Predictions are predictions. Observing a deviation and finding a historic event to match is not a prediction.
A prediction is saying something about something which hasn't happen yet. Saying "stuff will happen" is not a prediction, because there will always be some event you can associate with what you've previously observed. In order for this machine to predict something you need more information than just "Stuff will happen".
Actually (and this may seem counterintuitive, but if you think it through a bit, it makes sense), animation is hit much harder by compression/encoding effects than `live actors and stuff'. Effectively, there's less `visual noise' in animation for the effects to `hide behind'.
Animation also has lots of high frequency parts, with abrupt color changes and the large monocolored areas confuse the motion estimators.
The most blatant quantifiable lie is the 75 million polygon-remark. This is an highly idealized number of flat-shaded non-lighted polygons which the PS2 could possibly push if it had the cpu-power to actually supply enough meaningfull vertex data, or the memory to hold it.
My favorite part however is: And while today's games possess limited intelligence, future games will feature characters having not only brains, but simulated senses like sight and hearing, leading to an unprecedented level of realism.
For example, the system could run interactive movies that blend Hollywood's expensive production systems with intricate story lines and characters who interact realistically with the user.
Besides, that adds up to 40fps, which exceeds TV's ~30fps.
TVs are interlaced and actually refresh 60-half pictures a second. LCDs boasting 12ms refresh lie, because they use two refreshes to reach the pixels intended color.
Veering off-topic, but I've always wondered why some area of the software, be it OpenGL or the game itself, doesn't perfom some kind of dithering when converting the 24-bit image data down to 16-bit.
What makes you think they dont? (They do), but it's a double edged sword. When you do multiple passes, the ordered dithering reinforces itself and becomes ordered noise instead. Dithering was also an optional thing, and not every game turned it on.
The "Belgian Blue" breed of cows probably give some indication on what kind of problems this kid will have.
The double musculature of the cow doesn't add to the net mass, it simply occupies a larger portion. Belgian Blue bulls may look huge, but are in actually not larger than bulls of other breeds.
Bigger muscles means everything else is smaller, most notably the internal organs. Heart, lungs, etc. Smaller organs is pretty much the opposite you need when you have larger muscles, so this might be a big problem.
I've been unable to dig up the reference, but I read in an article about Pixar's "Monster's Inc." that for some frames it took longer to load the geometry than actually rendering the frame.
SETI and Folding@Home work because of the massive asymmetry between the amount of data and the CPU power required, and although you _perhaps_ could find subtasks that could easilly be "offsourced" so to speak, that made sense performance wise, I very much doubt that it would interface very nicely with the way the artists work, or make any sort of economic sense.
A smaller sensor is more noisy and more prone to chromatic abberation.
Short explanation: Chromatic abberation is a result of the fact that different wavelengths are refracted differently. They come out of the lens at slightly different angles.
The further away the sensor is from the lens, the more distance the light will have separated. The amount of separation is also a function of the curvature of the lens, i.e. how much it refracts.
The size of the CCD matters only indirectly, and the relation is not as you've stated. Besides, chromatic abberation can be pretty effectively counteracted through software software.
The amount of light hitting the sensor is limited by one thing and one thing only. The size of the lens, which is also what's limits resolution. This is basic physics, people.
Megapixels matter but only if the quality and size of the lens isn't limiting. Camera producers aren't idiots, so they haven't been placing oversized sensors in compacts that doesn't have the optics to provide a clear image, but as CCDs get cheaper and cheaper, a lot of them will, because people use that number to consider which camera to buy.
As an example, look at Canon S45 and Canon S50. It's basically the same camera, but the S45 has a 4MP CCD while the S50 has a 5MP CCD. For wide shots, this gives the S50 an edge, but when using the 3x optical zoom both cameras are equipped with, both are severely limited by the optics so you will not see any gain in resolution.
So, if there is no "energy in magnets" why does it take a huge amount of energy to seperate them?
