Maximum file size for demo (esp. @Assembly)
category: general [glöplog]
it amuses me to see people who more or less solely work on tiny intros or on other platforms, discussing the file size limit in the pc democompo they are highly unlikely themselves to enter. real-time is itself enough of a (very real, challenging) limitation.
that said, i totally think we should ban aga from amigacompos - a500 is where the action is!
that said, i totally think we should ban aga from amigacompos - a500 is where the action is!
Quote:
that said, i totally think we should ban aga from amigacompos - a500 is where the action is!
I actually wouldn't mind A500-only compos :D
Haha, I agree - but am I allowed to having zero A500 releases to my name? :)
This. I can't compare myself to you in terms of demomaking, let's get that out of the way first. But I've ran into plenty people who are very much into this circlejerk mentality that boils down to them either saying or thinking, or both, that doing a PC demo is easy as compared to size-constrained categories.
It's not.
Quote:
it amuses me to see people who more or less solely work on tiny intros or on other platforms, discussing the file size limit in the pc democompo they are highly unlikely themselves to enter. real-time is itself enough of a (very real, challenging) limitation.
This. I can't compare myself to you in terms of demomaking, let's get that out of the way first. But I've ran into plenty people who are very much into this circlejerk mentality that boils down to them either saying or thinking, or both, that doing a PC demo is easy as compared to size-constrained categories.
It's not.
OK, as an oldschooler I kept my mouth shut for long enough.
However, let us get real. As a community, what do we find more interesting: to see what Navis and his friends can do with 256Mb of volumetric data, or to bitch endlessly about o tempora, o mores?
Frankly, the recent obsession with size-limited prods gets me. Yes, saying that something shiny fits into 4K or whatever similar size is an easy way to explain demoscene to an outsider. At least some of the demoscene. I can see the temptation there. But until the day comes when we get Numb Res or Midnight Run or another prod of similar quality in 4K, let us not force arbitrary exercises in masochism upon everyone still on the scene.
However, let us get real. As a community, what do we find more interesting: to see what Navis and his friends can do with 256Mb of volumetric data, or to bitch endlessly about o tempora, o mores?
Frankly, the recent obsession with size-limited prods gets me. Yes, saying that something shiny fits into 4K or whatever similar size is an easy way to explain demoscene to an outsider. At least some of the demoscene. I can see the temptation there. But until the day comes when we get Numb Res or Midnight Run or another prod of similar quality in 4K, let us not force arbitrary exercises in masochism upon everyone still on the scene.
Quote:
that said, i totally think we should ban aga from amigacompos - a500 is where the action is!
but amiga demos then should be entitled to have the same size rules, i.e. no file size limit for them as well (so i can stream shitloads of pre-calced triangle vertex positions from HDD for each frame). revision has a current limit of meager 20MB.
Quote:
but amiga demos then should be entitled to have the same size rules, i.e. no file size limit for them as well (so i can stream shitloads of pre-calced triangle vertex positions from HDD for each frame). revision has a current limit of meager 20MB.
Agreed. Use as many floppies as you like.
it amuses me to see people discussing the file size limit in the pc democompo they are utterly unlikely themselves to organize.
Quote:
it amuses me to see people discussing the file size limit in the pc democompo they are utterly unlikely themselves to organize.
Now usually you're *the* voice of reason, bar none. But care to elaborate?
I feel we're all entitled to have an opinion for starters and it's been said plenty times that eventually it all comes down to what an organizer team decides for their own specific set of beliefs, opinions and technical/practical considerations.
So.. ? :)
soon the demoscene will discover MEGATEXTURE(TM) technology
Smash: even if I don't make demos, I'm still watching the demo compo and voting. And I'd like that to be an informed vote.
My 2 cents: the demoscene is based on a mix of technical and artistic skills. And as time goes by, the focus has shifted from the technical to the artistic side. Part of the reason is that everybody instinctively knows if he likes what he sees, whereas judging technical skill takes (arcane) knowledge. Is this a new effect or has it been done before? What are the limitations of the platform? Could this be done faster/in a smaller size? Who cares, it's so much easier to judge the color scheme!
That's why I want the organizers to help the public out: show the filesize of the entry, show the requirements (memory, nvidia-only? etc). Heck, I'd even like to see a FPS graph on a midrange card on the beamslide, because compo PCs are so much more powerful than even sceners PCs, that limiting yourself to midrange cards means handicapping yourself in the compo, and nobody will appreciate it until after the party when they try to run the entries on their home PC. (Yeah, I know orgos have enough work already, but I can dream. Of automated Partymeister FPS-graph generation!)
