Since causation cannot travel faster than light it could have happened 300 years ago 500 lightyears away and we wouldn't notice until 200 years in the future.
It has to be possible to implement from the patent specifications without any further inventing. Sticking a jet engine and wings to a car could mean duct taping them on. Any undefined parts used in the invention have to be invented already (e.g. you can say transistor but not anti-gravity generator). There was a time when patents required a reference implementation (e.g. to patent a car motor you'd have to build one and demonstrate it to the patent office), I think a patent should be clear enough to allow anyone with experience in the technical field to produce such a sample. If the patent has flaws that make an actual implementation without changes impossible the patent should be invalid.
Idea: "We should build a flying car!" Implementation: "We will build a flying car by doing this:____________"
An idea only describes the desired outcome, an implementation every single step to that outcome (which allows for others to arrive at the same outcome through different steps). The implementation has to be very specific so it only covers exactly the way you thought up and it has to be feasible, i.e. you can use it to archieve said outcome. You cannot patent perpetuum mobiles because it's physically impossible to create such a device and unless your desired outcome was a thing that'd look nice on a coffee table (even then it probably falls under copyright) you have no way of using your invention to do what you want it to do.
Patent durations have a downward trend unlike copyright because even the biggest corps know that while patents are a great weapon, he who lives by the patent will die by the patent. Patents are nice to wipe out upstart competition but patent hoarding companies are a severe threat to even the biggest megacorps, especially since they tend to be the first to get some overbroad patent accepted and wait until their patented stuff has become the industry standard before they start to issue lawsuits.
Next you'll tell me that line art takes as much space as a photo.
Unless you can find an algorithm to compress a photo into line art and turn it back that'll be a completely useless comparison since 99% of your source data is photos or comparable images. Some textures in games (e.g. normalmaps) cannot be compressed at all without creating severe artifacting.
Using JPEG as an algorithm will give nasty artifacts when the player comes close and increases the necessary resolution for textures to avoid visible artifacts. That means more work.
Then it's done badly, good procedural textures are far from random and often better then an avarage artist could do by hand in a reasonable time frame. The reason why procedural textures aren't used has nothing to do with their quality, it's because when you have big bucks and little brain it's easier to throw artists and storage at the problem.
Those seen in games aren't procedural, they're images that are tiled over the surface. You'll have to go with that route unless you want to add a pixel shader that calculates the details on the fly (because otherwise you have to store an image of it in memory and no algorithm can create better source data than a good artist yet). Since there are plenty of shaders in any given scene already that'd cause more trouble than it's worth.
Don't think people avoid procedurals because they're strange, in fact procedural generation is the holy grail of graphics development but there just aren't any procedurals that would be usable in a game and produce realistic results (hell, even prerendered CG usually uses hand painted everythings because the procedurals even the high end 3d programs offer are worthless if you want to make anything that looks decent).
Of course. Noone can tell you your textures aren't allowed to blur. It doesn't look as good as non-blurring textures of course but if the hardware can't do it any better without making even bigger sacrifices that's the only way to do it.
There are no lightmaps in Doom 3. That's also the reason it's so dark, only direct light is calculated, there is no light bouncing, which accounts for most of the light we see in real life. Radiosity is a BIG factor in the realism of a scene.
1. Go multi-disc. Not really an option when streaming but maybe it'd work with something like SSX3. 2. Use procedurals, utilize only a fraction of the machine's power for gameplay while the rest is spent on generating the graphics data. 3. Just downsample the textures, keep disc costs and development effort low and framerate high.
It looks random because it doesn't seem to reflect any surface properties. If there's a larger rock painted on the cave wall the detail texture will still treat it like normal dirt, it won't even flow along the painted surface of the object. Scratches in the concrete will have the same texture as well. Plus there's the whole deal about which parts of a surface would be most exposed to withering. If you can spend enough development ressources on creating a logic that would add those details you'd sit there five minutes during load while the game calculates the withering on various surfaces. Hell, we could let it render the lightmaps at load time! Just wait an hour until the radiosity simulation is complete!
I think procedural objects are more useful for meshes, e.g. trees (Speed Tree) and other scenery parts that appear in large numbers and should look different each. That's more doable because meshes consist of a bunch of vectors the engine could freely modify. Textures aren't a good target.