It's kinda like gravity, only much stronger. It takes energy to lift a rock off the ground, and it hurts to drop the rock on your foot, but you'd find it difficult to pull a lot of energy out of this system.
You could create energy by letting magnets pull themselves together, but then that energy would be spent, and you'd have to use the same amount of energy (if the system was 100% effective) to pull the magnets apart.
To continue the gravity-analogy: Consider a powerplant creating energy by lowering rocks into a well. The rock is tied to a rope connected to a generator. As the rock decends you get a bit of energy, but when the rock hits the bottom, the fun is over. Now you have to get the rock out of there or lower down a new one, but eventually the well would fill up, and you'd have to dig a new one, and you can be certain that will take a lot more energy than you got out of putting them in there in the first place.
Actually, J2SE 1.5 already fixes a lot of these things in Java.
Yeah. And don't you think JWZs rant is fairly dated now? A lot of his points are no longer valid, and more than a couple of them only reveal that he's still trying to code C in Java.
Yeah. The 75GXP-thing happened to me too, but at least it gave me a good year before it started to fail. I was lucky though, and no data was lost. I'll probably consider Hitatchi, but I won't be first in line like I was with the 75GXP.
I am a bit surprised that Hitachi is still sticking with the Deskstar-name.
Companies get ragged on for 'ripping off the consumer' over print head costs. But you can see it as a choice too.
It would be easier to see this as a choice if not for the fact that the printer companies are making a huge bundle selling inks. Far more than they are not making selling cheap printers, and the different tactics they use to prevent third-party ink producers are definately not in the customers best interest.
Likewise, ink-economical-printers are overprised, both because fewer of them are sold (since joe sixpack wants cheap printers), and because they factor in not only the increased cost of production, but the money they LOSE not selling expensive inks.
There is a reason why they call inks the black gold of the information age.
What we need is a new printer company with a fresh perspective, or that one of the old ones decides to screw all the others, but right now, nobody wants to, because they're all sitting pretty. It's a Nash equilibrium. What's needed is a new player.
I suspect in this picture, there was large pieces of the sky missing, that they'd decided not to capture, to save time, so the sky was just filled in later.
Especially if I could get commercials tailored to my demographic, and watch them before the show, instead of in the middle of them, if this means I get to watch episodes of my favorite shows "for free".
The problem with advertise-driven TV is that the contract between the advertiser/broadcaster/viewer is fuzzy. There is no delivery system which can target a specific viewer.
I suspect that in the future, this contract may be explicitly defined, and applications like TiVo will help with the implementation. I'm willing to make some information about myself available to advertizers for a specific amount of time, if feel like it's a good trade. And it is a good trade if I don't have to watch diaper commercials, and get to see my favorite shows without interruption, even if I need to press and hold a "dead man's switch" during the commercials.
It is not a good trade when I'm constantly interrupted by waves of ads because they need to fling a lot at the wall to make something stick.
It might be the perfect game if you have to pick ONE game to give to a million families. It is not the perfect game for a specific familiy, just the perfect game when the familiy is not known.
By the same logic, you can find out that the perfect food is a Big mac, since nobody really hates it (You can't hate something which tastes nothing).
Whenever you create something with the ultimate all-encompassing demographic, you end up with something which is infinitely bland and infinitely inoffensive.
In beauty contests, you typically have several rounds with different jurys, a mechanism which is sure to filter out someones ideals and move towards the average, which is why you'll find that Miss Universe can be less attractive than the girl next door.
Of course, there are objective parameters you can measure, but if you get all or most of them right, you just end up with something that doesn't totally suck. To create something brilliant you have to narrow your appeal, to match the individual preferences of a spesific group.
Oh really? If as much resources were put into creating a poker-playing machine as were put into deep blue, I think you'd be in for a surprise. Of cource, this machine would have sensors to detect everything from pupil dialation and voice patterns to body temperature and perspiration.
Poker is almost the opposite to chess as the bare game mecanics is mostly chance and all that is required is a bit of statistics. The problem of poker is mainly about reading your opponents. In chess, the computer has the added benifit of being unaffected by Kasparovs frightening scowl.
