What do we need to listen? The engine as it revs, tires stressing against the road or other surfaces, wind, maybe the sound of a crash. What else?
What do we need to listen? The engine as it revs, tires stressing against the road or other surfaces, wind, maybe the sound of a crash. What else?
My bad, sorry about the attitude. Yes sometimes Im arrogant.
That said, I believe you agree with me then?No, as i dont feel samples can do that - i used to think this "samples sound fake" was a bunch of BS but playing GTR and i believed it - it sounds "WOW" at the start, but you start to realise it's repetitive and you can hear alot about the load on the engine etc through it's noise, which you just don't get with samples.
Cars do exist, thats not a problem. Or will you argue about that it would matter if the samples are taken from a ford k instead of a opel corsa? As for the resources needed. Thats true, but that is also why I said I wish they had them.
Regarding the flexibility of sound, nope that would be a mistake in programing, not in flexibility of sampled sounds. Again, take a look a the music industry.
Bad programing. Anything that can be synthesized can be sampled (or synthesized above the samples). And it will sound a hole lot better.
Just drive a lap in the F1 of rFactor and then in LFS. Sorry to say it, but it sounds pathetic in comparison.
If you mean that a sample can be re-mixed to make it different it doesn't help. The sample will still be the same and it will get repetitive.
Infinite number of samples would be of course better than synthetic sound, but that's hardly a realistic task to do. For example simple thing like listening the sound from two different directions makes a big difference. Basically synthetic sound will need a less complex sound engine when closing in on the realism.
Hmm we are not communicating. The sound of ANY car would be repetitive for your tastes, I guess
because its always the same!Again, if you knew something about sampled music you would understand. To replicate the sound of a piano, for example, I would need to sample an incredible amount of individual sounds. Right?
Wrong. I have to know what sounds I need to sample, and then, from them, extrapolate others. Hint, a change in the pitch can be done without having to sample the change in some factual waveform.
Absolutely not... What is better: A music that's being played on the instruments or one where it comes from a box playing the prerecorded notes?
Actually not. Real cars sound is never repetetitive, every little explosion inside the pistons sound different. Not to mention that everytime the engine load, throttle position, speed, position of the listener, position of the moon, ... the sound changes.
Real life has syntethic sound with infinite amount of variables
You could just create a similar sound without samples at all? By synthesizing, you know.
This is the same problem here too. A sample will always sound the same. So to simulate a piano sound accurataly you would indeed need a infinite amount of samples
That's just it. In real life many other things change in the sound, not just pitch or volume.
You've got something wrong with that Piano theory. How come proffesionals use real pianos and klaviers (it's klavier in English, right?). They use them, because a synthenizer is NOT AS GOOD. Piano doesn't only change the pitch, it changes a lot. If you've ever looked inside a piano, you'd know.
He's right that a sampled piano will always sound better than a synth piano, and of course a REAL piano will sound the best- but we can't use real engines in LFS now can we?
It's probably not a good idea to compare pianos and engines though. There are some very good synths around, I'm thinking especially of the Steam Pipe synth in Reaktor, which simulates air passing through a tube and can turn out some very passable likenesses of real flutes and other instruments.
The more a synth system actually properly models what it's trying to replicate, the better the result I imagine. For now, the LFS engine synth isn't doing too badly, but It's still in the simple early stages. If ever perfected, it would totally leave the sampled route for dead. It would be an incredible innovation and I hope Scawen gets there some day.
It's probably not a good idea to compare pianos and engines though. There are some very good synths around, I'm thinking especially of the Steam Pipe synth in Reaktor, which simulates air passing through a tube and can turn out some very passable likenesses of real flutes and other instruments.
The more a synth system actually properly models what it's trying to replicate, the better the result I imagine. For now, the LFS engine synth isn't doing too badly, but It's still in the simple early stages. If ever perfected, it would totally leave the sampled route for dead. It would be an incredible innovation and I hope Scawen gets there some day.
You are right, for music there are good synths, and there are even synths that are just effects for samples. Thats what I think LFS should have, some sampled sounds with synth effects.
But I dont see how just synth sounds could do it.
In any case, the question would be... when can we expect an improvement on LFS sound system?
Well, I happen to play the piano...and keyboards, and computers.
Well, my girlfriends used to recognize me by the sound of my car... go figure!
You are not really thinking now. Of course the car sounds similar all the time, but there are still subtle changes. For example try this next time you drive your car: 1st rev it and listen to the sound, then go driving first push throttle fully on 1st and listen to the sound then do the same thing only on half throttle. Now repeat on 3rd gear. Is the sound the exactly the same everytime? No imagine that there are 6 gears in the car,there are in theory infinite amount of different throttle positions and up t 20 000 different rpm amounts. Now that's just the beginning. The sound also changes when load is different for example when you are going uphill or downhill. And that's not even scratching the surface yet.
There are infinite amount of variables that do change the sound (and not just the pitch or volume). As I said with samples you would need infinite amount of them. With a synthetic sound you would need infinitely complex algorithm. In a sim you don't of course need to take all the variables in to account, but still when closing in to the real life sound a complex algorithm is easier to make than record millions of samples of different states of the car.
And just to make my point even clearer here is a even simpler simplification: A simple addition algorithm. A number plus a number equals. With samples you would basically only know that 1 + 2 = 3 and 2 + 3 = 5 (two answers). With a synthetic algorithm you would simply know the algorithm: a + b = c. a and b could be replaced by any two numbers which then the computer would calculate and show the result. With the algorithm you would know the sum of any two numbers.