The problem is that even a rough texture takes as much space as a full one. Adding some random stuff to keep them from blurring is nothing new, you can see that being done in games like Serious Sam. It's not on par with handpainted detail because it looks very random.
The artists draw those anyway, textures are usually downsampled at the end when it turns out they don't fit. And procedural synthesis doesn't give you bark if you tell it "give me bark", it requires workers to feed it the proper base data and if it's as versatile as painting textures directly it'll require similar work as well.
Steam is one of the worse implementations I've come across. Ressource hogging, requiring an internet connection and failing if it thinks you're connected but you are not. Earth 2160, for example, allows you to activate via the phone and then the game just runs, no more needing to start some extra client, logging into an account or connecting to the internet (not even needing a CD). Sure, it's easier to crack than Steam but nothing is uncrackable and hell, I, as the customer, don't care how many warez copies it prevents, if it inconveniences me I don't like it.
You're ignoring that a lot of stuff isn't level dependent, e.g. the player character and most of his equipment, repeating enemies, textures that were in another level as well, etc. Plus there's the gamestate that takes up RAM and is not part of the disc content.
Vector graphics are useless for realistic stuff. They're nice for simple structures and would maybe work for cartoon textures but that's where it ends.
1. Kkrieger took longer to develop than most commercial titles (though admittedly they used a smaller team) and the procedural system is VERY limited. 2. Many games use streaming to avoid loading screens, if you do that you'll need that CPU power for the game (unless you're not using the system's full power normally). 3. Kkrieger takes longer to load than e.g. Earth 2160 on my system.
PC games are for some reason MUCH smaller than console titles. Some PS2 games span 2 DVDs yet they don't have nearly the texture sizes PC games have while among the PC titles only FPSes and console ports seem to exceed 2GB.
Slashdot Mirror
Since causation cannot travel faster than light it could have happened 300 years ago 500 lightyears away and we wouldn't notice until 200 years in the future.
It has to be possible to implement from the patent specifications without any further inventing. Sticking a jet engine and wings to a car could mean duct taping them on. Any undefined parts used in the invention have to be invented already (e.g. you can say transistor but not anti-gravity generator). There was a time when patents required a reference implementation (e.g. to patent a car motor you'd have to build one and demonstrate it to the patent office), I think a patent should be clear enough to allow anyone with experience in the technical field to produce such a sample. If the patent has flaws that make an actual implementation without changes impossible the patent should be invalid.
Trivial patents should not be awarded period. That's one of the requirements for a patent to be valid, that it's not obvious.
Your sig goes along perfectly with that post...
Idea: "We should build a flying car!"
Implementation: "We will build a flying car by doing this:____________"
An idea only describes the desired outcome, an implementation every single step to that outcome (which allows for others to arrive at the same outcome through different steps). The implementation has to be very specific so it only covers exactly the way you thought up and it has to be feasible, i.e. you can use it to archieve said outcome. You cannot patent perpetuum mobiles because it's physically impossible to create such a device and unless your desired outcome was a thing that'd look nice on a coffee table (even then it probably falls under copyright) you have no way of using your invention to do what you want it to do.
Perhaps the OP is thinking of countries where patents don't exist?
Sony Playstation?
Microsoft XBox?
Nintendo Gamecube?
Patent durations have a downward trend unlike copyright because even the biggest corps know that while patents are a great weapon, he who lives by the patent will die by the patent. Patents are nice to wipe out upstart competition but patent hoarding companies are a severe threat to even the biggest megacorps, especially since they tend to be the first to get some overbroad patent accepted and wait until their patented stuff has become the industry standard before they start to issue lawsuits.
They keep coming and coming and even if you defeat them they come again... Sounds like Zombies. Martha, bring me the shotgun.
Hm, just a thought: If the EU were to allow software patents, could an EU court invalidate a US patent?
Well, it might have been invented but the effect is still travelling towards us...
Next you'll tell me that line art takes as much space as a photo.
Unless you can find an algorithm to compress a photo into line art and turn it back that'll be a completely useless comparison since 99% of your source data is photos or comparable images. Some textures in games (e.g. normalmaps) cannot be compressed at all without creating severe artifacting.
Using JPEG as an algorithm will give nasty artifacts when the player comes close and increases the necessary resolution for textures to avoid visible artifacts. That means more work.