But this has nothing to do with AI, you may think. A machine just computes. It measures the world, compares to a database of experiences, either learned or installed by it's creators weighs pros and cons and selects the path which leads to the desired outcome.
But how is this different from how humans "work"?. I suspect, not much at all. It's simply that we lack the degree of introspection with the human brain, that we have with computers. We cannot antropomorphise something which we know the nuts and bolts of. This will most likely change as we get more and more neural nets into normal software and programs evolved though genetic algorithms rather than being punched in by code monkeys like myself.
As this is a hobby project, I won't make any promises time-wise, but when I've filed down most of the sharp corners of my app, the knobs will indeed be made available. Not the source, at least not while the program is still in beta, but I'll most likely include som instructions on how to implement new functions.
Because of the special focus of my program, it will not please everyone. For instance, stereo or multi-channel sound is out of the scope of the program, since positioning in space is strictly in the domain of the game engine.
To use a graphics programming analogy, the program can be used to create textures, but not complete images.
By the way, noise^3 makes an excellent rain-sound as you said, and by tweaking the frequencies I could change it from rain hitting asphalt to rain on a large umbrella or a marquee.
noise^5 or higher makes a pretty good pickup-noise.:)
Actually, I'm currently in the process of writing a Java application for creating sound effects from scratch, using basic mathematical functions as a base. It is a fork of my original project, which is a game, and I wanted original and free sound effects for it.
The page shows a screenshot of the gui showing an attempt to recreate the waterfall described in the parent comment, and have for your enjoyment(?) included a wav+mp3 of what the output sounds like. Note that some of the parameters are not visible in the shot.
Slashdot Mirror
Predictions are predictions. Observing a deviation and finding a historic event to match is not a prediction.
A prediction is saying something about something which hasn't happen yet. Saying "stuff will happen" is not a prediction, because there will always be some event you can associate with what you've previously observed. In order for this machine to predict something you need more information than just "Stuff will happen".
Actually (and this may seem counterintuitive, but if you think it through a bit, it makes sense), animation is hit much harder by compression/encoding effects than `live actors and stuff'. Effectively, there's less `visual noise' in animation for the effects to `hide behind'.
Animation also has lots of high frequency parts, with abrupt color changes and the large monocolored areas confuse the motion estimators.
I dug up an old article on the PS2.
The most blatant quantifiable lie is the 75 million polygon-remark. This is an highly idealized number of flat-shaded non-lighted polygons which the PS2 could possibly push if it had the cpu-power to actually supply enough meaningfull vertex data, or the memory to hold it.
My favorite part however is:
And while today's games possess limited intelligence, future games will feature characters having not only brains, but simulated senses like sight and hearing, leading to an unprecedented level of realism.
For example, the system could run interactive movies that blend Hollywood's expensive production systems with intricate story lines and characters who interact realistically with the user.
Ooooh. I gotta get me some of that!
Besides, that adds up to 40fps, which exceeds TV's ~30fps.
TVs are interlaced and actually refresh 60-half pictures a second.
LCDs boasting 12ms refresh lie, because they use two refreshes to reach the pixels intended color.
Excuse me? eDonkey/mule has a 4:1 download/upload ratio, and as long as upload is over 10KB/s the download isn't capped at all.
Veering off-topic, but I've always wondered why some area of the software, be it OpenGL or the game itself, doesn't perfom some kind of dithering when converting the 24-bit image data down to 16-bit.
What makes you think they dont? (They do), but it's a double edged sword. When you do multiple passes, the ordered dithering reinforces itself and becomes ordered noise instead. Dithering was also an optional thing, and not every game turned it on.
The "Belgian Blue" breed of cows probably give some indication on what kind of problems this kid will have.
The double musculature of the cow doesn't add to the net mass, it simply occupies a larger portion. Belgian Blue bulls may look huge, but are in actually not larger than bulls of other breeds.