I'm perfectly happy to have no file limitation in return for the filesize on the beamslide. If 2 demos have exactly the same visuals/audio but the first is half the size of the other, that makes the first objectively better and it deserves more votes. Navis insistence that it's up to the creators to decide to share this info feels disingenuous to me, of course this will only be shared if the creators think it will benefit them (hello Debris/Farbrausch).
If you want to use a gigabyte of volumetric data, fine, but don't hide the filesize to fool people into thinking you generated it procedural at runtime. It's up to you to use it in an impressive enough way that the public appreciates it despite the bigger size.
My 2 cents: the demoscene is based on a mix of technical and artistic skills. And as time goes by, the focus has shifted from the technical to the artistic side. Part of the reason is that everybody instinctively knows if he likes what he sees, whereas judging technical skill takes (arcane) knowledge. Is this a new effect or has it been done before? What are the limitations of the platform? Could this be done faster/in a smaller size? Who cares, it's so much easier to judge the color scheme!
That's why I want the organizers to help the public out: show the filesize of the entry, show the requirements (memory, nvidia-only? etc). Heck, I'd even like to see a FPS graph on a midrange card on the beamslide, because compo PCs are so much more powerful than even sceners PCs, that limiting yourself to midrange cards means handicapping yourself in the compo, and nobody will appreciate it until after the party when they try to run the entries on their home PC. (Yeah, I know orgos have enough work already, but I can dream. Of automated Partymeister FPS-graph generation!)
I'm perfectly happy to have no file limitation in return for the filesize on the beamslide. If 2 demos have exactly the same visuals/audio but the first is half the size of the other, that makes the first objectively better and it deserves more votes. Navis insistence that it's up to the creators to decide to share this info feels disingenuous to me, of course this will only be shared if the creators think it will benefit them (hello Debris/Farbrausch).
If you want to use a gigabyte of volumetric data, fine, but don't hide the filesize to fool people into thinking you generated it procedural at runtime. It's up to you to use it in an impressive enough way that the public appreciates it despite the bigger size.
Quote:
heck, I'd even like to see a FPS graph on a midrange card on the beamslide
Do you honestly believe this registers with people watching and even if it does has any kind of impact that carries significance over the overall experience?
Come on :)
Superplek: I know it's not going to happen, I'm just stating my own preference.
I said graph because average FPS can easily be gamed. But even if there'd be just some 5-star rating (1 star = needs dual Titans for 60 FPS, 5 stars = Intel 4000 GPU), I think people would keep that in mind when watching the demo. The answer to "Could MY pc run this?" can legitimately influence voting.
Then again, I might be in the minority, running demos instead of clicking the YouTube link :)
I said graph because average FPS can easily be gamed. But even if there'd be just some 5-star rating (1 star = needs dual Titans for 60 FPS, 5 stars = Intel 4000 GPU), I think people would keep that in mind when watching the demo. The answer to "Could MY pc run this?" can legitimately influence voting.
Then again, I might be in the minority, running demos instead of clicking the YouTube link :)
Seven: this is pressuring coders to write stuff for mid/low end cards. I can see why that’s desirable, but when we’re doing this unpaid, for own own enjoyment, we generally prefer to decide what hardware we want to target and where to spend our time ourselves.
I guess this is what it all boils down to: some want demos that run well on older hardware, some want demos size coded... well, sorry, but no. I’ll make what I want to make, and if that doesn’t fit the compo requirements I won’t make it at all, since I have other stuff to be doing.
I guess this is what it all boils down to: some want demos that run well on older hardware, some want demos size coded... well, sorry, but no. I’ll make what I want to make, and if that doesn’t fit the compo requirements I won’t make it at all, since I have other stuff to be doing.
Quote:
If 2 demos have exactly the same visuals/audio but the first is half the size of the other, that makes the first objectively better and it deserves more votes.
Why? Why is it objectively better?
Also, keep in mind that a lot of the stuff we do isn’t obvious. A new rendering technique might need models storing in formats that eat disk space, and that fancy lighting might cost a lot of FPS. Probably only those of us working on this stuff will understand why, and that’s fine. But if you show the FPS on a midrange card or the file size and people don’t understand the details, you could be punishing people for doing cool new things.
