Yeah, I agree - as far as I know Blizzard isn't using spyware to create targeted advertising aside from "you're advertising to gamers." In addition, you don't have to even play on Battle.NET to enjoy multiplayer, though if you want to avoid cheaters it helps. Blizzard pays their bills for the service thru advertising, so I understand the need to have ads there. I've heard it's profitable, too.
The main thing that bugs me, though, is that I would think Battle.NET is MUCH more expensive to run since games are hosted on Battle.NET servers. As far as I can tell, EA is providing nothing more than a matching service for multiplayer so I would expect they have trivial expenses in comparison.
I can skip BF2142 easily enough, but I hope this isn't a trend. If Midway ships UT2007 that way, I'll be much more disappointed, and if they add unlocks too (like BF), well, ew. I hate locks with a passion on that type of game - it's bad enough that skilled players already can beat the crap out of unskilled players, but then they tack on better equipment for those skilled players and give them an unfair edge, as well... yeah, well I'm not a fan. I might be ok with it if the game at least handicapped by skill (balancing the unskilled players with more numbers). In the real world, that would be like expecting a SEAL to be able to take on a dozen poorly trained militia alone - he's got better equipment and training, but a lucky shot can still take him out.
what frightens me is that I got an unlockable eVGA AGP 7800GS ~6 months ago for $4 cheaper (~$50 if you only count 4 * or better vendors like I ordered thru) than the cheapest price on pricewatch. It was $198 with express 2 day shipping (not sure if it was cheaper with slower shipping). It also proved unlockable with CoolBits, so I essentially got a 7800GT for the price of a 7800GS.
sigh - I need to upgrade soon - I'm jealous of my dual 3GHz's at work (with an extremely crappy GPU that actually does affect my work since my company is now adding GPU features into our products... sigh - if you spend $3000 on a PC, you can put in at LEAST a $50 GPU... this one listed at $27 last year when I got the machine).
heh - why not? I'll be there in a couple of days, so I can lend 'em a hand - as long as I've got a maß (beer glass) in the other. I'm college educated, so I can even drink and program at the same time (expect functional code, not legible code;) ).
p.s. Please pay me in Euro (and yes, I will even pronounce it "oi row" just for the brownie points) - the dollar is practically toilet paper these days.
If you're a gamer, because Microsoft is pushing Vista Only (DirectX 10, exclusive to Vista by cabal) games like age of conan and Halo 2. I'm sure there will be many more that follow, and pretty soon a good chunk of the Windows gaming market will be forced into Vista through Microsoft's strong-arm tactics, whether they feel like they need the upgrade or not.
For that reason alone I will probably build a Vista/Linux dual-boot machine sometime next year. I certainly will continue to avoid WiMP for ripping, and I already planned to use my "old" PC's linux partition to become a "Media center PC" (using MythTV) if TiVO manages to kill my current Dish PVR. That 3 day thing would never work for me, as I often get weeks or months behind during work crunches. Micro$oft of all people should relate to that one.
A state based injunction like that isn't even valid across state lines unless the business also operates there. If I were running my business in, say, Washington and had operations in Oregon, he would have to file suit in Washington, Oregon or in federal court for it to be valid. Filing a local case like that is a waste of time.
In the late '40s through late '60s the US military tested chemical weapons on its own people without their knowledge. I read about clean up efforts from some of these secret tests just a few years ago in a local rag (that's newspaper, if you don't know the idiom), so I dug around google for at least one reference.
The last time I know of that the US ASKED for volunteers, they were doing nuclear testing with citizens well within the danger zone to see the effects of nuclear weapons on people just outside the immediate blast radius (probably because PETA was protesting the use of animals - jk).
yeah, well he's right - Bush has proven without a reasonable doubt that the evidence doesn't show whether there is or is not global warming at this time and until we can prove it undoubtably, we should not put any unnecessary restrictions on campaign funde... er, I mean big businesses.
For that matter, Al Gore is an idiot*, so don't believe everything he says - I'm pretty sure he's secretly in league with law breaking PETA militants. George Bush's far superior IQ** always prevails.
* compared to a genius monkey ** compared to a dead monkey
p.s. this was intended to be funny. If you didn't find it mildly amusing, I'm sorry, and please don't grenade my house.
Ah, but in fact they bought the second publisher of D&D, BUT the first publisher of AD&D. Since the original version of D&D was discontinued and AD&D then became known as D&D, technically TSR, Inc. was the first publisher of D&D! If you were a true Sicilian, you would know that! Ah ha ha ha. Ah ha ha ha... klunk (sorry, that bit of logic was hitting Vizzini levels;).