Bigger muscles means everything else is smaller, most notably the internal organs. Heart, lungs, etc. Smaller organs is pretty much the opposite you need when you have larger muscles, so this might be a big problem.
I've been unable to dig up the reference, but I read in an article about Pixar's "Monster's Inc." that for some frames it took longer to load the geometry than actually rendering the frame.
SETI and Folding@Home work because of the massive asymmetry between the amount of data and the CPU power required, and although you _perhaps_ could find subtasks that could easilly be "offsourced" so to speak, that made sense performance wise, I very much doubt that it would interface very nicely with the way the artists work, or make any sort of economic sense.
A smaller sensor is more noisy and more prone to chromatic abberation.
Short explanation: Chromatic abberation is a result of the fact that different wavelengths are refracted differently. They come out of the lens at slightly different angles.
The further away the sensor is from the lens, the more distance the light will have separated. The amount of separation is also a function of the curvature of the lens, i.e. how much it refracts.
The size of the CCD matters only indirectly, and the relation is not as you've stated. Besides, chromatic abberation can be pretty effectively counteracted through software software.
Sweet Jebus!
The amount of light hitting the sensor is limited by one thing and one thing only. The size of the lens, which is also what's limits resolution. This is basic physics, people.
Megapixels matter but only if the quality and size of the lens isn't limiting. Camera producers aren't idiots, so they haven't been placing oversized sensors in compacts that doesn't have the optics to provide a clear image, but as CCDs get cheaper and cheaper, a lot of them will, because people use that number to consider which camera to buy.
As an example, look at Canon S45 and Canon S50. It's basically the same camera, but the S45 has a 4MP CCD while the S50 has a 5MP CCD. For wide shots, this gives the S50 an edge, but when using the 3x optical zoom both cameras are equipped with, both are severely limited by the optics so you will not see any gain in resolution.
So, if there is no "energy in magnets" why does it take a huge amount of energy to seperate them?
It's kinda like gravity, only much stronger. It takes energy to lift a rock off the ground, and it hurts to drop the rock on your foot, but you'd find it difficult to pull a lot of energy out of this system.
You could create energy by letting magnets pull themselves together, but then that energy would be spent, and you'd have to use the same amount of energy (if the system was 100% effective) to pull the magnets apart.
To continue the gravity-analogy: Consider a powerplant creating energy by lowering rocks into a well. The rock is tied to a rope connected to a generator. As the rock decends you get a bit of energy, but when the rock hits the bottom, the fun is over. Now you have to get the rock out of there or lower down a new one, but eventually the well would fill up, and you'd have to dig a new one, and you can be certain that will take a lot more energy than you got out of putting them in there in the first place.
Being american makes you a twunt.
Way to assume, nobrains.
I knew it,
Soccer rots your brain.
Actually, J2SE 1.5 already fixes a lot of these things in Java.
Yeah. And don't you think JWZs rant is fairly dated now? A lot of his points are no longer valid, and more than a couple of them only reveal that he's still trying to code C in Java.
Yeah. The 75GXP-thing happened to me too, but at least it gave me a good year before it started to fail. I was lucky though, and no data was lost. I'll probably consider Hitatchi, but I won't be first in line like I was with the 75GXP.
I am a bit surprised that Hitachi is still sticking with the Deskstar-name.
Like Robocop I and II did.
Companies get ragged on for 'ripping off the consumer' over print head costs. But you can see it as a choice too.
It would be easier to see this as a choice if not for the fact that the printer companies are making a huge bundle selling inks. Far more than they are not making selling cheap printers, and the different tactics they use to prevent third-party ink producers are definately not in the customers best interest.
Likewise, ink-economical-printers are overprised, both because fewer of them are sold (since joe sixpack wants cheap printers), and because they factor in not only the increased cost of production, but the money they LOSE not selling expensive inks.
There is a reason why they call inks the black gold of the information age.
What we need is a new printer company with a fresh perspective, or that one of the old ones decides to screw all the others, but right now, nobody wants to, because they're all sitting pretty. It's a Nash equilibrium. What's needed is a new player.