Where I would like to see the file size though: size coded stuff. Often after a compo I’ve learned one entry is say 30KB, and in a compo where size is the focus that’s meaningful.
Where I would like to see the file size though: size coded stuff. Often after a compo I’ve learned one entry is say 30KB, and in a compo where size is the focus that’s meaningful.
A few things worth addressing.
I don't. And if so, there have been competitions in the past for lo-fi PC platforms (to name an example), demos that'd run on Intel HD sets/netbooks basically. This is a coarser but more comprehensible way to achieve the same.
First of all that's a pure hypothetical. Secondly, who are you to define what's deserving of a vote? The experience anyone gets from any production is a very personal matter that more often than not does *not* boil down to numbers. Add to that you'd be a making a snap judgement not being in the know of the hurdles, both human and technologically, that lie at the source of what you're seeing.
It's. Just. Not. That. Simple.
Quote:
But even if there'd be just some 5-star rating (1 star = needs dual Titans for 60 FPS, 5 stars = Intel 4000 GPU), I think people would keep that in mind when watching the demo. The answer to "Could MY pc run this?" can legitimately influence voting.
I don't. And if so, there have been competitions in the past for lo-fi PC platforms (to name an example), demos that'd run on Intel HD sets/netbooks basically. This is a coarser but more comprehensible way to achieve the same.
Quote:
If 2 demos have exactly the same visuals/audio but the first is half the size of the other, that makes the first objectively better and it deserves more votes.
First of all that's a pure hypothetical. Secondly, who are you to define what's deserving of a vote? The experience anyone gets from any production is a very personal matter that more often than not does *not* boil down to numbers. Add to that you'd be a making a snap judgement not being in the know of the hurdles, both human and technologically, that lie at the source of what you're seeing.
It's. Just. Not. That. Simple.
yeah, let's have digitalfoundry analyze democompo entries :) we need in-depth frame analysis, FPS fluctuation meters and resolution pixel counting.
of course they would need a highly secret demoscene debugkit and a signed NDA with Stiegler Legal.
of course they would need a highly secret demoscene debugkit and a signed NDA with Stiegler Legal.
That'd be really cool though, "<...> technology, powered by Stiegler Legal"
Preacher & Superplek: I'm making the (hypothetical,yes) point that if demos are artistically identical, you can judge them objectively on technical grounds. Most people would agree that smaller and faster is better than bigger and slower, no? But you need to know that info before you can take it in account, hence I'd like to see it on the beamslide, so I can vote for a mix of artistical and technical resons. I'm obviously not forcing anyone to vote automatically for smaller/faster entries.
Psonice: not showing such info pressures coders to write to the compo PC specs (if you want to win, which I assume most people want). And now you're punishing people who care about running on average hardware, and people who like to run compo winners at home without buying a 600$ videocard...
Look, I know we can discuss endlessly about this, and it's a matter of personal opinion. If a party organizer reads this tread, I just want to cast my vote to "if no size limits, please show filesize". The rest was not relevant to the thread, so I shouldn't have mentioned it, sorry for that.
Psonice: not showing such info pressures coders to write to the compo PC specs (if you want to win, which I assume most people want). And now you're punishing people who care about running on average hardware, and people who like to run compo winners at home without buying a 600$ videocard...
Look, I know we can discuss endlessly about this, and it's a matter of personal opinion. If a party organizer reads this tread, I just want to cast my vote to "if no size limits, please show filesize". The rest was not relevant to the thread, so I shouldn't have mentioned it, sorry for that.
Quote:
you can judge them objectively on technical grounds
But why would size be somehow relevant? You would be assuming that the size of the binary somehow correlates with the technical quality, which is quite a dubious assertion in my opinion. I think metrics like source code quality (how would you measure that? via some static analysis tool?) or compatibility of the demo with different sets of hardware (which, incidentally, would probably correlate negatively with the binary size. It's smaller if you take out error handling) would speak a lot more for the technical quality than a random number like that. And assuming identical source code, would the person who happened to have a slightly different set of compiler flags win?
I guess in the end it’s down to what the people making the actual demos want to do, plus what’s actually practical for the organisers.