And to be nitpicky, it was Gygax and (Dave) Arneson that invented D&D, not Gygax and Kaye, who were the publication financers. I don't believe Kaye had much input, if any, to the material.
The idea is to have a database of signal patterns and compare this to an incoming signal, so it means that one time in 20 the signal was not identified as one of the stored signals. Since the method never gave a false identification (positive) as one of the stored signal patterns, your first assertion is true - 1 in 20 valid attempts is refused.
Ah - I see what you are getting at - you are saying that a ray tracer can find each pixel on the screen by the sort alone. I was giving worst case scenario estimates (big-Oh), which favor the rasterizer with z-buffering because it doesn't require a sort, but average case scenario estimates may well indeed favor ray tracing. Going back to the original topic, it seems to me ray tracing had a better big Oh than the old raster standard the painter's algorithm (like nlogn vs n^2) and had better average case analysis as n approached infinity. In school zbuffering was really a future thing, and I don't recall doing any analysis on it (I think it was even new in the book we used - Computer Graphics: Principals and Practices [2nd edition?] that year). For that matter, I was still writing code in PHIGS and GL, at that time, not even OpenGL, though I picked up some OGL by my second graphics class the next spring.
There are occlusion sort algorithms on the polygon side that may give it a good run for its money, like cPLP (since standard PLP is lossy) or hierarchal z-buffers, though there isn't really a "one shoe fits all feet" algorithm like you're talking about.
I've now read that even polygon data can be sorted into KD-trees, so it's not exclusive to ray tracing (as I stated earlier, it's been a long time since I've used them if at all...), so I don't think your statement is correct. I also don't think my first ray tracer had any sorting at all, though obviously sorting has some speed benefits.
The LOD culling that is most common in raster games lately is probably a chunked level of detail algorithm such as GeoMipMapping (sometimes combined with GeoMorphing for smoothness), which has a horrible problem with popping because most programmers intentionally choose to set the error values to noticable levels rather than have framerates suffer. The old standard was ROAM, but that was CPU based and GeoMipMapping is optimized for use with a GPU, making it much more appealing on today's hardware. A ray tracer won't help with LOD unless you reduce the shape count (as opposed to polygon count). Your best hope is probably to have less shapes to form your model to begin with, because you have more powerful primitives (cube, sphere, etc) and you could also make much more efficient use of constructive solid geometry to make simpler objects without polygons (yes, you can do CSG with polygons, too, but most of the time that's used to create a polygon based subsurface). Still, if you have too many objects in the scene you either need to distance cull them or simplify them, so the solution is non-trivial.
Another, newer form of raster detail ugliness can be seen in games like Oblivion, which I believe uses a technique called distance based detail mapping (look at the ground and move - you should see a sphereical detail section around your avatar - it's most obvious on the paved roads in towns) and distance based object culling (like trees). I ask, though, how Ray Tracing would fix that? Anything I can think of will still cause popping or repetitive detail. Terrain simplification will probably be the same with ray-tracers as it is with polygons because height maps are so much more disk efficient than other methods and therefore will still have the same sort of culling issues. If anything, I would expect to see programmers adding splines for smoothness before moving to a more disk intensive model.
I still don't see, but perhaps its because a KD-tree isn't "grid-locked" like a quad or oct tree (I vaguely remember them from classes, but it's been a long time). The raster code I've been working on does trivial adds to the view volume of blocks of data, then it does trivial rejects to the actual polygons after that, and I believe that has a similar effect to that of a KD-tree. The scene is potentially infinite since it uses terrain paging (in reality, it's a 4GB data file), so trivial rejection of everything would potentially take infinite time. The potential terrain that is in view is found using a surface projected cone, since it's mathematically much simpler than a frustum at the expense of accuracy (we leave the accuracy to hardware). In practice, this has proven extremely fast.
In ASCII-art below, O is the observer (camera), and X and Y are potential view volume. * is cached data that is out of the view volume that is paged in background. O is added by default. Up to 3 additional "blocks" of code can be added to the data that needs to be parsed based on camera direction - the Ys show what should be the worst case scenario (viewer looking toward a corner with parts of two other areas in view).
***** *XYY* *XOY* *XXX* *****
Although the code I'm working on only does this in 2 directions, it can easily be expanded into 3 - you could even use the projected cone method and just project it onto another side. Border crossing polygons can be split, if necessary, but I've never had to do anything like that in practice (for static objects that overlap the terrain I just add vertices on the border in a pre-processing step). So far I haven't had to do any simplification of dynamic objects outside of distance based detail reduction (a sqrt on each actor), but that could become an issue later.
Anyhow, I'm not sure if that is the equivalent of what you were talking about (I think it is, since ray traced objects are more difficult to split, thus the KD-tree instead of oct-tree), but it is a way to reduce an infinite dataset to a manageable one using data structures.