I suspect in this picture, there was large pieces of the sky missing, that they'd decided not to capture, to save time, so the sky was just filled in later.
No, but you've confused wavelength with frequency.
Especially if I could get commercials tailored to my demographic, and watch them before the show, instead of in the middle of them, if this means I get to watch episodes of my favorite shows "for free".
The problem with advertise-driven TV is that the contract between the advertiser/broadcaster/viewer is fuzzy. There is no delivery system which can target a specific viewer.
I suspect that in the future, this contract may be explicitly defined, and applications like TiVo will help with the implementation. I'm willing to make some information about myself available to advertizers for a specific amount of time, if feel like it's a good trade. And it is a good trade if I don't have to watch diaper commercials, and get to see my favorite shows without interruption, even if I need to press and hold a "dead man's switch" during the commercials.
It is not a good trade when I'm constantly interrupted by waves of ads because they need to fling a lot at the wall to make something stick.
It might be the perfect game if you have to pick ONE game to give to a million families. It is not the perfect game for a specific familiy, just the perfect game when the familiy is not known.
By the same logic, you can find out that the perfect food is a Big mac, since nobody really hates it (You can't hate something which tastes nothing).
Whenever you create something with the ultimate all-encompassing demographic, you end up with something which is infinitely bland and infinitely inoffensive.
In beauty contests, you typically have several rounds with different jurys, a mechanism which is sure to filter out someones ideals and move towards the average, which is why you'll find that Miss Universe can be less attractive than the girl next door.
Of course, there are objective parameters you can measure, but if you get all or most of them right, you just end up with something that doesn't totally suck. To create something brilliant you have to narrow your appeal, to match the individual preferences of a spesific group.
Poker comes to mind immediately.
Oh really? If as much resources were put into creating a poker-playing machine as were put into deep blue, I think you'd be in for a surprise. Of cource, this machine would have sensors to detect everything from pupil dialation and voice patterns to body temperature and perspiration.
Poker is almost the opposite to chess as the bare game mecanics is mostly chance and all that is required is a bit of statistics. The problem of poker is mainly about reading your opponents. In chess, the computer has the added benifit of being unaffected by Kasparovs frightening scowl.
But this has nothing to do with AI, you may think. A machine just computes. It measures the world, compares to a database of experiences, either learned or installed by it's creators weighs pros and cons and selects the path which leads to the desired outcome.
But how is this different from how humans "work"?. I suspect, not much at all. It's simply that we lack the degree of introspection with the human brain, that we have with computers. We cannot antropomorphise something which we know the nuts and bolts of. This will most likely change as we get more and more neural nets into normal software and programs evolved though genetic algorithms rather than being punched in by code monkeys like myself.
Blatand was loosely translated into Bluetooth
What do you mean loosely? Blatand literally means "blue tooth".
I mean "a".. no i mean "" argh.
Can we get ISO8859-1 in slashdot please?
As this is a hobby project, I won't make any promises time-wise, but when I've filed down most of the sharp corners of my app, the knobs will indeed be made available. Not the source, at least not while the program is still in beta, but I'll most likely include som instructions on how to implement new functions.
:)
Because of the special focus of my program, it will not please everyone. For instance, stereo or multi-channel sound is out of the scope of the program, since positioning in space is strictly in the domain of the game engine.
To use a graphics programming analogy, the program can be used to create textures, but not complete images.
By the way, noise^3 makes an excellent rain-sound as you said, and by tweaking the frequencies I could change it from rain hitting asphalt to rain on a large umbrella or a marquee.
noise^5 or higher makes a pretty good pickup-noise.
Actually, I'm currently in the process of writing a Java application for creating sound effects from scratch, using basic mathematical functions as a base. It is a fork of my original project, which is a game, and I wanted original and free sound effects for it.
The page shows a screenshot of the gui showing an attempt to recreate the waterfall described in the parent comment, and have for your enjoyment(?) included a wav+mp3 of what the output sounds
like. Note that some of the parameters are not
visible in the shot.