stereo. So why you think that the extra 32mb of filesize is because of musicians is beyond me. However, if you have read the thread, you'd probably noticed that Smash^FLT has some ideas on how to use the extra space. For me, I'd like a couple of megabytes more, so that I can encode in something which I find acceptable, ie 192 VBR MP3 or OGG at equivalent rates.
lug00ber: So what you're saying is prerendering audio lets you do things that aren't possible (yet) in realtime. It's still not the reason I'm looking for. Because prerendering effects also lets you do things that aren't possible (yet) in realtime.
ryg: But what about the inevitable near future when PCs can handle video in a resolution and framerate that's comparable to what demos use? Will realtime effects be pointless then?
About realtime audio: complex numbers means basically "let i^2=-1", and linear algebra should be easy for coders to pick up too, though I guess all the rest of the theory is pretty inaccessible. But shouldn't that just make it more interesting? Or is it only fun if the first hit on Google tells you exactly how to do it? ;)
ryg: But what about the inevitable near future when PCs can handle video in a resolution and framerate that's comparable to what demos use? Will realtime effects be pointless then?
About realtime audio: complex numbers means basically "let i^2=-1", and linear algebra should be easy for coders to pick up too, though I guess all the rest of the theory is pretty inaccessible. But shouldn't that just make it more interesting? Or is it only fun if the first hit on Google tells you exactly how to do it? ;)
Quote:
Lately I can think of ASD's productions as having real standout music that sets them apart. Just think what those would be with the standard techno beat?
Even though I agree with you on principle, I think that was a poorly chosen example, as _any_ ASD-demo would set themselves apart from the rest, even with "standard techno-beat". :) Okay, so I might prefer a bigger and beatisher approach, but still; put any half-decent soundtrack to a Navis production, sync it up properly, and you have a hell of a demo -- progrock or no progrock. :)
..and let me just clarify what I meant by my statement earlier; people care about the soundtrack of a demo, of course they do. The difference is just that as long as the soundtrack is "okay", the demo sure as hell gets more attention (and so it should; it is only natural). Now; I've heard plenty of _outstanding_ soundtracks in demos as well, but the bad thing is that the rest of the demo rarely follows up on the quality.
There are exceptions of course (Kewlers, MFX, ASD and Fairlight-demos comes to mind) where the soundtrack and the demo seems almost fused together in perfect harmony, but this is very rare if you use the total of all demos released as a baseline.
..and to whoever said it; no - the musicians are _definitively_ not the ones pushing for larger filesizes in compos. :) I think I used 80 kbps VBR Ogg Vorbis-files for my latest demo-soundtracks and I'm pretty darn happy with the results. I guess I could have used higher bitrates as well, but I'm also kind of old fashioned when it comes to "wasting" bytes. :) In that regard I really have no objection to letting the 3D-animator/modeller have my "spare" megabyte, as I am fully aware that more people will appreciate a kickass 3D-scene than slightly better soundquality in the top-15kHz area. :)
Quote:
lug00ber: So what you're saying is prerendering audio lets you do things that aren't possible (yet) in realtime. It's still not the reason I'm looking for. Because prerendering effects also lets you do things that aren't possible (yet) in realtime.
What a pointless argument. What reason _are_ you looking for? I'm pretty sure both ryg, lug00ber and I have given you plenty of good reasons why doing music realtime is a bad idea:
- It is extremely hard to create a softsynth that has high enough quality/diverse enough sounds to compete with pre-rendered music.
- IF, somehow, someone DID make one, it would require the musician to learn a new tool instead of the one he is used to already = idiotic.
- IF, again somehow, the Super Ultra Softsynth 2000(tm) existed, it would take up _massive_ amounts of memory and CPU-power -- and for what? Proving that the programmer is talented? Visuals are better for that (as I think Gargaj and pretty much everyone else already have made pretty darn clear)
- Doing things the hard (= impossible) way just for the sake of it is pretty stupid. Sure; when someone creates a MAD COOL GBA or Amiga-demo it impresses you, but guess what; it's pretty much all fake anyway.
Let me rephrase that last point: I love it when people create "impossible" things in demos, it's just that doing it when 0.5% of the demoscene "gets it" it starts to become pointless. I mean; sure -- I guess the next excess or portal process-demo could be coded upside down, or that all the 3D-models are created on Lightwave running on an unexpanded A1200.. or the all of the music could be made by sampling sounds made by alaskan salmon in pink sombreros, but what the hell would the point be?
There comes a point when doing things "just for the challenge" becomes plain old stupid. :)
There comes a point when doing things "just for the challenge" becomes plain old stupid. :)
Doom: I will accept the argument lost if you would agree that all musicians who ever sampled a synthesizer are as pathetic as me for using prerendered sounds when they should have their had coder make a realtime softsynth for synthesized sounds. (Samples of instruments such as drums, or vocals for that matter, should be acceptable to all, unless you believe that instead of sampling coders should do a synth that enables physical modelling of every object in the world that can make a sound.)
Quote:
complex numbers means basically "let i^2=-1"
Slightly more to it than that (massive understatement) and it's a fairly hideous subject even for anyone confident with maths.
parapete: Many people consider it the most un-hideous subject of maths.