Then it's done badly, good procedural textures are far from random and often better then an avarage artist could do by hand in a reasonable time frame. The reason why procedural textures aren't used has nothing to do with their quality, it's because when you have big bucks and little brain it's easier to throw artists and storage at the problem.
Those seen in games aren't procedural, they're images that are tiled over the surface. You'll have to go with that route unless you want to add a pixel shader that calculates the details on the fly (because otherwise you have to store an image of it in memory and no algorithm can create better source data than a good artist yet). Since there are plenty of shaders in any given scene already that'd cause more trouble than it's worth.
Don't think people avoid procedurals because they're strange, in fact procedural generation is the holy grail of graphics development but there just aren't any procedurals that would be usable in a game and produce realistic results (hell, even prerendered CG usually uses hand painted everythings because the procedurals even the high end 3d programs offer are worthless if you want to make anything that looks decent).
No, you cannot patent discoveries, you can only patent some useful application you derived from them.
Of course. Noone can tell you your textures aren't allowed to blur. It doesn't look as good as non-blurring textures of course but if the hardware can't do it any better without making even bigger sacrifices that's the only way to do it.
There are no lightmaps in Doom 3. That's also the reason it's so dark, only direct light is calculated, there is no light bouncing, which accounts for most of the light we see in real life. Radiosity is a BIG factor in the realism of a scene.
Well, we have three options here:
1. Go multi-disc. Not really an option when streaming but maybe it'd work with something like SSX3.
2. Use procedurals, utilize only a fraction of the machine's power for gameplay while the rest is spent on generating the graphics data.
3. Just downsample the textures, keep disc costs and development effort low and framerate high.
It looks random because it doesn't seem to reflect any surface properties. If there's a larger rock painted on the cave wall the detail texture will still treat it like normal dirt, it won't even flow along the painted surface of the object. Scratches in the concrete will have the same texture as well. Plus there's the whole deal about which parts of a surface would be most exposed to withering. If you can spend enough development ressources on creating a logic that would add those details you'd sit there five minutes during load while the game calculates the withering on various surfaces. Hell, we could let it render the lightmaps at load time! Just wait an hour until the radiosity simulation is complete!
I think procedural objects are more useful for meshes, e.g. trees (Speed Tree) and other scenery parts that appear in large numbers and should look different each. That's more doable because meshes consist of a bunch of vectors the engine could freely modify. Textures aren't a good target.
The problem is that even a rough texture takes as much space as a full one. Adding some random stuff to keep them from blurring is nothing new, you can see that being done in games like Serious Sam. It's not on par with handpainted detail because it looks very random.
The artists draw those anyway, textures are usually downsampled at the end when it turns out they don't fit. And procedural synthesis doesn't give you bark if you tell it "give me bark", it requires workers to feed it the proper base data and if it's as versatile as painting textures directly it'll require similar work as well.
Steam is one of the worse implementations I've come across. Ressource hogging, requiring an internet connection and failing if it thinks you're connected but you are not. Earth 2160, for example, allows you to activate via the phone and then the game just runs, no more needing to start some extra client, logging into an account or connecting to the internet (not even needing a CD). Sure, it's easier to crack than Steam but nothing is uncrackable and hell, I, as the customer, don't care how many warez copies it prevents, if it inconveniences me I don't like it.
You're ignoring that a lot of stuff isn't level dependent, e.g. the player character and most of his equipment, repeating enemies, textures that were in another level as well, etc. Plus there's the gamestate that takes up RAM and is not part of the disc content.
Vector graphics are useless for realistic stuff. They're nice for simple structures and would maybe work for cartoon textures but that's where it ends.
1. Kkrieger took longer to develop than most commercial titles (though admittedly they used a smaller team) and the procedural system is VERY limited.
2. Many games use streaming to avoid loading screens, if you do that you'll need that CPU power for the game (unless you're not using the system's full power normally).
3. Kkrieger takes longer to load than e.g. Earth 2160 on my system.
PC games are for some reason MUCH smaller than console titles. Some PS2 games span 2 DVDs yet they don't have nearly the texture sizes PC games have while among the PC titles only FPSes and console ports seem to exceed 2GB.
The code is compatible but I think the OP wanted to suggest that the Rev's specs are identical to the GC's.