(Currently working on a 4/8k here... not sure if the one after that will be intro or demo, depends on how the tech works out but if it ends up as a demo >256mb would be welcome as I’ll easily hit it :)
(Currently working on a 4/8k here... not sure if the one after that will be intro or demo, depends on how the tech works out but if it ends up as a demo >256mb would be welcome as I’ll easily hit it :)
Quote:
I'm making the (hypothetical,yes) point that if demos are artistically identical, you can judge them objectively on technical grounds. Most people would agree that smaller and faster is better than bigger and slower, no?
That's never true, so that remains in the realm of hypotheticals. I've written tight code and that's basically all I'm good for - but usually a lack of discipline and content had me coming in second or flat out losing to whatever pleased the viewers more. I don't think any excuse of the "but my cycle count / technique" kind would've been of any help and it's really not what matters, at least not as significantly as it once did (pre-2000s).
It's all very personal, as I said before. Remember The Popular Demo? Not only was it a killer performance, it also turned out to run smooth on a GeForce2 MX. But that's something you figure out after the fact (and sure we ran demos back then instead of watching them on YT, but if I'm on the go with my tiny MacBook it's a damn good solution to stream HD video and enjoy anyway) and if you were into that kind of shit you'd go "oh shit that's fucking neat". Not everything that's not immediately on the surface needs praise, does it? Sleep tight knowing that like minded people will recognize your effort.
I'm here as a party goer and compo watcher. And I say that increasing what already seemed like a big size limit to several gigabytes, feels like it would make the demo compo even less, not more, interesting to watch, for me. It already feels like an anti-climactic wild-compo finish for a big party. I'd like to be more impressed and more in awe. But what is the thing that's supposed to impress me? Certainly not the size limit. The music? A lot of demos have crap and/or depressing or otherwise "artistic" music I'd rather not listen to, and boring and/or depressing visuals I'd rather not have to watch again. Am I supposed to judge it as general video art? Or video art that was made by an amateur, so it deserves more pity compensation? I don't know. Making the size limit SMALLER would feel like the better thing to do, because then it would bring the demo compo (which they seem to want to keep as the "main" compo) more towards an area where I'm able to get "wow" feelings.
Usually, if I haven't gone to sleep before the demo compo (which has happened more often than not in the past years), I end up voting for things like DEMO2 by Ekspert. It's funny, not depressing, and has groovy music. Might be made with Unity or Red Sector Demomaker or whatever, I don't care.
I don't pretend to be any kind of artsy fartsy art hippie who "understands" the "pure art" side of things.
Anyway, maybe removing all size restrictions will break flood barriers and bring in people who can make demos great again. ;) At least I hope it will bring in more entries, because I think that with quantity comes quality as well, eventually. Having lots of entries is a sign of a healthy scene, IMO.
Usually, if I haven't gone to sleep before the demo compo (which has happened more often than not in the past years), I end up voting for things like DEMO2 by Ekspert. It's funny, not depressing, and has groovy music. Might be made with Unity or Red Sector Demomaker or whatever, I don't care.
I don't pretend to be any kind of artsy fartsy art hippie who "understands" the "pure art" side of things.
Anyway, maybe removing all size restrictions will break flood barriers and bring in people who can make demos great again. ;) At least I hope it will bring in more entries, because I think that with quantity comes quality as well, eventually. Having lots of entries is a sign of a healthy scene, IMO.
Preacher: I'm not talking about the last few kilobytes, which are rounding errors in modern demos sizes. In my example, I talked about a factor 2, and Navis asked about 600+MB data.
"Pure" movie players may be forbidden by compo rules, but there's a whole continuum of faking impressive effects with a ton of data. Are you calculating that fractal zoom realtime, or simply zooming into precalculated images ? Are you doing real fluid dynamics on a million particles, or just interpolating between a ton of keyframes? That's what I'm talking about, and knowing the filesize can help me judge what is real vs faked (or faked cleverly vs stupidly)
"Pure" movie players may be forbidden by compo rules, but there's a whole continuum of faking impressive effects with a ton of data. Are you calculating that fractal zoom realtime, or simply zooming into precalculated images ? Are you doing real fluid dynamics on a million particles, or just interpolating between a ton of keyframes? That's what I'm talking about, and knowing the filesize can help me judge what is real vs faked (or faked cleverly vs stupidly)
Seven: could you tell the difference between 600MB of keyframe data and 600MB of ML data used to generate those fluids in real-time? One would be an animation, the other would be a very, very cool effect.