Apart from that, I can not believe you need to _understand_ infinite series to get a grip on sound engineering. At least not to get an OK understanding of what is going on. It's just like 3d-vector maths. Very few people are really aware of what is going on there, and still there are some cool demos around doing 3d :)
(this is just my feeling. I have no clue on signals&waves myself.)
Apart from that, I can not believe you need to _understand_ infinite series to get a grip on sound engineering. At least not to get an OK understanding of what is going on. It's just like 3d-vector maths. Very few people are really aware of what is going on there, and still there are some cool demos around doing 3d :)
(this is just my feeling. I have no clue on signals&waves myself.)
Quote:
I guess the next excess or portal process-demo could be coded upside down
the Excess demos are already running upside down on my machine :(
gloom: Like I said, I'm looking for a reason that only justifies prerendered audio and doesn't justify realtime visuals. Look:
- It is extremely hard to create realtime 3D that has enough detail/quality to compete with the output of 3DS MAX.
- IF, somehow, someone did make all this possible, it'd be much harder to implement than just using the output of 3DS MAX.
- Realtime 3D takes up massive amounts of memory and CPU power.
- Doing things the hard (= impossible) way just for the sake of it is pretty stupid.
But thank you for pointing out one difference, and no, you didn't make that clear before, namely that visuals are better for showing the programmer's talents. So I've heard these reasons for why visual coding is a place for programmers to show their talent but audio coding isn't:
1. Visuals are a better way for the coder to show off
2. Audio coding is very difficult and the techniques are little-known
3. Realtime visuals are better in terms of image quality and framerate than video, at least right now, as opposed to realtime audio vs. streams
This is good, but: 1 and 2 can be seen as disagreeing somewhat. 1 implies that it's important for the coder to show off, while 3 implies this is or will soon be irrelevant, it's more about picture quality. 3 in itself will be moot in a couple of years, 2 in itself is seriously questionable (shouldn't coders embrace the challenge?), and 1 begs the question if the coder can't show off even better if he has unlimited CPU power (ie. prerenders all his stuff completely).
Please don't get me wrong, I'm not trying to prove a point as much as to figure out what to expect if I wanted to consider switching to the PC. The emphasis on realtime graphics which in no way applies to audio has me thoroughly puzzled. Those three reasons above do clarify it a little even if it's still fuzzy to me.
- It is extremely hard to create realtime 3D that has enough detail/quality to compete with the output of 3DS MAX.
- IF, somehow, someone did make all this possible, it'd be much harder to implement than just using the output of 3DS MAX.
- Realtime 3D takes up massive amounts of memory and CPU power.
- Doing things the hard (= impossible) way just for the sake of it is pretty stupid.
But thank you for pointing out one difference, and no, you didn't make that clear before, namely that visuals are better for showing the programmer's talents. So I've heard these reasons for why visual coding is a place for programmers to show their talent but audio coding isn't:
1. Visuals are a better way for the coder to show off
2. Audio coding is very difficult and the techniques are little-known
3. Realtime visuals are better in terms of image quality and framerate than video, at least right now, as opposed to realtime audio vs. streams
This is good, but: 1 and 2 can be seen as disagreeing somewhat. 1 implies that it's important for the coder to show off, while 3 implies this is or will soon be irrelevant, it's more about picture quality. 3 in itself will be moot in a couple of years, 2 in itself is seriously questionable (shouldn't coders embrace the challenge?), and 1 begs the question if the coder can't show off even better if he has unlimited CPU power (ie. prerenders all his stuff completely).
Please don't get me wrong, I'm not trying to prove a point as much as to figure out what to expect if I wanted to consider switching to the PC. The emphasis on realtime graphics which in no way applies to audio has me thoroughly puzzled. Those three reasons above do clarify it a little even if it's still fuzzy to me.
If we were all blind, then demos would have kick ass audio effects instead of music.
eh.. first line should be "doesn't justify prerendered visuals"
Blech.
Blech.
@_-__-_: Perhaps that's why there are no smell effects. (Or are there? I've never been to a demoparty)
This is starting to sound more and more like the arguments a month or so ago about tracked versus prerendered music, which eventually seemed to come down to one side claiming that the tracker was superior because it required far more creativity to produce sound in a well-defined environment, and the other side claiming that most of that 'creativity' had nothing to do with musical ability.
I dunno, if you want a next generation of audio soundtracks, start rendering the music as 5.1 audio and include lots of positional audio effects (when the letters in that scroller fly in from behind I wanna FEEL them, damnit!). Granted, it'll be lost on nearly all people who may have two speakers and a subwoofer (or maybe an AC'97 card and a pair of headphones), but hey, it's time for better equipment (and unlike video, I doubt downmixing would lose 'framerate'). Also, use Vorbis 1.12 instead of Vorbis 1.00.
*cough*
what?
This is starting to sound more and more like the arguments a month or so ago about tracked versus prerendered music, which eventually seemed to come down to one side claiming that the tracker was superior because it required far more creativity to produce sound in a well-defined environment, and the other side claiming that most of that 'creativity' had nothing to do with musical ability.