Simulating a sound is harder because there are variables that aren't really known or making the algorithm needs too much computer power to be realistic. Samples are a shortcut and they only represent the final result. Which is fine if you can record enough samples which turn out to be the downfall of the sampled system because you can never have enough samples.
Also considering that the current system in LFS gives so much more info than the sampled systems in other games proves why synthetic is the better tech. The thing that is wrong with the current system is that there are not enough variables and the sound is not random enough. The sound waves are smooth compared to real life where the waves are not smooth at all. This is what makes the sound sound wrong compared to real life. Samples won't just magically fix the situation, samples only work in the situation where they were recorded. The do sound right on those few situations, but anywhere else they don't.
Phew, a long post. Hope it makes atleast somekind of sense, can't be bothered to edit it anymore
Great, now we are going somewhere. Ok, lets take the sound of your voice and my voice. They are different, yet, people can recognize them all the time, because they are "consistent with themselves".
Why don´t we hear simulations of a particular voice? Because it would be impossible to accomplish it with synthesized sound.
If I really want to emulate the voice of someone I should first SAMPLE his voice, then polish the model with the variables you talk about (the ones we can simulate or replicate of course) and then we could have something resembling the original. Not at rich you will say. Well of course! but RECOGNIZABLE.
Yes, sampled sounds are really different from the real thing, but are closer to it than simulated waves.
Okay, voices. Lets say you have 10 samples of voice. How do you make it speak? The answer, of course, is you can't. If you synthesis the sounds in real time, based on air velocity, mouth shape, larynx tension, then you can make any sound (albeit an approximation). The more variables you add, the richer and more realistic sounds you can produce.
Same with engines. You could have 100 samples, but there is no way you'll ever be able to manipulate those samples well enough to make every sound that car is capable of.
Then throw in engine damage. How do you sample this? Find an engine with the right amount/type of damage? You don't - simple. With synthesised sound you can do what you like, and it'll sound (given decent inputs and algorithms) like real damage.
GTR/GTL/rFactor sound terrible with samples (for engines - they are quite good for other sounds that don't 'vary').
Same with engines. You could have 100 samples, but there is no way you'll ever be able to manipulate those samples well enough to make every sound that car is capable of.
Then throw in engine damage. How do you sample this? Find an engine with the right amount/type of damage? You don't - simple. With synthesised sound you can do what you like, and it'll sound (given decent inputs and algorithms) like real damage.
GTR/GTL/rFactor sound terrible with samples (for engines - they are quite good for other sounds that don't 'vary').
Ill add my opinion on sound and graphics, ill start with the sound.
Personally, i prefer sampled sounds.. even when it doesn't give as much info to the driver as synthesized sound would do. I played gtr2 for a (very short) while.. (then i found out about LFS) but i was blown away by the sound of gtr2.. especially the shifting sounds. In my opinion this realistic sound adds alot to the driving experience. So after playing gtr2, i found out about LFS and i was kinda dissapointed with the sound.. but hey everything else is better so the choise was easy.
Now for the graphics:
I think LFS actually looks really good. So im happy.. but of course, if i had an expensive superfast pc i probably would be begging for more.
Personally, i prefer sampled sounds.. even when it doesn't give as much info to the driver as synthesized sound would do. I played gtr2 for a (very short) while.. (then i found out about LFS) but i was blown away by the sound of gtr2.. especially the shifting sounds. In my opinion this realistic sound adds alot to the driving experience. So after playing gtr2, i found out about LFS and i was kinda dissapointed with the sound.. but hey everything else is better so the choise was easy.
Now for the graphics:
I think LFS actually looks really good. So im happy.. but of course, if i had an expensive superfast pc i probably would be begging for more.
But cars don't really make a "shifting sound", do they? I don't hear it in any videos on the internet of race and road cars.
In the cockpit you definitly hear sound from changing gears. (from the gearbox etc) i just thought it sounded lovely in gtr2.. but i also loved the sampled engine sound.
This is a good point. But its relatively easy to keep current graphics for "not that fast" machines, while giving the ones who do have the ultimate gaming PC the proper eye candy.
In fact, its my believe that more people would buy LFS if it was "prettier". Not everybody is spartan, you know
What gearboxes do you drive? Real gearboxes make very little noise when you shift - if they did, something will break/wear very very quickly.
Most sounds in LFS are sampled, including the gear shift sound, if I'm not mistaken. Even the engine uses samples to do what it does. We just need more appropriate sounds to deal with situations as they arise. Hitting a stack of tires shouldn't sound the same as scraping across a guard-rail, etc. DaveWS created a nice thread of sound improvement ideas- I'll try and find it if someone doesn't beat me to it.
http://www.lfsforum.net/showth ... ghlight=sound+improvement
http://www.lfsforum.net/showth ... ghlight=sound+improvement
(wheel4hummer)
DELETED
by wheel4hummer
Yes they are - sort of. Even the engine sound comes down to a sample - the 'combustion noise', which is pretty constant, therefore can be sampled, and modified in real time by the parameters in Shift-A.
Some things could be added like that like realistic shadows and reflections to some extent. But some things need to work on everyones computer, things like realistic smoke and blinding reflections and weather visuals. Having those disabled would be a serious advantage.
PS. Shifting gears makes a sound, but you really can't hear it when the car is moving. Imho the sound is about right currently in LFS.
DX9 or DX10 graphics? what about sampled sound?
(52 posts, started )
FGED GREDG RDFGDR GSFDG