I dunno, if you want a next generation of audio soundtracks, start rendering the music as 5.1 audio and include lots of positional audio effects (when the letters in that scroller fly in from behind I wanna FEEL them, damnit!). Granted, it'll be lost on nearly all people who may have two speakers and a subwoofer (or maybe an AC'97 card and a pair of headphones), but hey, it's time for better equipment (and unlike video, I doubt downmixing would lose 'framerate'). Also, use Vorbis 1.12 instead of Vorbis 1.00.
*cough*
what?
It's not only about understanding DSP or complex numbers. I'm a bit what out of spain is called "electrical engineer"; I very well know all the theory of fourier series, fourier transform, laplace transform, cosine transform and of course the queen of the family - the zeta transform. I have also applied all this theory many times. I also know designing analog and digital filters, I now how LPC works, I have done my own image and sound compressors. Still, I'm completely unable to generate a stupid sound out of noise/LFOs/filters/reverb machines. Why? I think, you need more than knowing DSP theory and programming to build a soft synth. I strongly believe that you _also_ need to be a musician , _and_ a frequent user of one or more sequencers. Now, find more than 4 such coders in the demoscene.
[the other approach, physical synthesis (that's what we used in our last prod) is more "simple" to understand, but definitively this is an even less documented area than traditional synthesis.]
Still, I wonder if indeed everyone really understands the slerp() of quaternion interpolation or the inverse transpose for normal vectors transformation. It obviously doesn't prevent that people from creating 3D effects...
[the other approach, physical synthesis (that's what we used in our last prod) is more "simple" to understand, but definitively this is an even less documented area than traditional synthesis.]
Still, I wonder if indeed everyone really understands the slerp() of quaternion interpolation or the inverse transpose for normal vectors transformation. It obviously doesn't prevent that people from creating 3D effects...
Hyde : Well I meant hideously difficult, rather than inelegant. Opinions will vary on that anyway...
Seeing where this discussion goes, there's just one thing to point out: There's no reason for a higher sizelimit, since the audience won't be able to see whether it's hires textures and resolution on the bigscreen anyway, and likewise with music - in the heat of the Bingen morning, it doesn't matter whether it's a 128kb or 320kb MP3, since the speakers will probably mostly play the bassdrum anyway ... Well, who am I to be bothered, I have other priorities than updating my hardware ;).
Anyway, where does this all go? Well, where it has been going for quite a few years now, which means that no matter what quality the textures are - in pc demos, the most important thing is the script, the storyline, the 3D and colourscheme and the theme and how it all fits together, rather than the coding ... If big 3D-demos are your game, then the 64Mb will make it look a bit nicer, but ultimately, it's the above that counts - so, to use a phrase used here before about music: Bad code can spoil the production, but is the really good code noticed in a 3D accelerated demo? Hasn't the pc demo compo become a compo for the 3D artists rather than the coders? Well, basically, I guess this discussion about the sizelimit is all about what kind of demos we want ... Personally, I prefer demos, everyone can run without needing to buy a new gfxcard every year ...
Anyway, where does this all go? Well, where it has been going for quite a few years now, which means that no matter what quality the textures are - in pc demos, the most important thing is the script, the storyline, the 3D and colourscheme and the theme and how it all fits together, rather than the coding ... If big 3D-demos are your game, then the 64Mb will make it look a bit nicer, but ultimately, it's the above that counts - so, to use a phrase used here before about music: Bad code can spoil the production, but is the really good code noticed in a 3D accelerated demo? Hasn't the pc demo compo become a compo for the 3D artists rather than the coders? Well, basically, I guess this discussion about the sizelimit is all about what kind of demos we want ... Personally, I prefer demos, everyone can run without needing to buy a new gfxcard every year ...
Quote:
There's no reason for a higher sizelimit, since the audience won't be able to see whether it's hires textures and resolution on the bigscreen anyway, and likewise with music - in the heat of the Bingen morning, it doesn't matter whether it's a 128kb or 320kb MP3, since the speakers will probably mostly play the bassdrum anyway
Thousands people watch demos at home on a decent monitor and a sound system after a party compo you know...
See how many times famous demos are being downloaded on scene.org
make that, thousands of people watch demos at home on average spec systems :P
parapete: what is so hard about "complex numbers" ?
Yeah, they do, and nothing hinders the groups from releasing a HQ highsize version for those who want it. Even to be distributed at the party where they are released, a scenario which has been seen many times before.
Oh, and about the ongoing music/sound discussion discussion [offtopic?]: Music is not about sound quality. Guess I'll never be/never was particularly succesful, but I can't really be bothered about much else than the actual composition (the fun part)... All the fiddling about with making it sound good I find really boring and usually never get around to actually doing.
Oh, and about the ongoing music/sound discussion discussion [offtopic?]: Music is not about sound quality. Guess I'll never be/never was particularly succesful, but I can't really be bothered about much else than the actual composition (the fun part)... All the fiddling about with making it sound good I find really boring and usually never get around to actually doing.
Keops: Oh, and besides, how many of these downloads are followed by a sigh of disappointment from the user who can't see the production on his/her humble machine?