I've written a ray tracer and I don't know what you're talking about. A ray tracer needs to test each ray against every object in the scene. You can use bounding boxes or spheres to trivially reject objects, but that's true of either model (or you can use pre-sort data structures such as quad-tree, oct-tree, BSP tree, etc). For specular objects the ray is also checked against every object in the scene again for what it reflects, which can increase time significantly; however, ray tracing is faster in the respect that it only checks one point per ray on all objects rather than all points on the rasterizer that contain the triangle, which should be faster in a large scene (but in practice takes little time since it's handled by graphics hardware - I haven't had to write a software blitter for that sort of thing in ages).
I'm pretty sure they both rasterization using a depth buffer and ray tracing would take linear time (O(n)), while the painter's algorithm was O(nlogn) because of the required logn sort.
I certainly didn't mean ray tracing was any better than rasterization on diffuse - they're both bad. I just wanted to point out that ray tracing has flaws and shouldn't be considered the be-all end-all solution for computer graphics - there are still issues that need to be resolved. I have mixed feelings about ambient occlusion as the diffuse technique as presented by the paper, but I have the same feelings about it being used in polygons (it's fast, but not realistic). I certainly don't mean to bad-mouth ray tracing, because I think it's a great technique and I even predicted that it would become the dominant technique starting around 2010 or 11 (prediction was from 1996 and based on Moore's law), because that's when I believed single processor machines would hit the 17GHz mark which I had calculated was required to maintain 640x480x30FPS. Yes, that prediction was a bit short-sighted, as resolution expectations, CPU technology improvements, GPUs (not that current GPUs would help much because they're not meant to handle scenes), and multiprocessing were not considered.
you are correct - pixel/fragment shading is not free - both use the same default shading model in OpenGL and probably DirectX.
the default shader is, I believe, Lambert (a close relative to Phong - if not, it's Phong) for OpenGL and probably DirectX as well. Programs in the shader can change this to whatever you want it to be (e.g. a cel shader) and you would need to do that in either a ray tracer or rasterizer.
there's a lot of things I like about ray tracing, but it's not without flaw - it handles specular highlights fantastically, but doesn't handle diffuse well at all, so you have to bolt on other techniques. Most people (including Intel) use ambient occlusion since it's a quick technique (also commonly used in polygon based graphics), but it tends to make muddy shadows (see the wikipedia entry). Radiosity is more realistic, but the patch computations are incredibly expensive (but parallel-able). photon mapping is another method that could be used, but I haven't used it myself. In college I wrote (with a team) a simple ray tracer and shortly after that class wrote a radiosity engine, so I'm familiar with both techniques. I never did really understand how to combine them, but I remember seeing POVRAY do it in the mid 90s and really wanted to figure out how they did it (but I graduated and was putting in startup hours, so that never happened).
Oh, and waves on a lake are non-trivial - to be completely realistic, you need to deal with subsurface diffusion (or estimation of), foam and caustics (if you can see through the semi-transparent water surface). The specular mirror effect would be nice, but I don't see true caustics from either a raytracer or a rasterizer (you'd need to use ray bars or cones, probably).
I learned the same thing as you, but I believe the geometric complexity towards infinity favoring raytracing was only true for painter's algorithm. Actually, I think the magic number was actually somewhere around 100000 polygons for painter's algorithm, but the occlusion characteristic of the z-buffer negated the advantage.
The main advantages of ray tracing is physically correct lighting, shadows, and reflections, as well as the ability to use any geometric shape that can be mathematically defined (e.g. a sphere). The downside of parallelizing this is each additional parallel ray comes with a concurrent increase in memory requirements. Ray tracing also tends to do specular well, but isn't as good at diffuse, so many ray tracing programs hybrid-ize with something like radiosity or ambient occlusion.
clipping is less of an issue now than it was pre z-buffer (or depth buffer, if you prefer) and artifacting from it should be no worse than ray tracing. The painter's algorithm had the three triangle problem (same as the three polygons on wikipedia) that caused odd artifacting, but I don't see how zbuffer is any worse than ray tracing, outside of the physical shape of round objects - both traditionally keep the closest pixel.
To answer the parent GPU question, the short answer is it's not easily possible with this generation of GPUs. The current generation of GPUs are designed to handle meshes and texture sets fed in one at a time, not en-masse (e.g. a scene), though actions on these meshes are done in massive parallel fashion. That's not to say that ray tracing isn't done on GPU - it is - just that it's done on individual meshes. For some examples of ray tracing on GPU, check out Displacement mapping and relief texture mapping (though both use pseudo-ray tracing in practice for increased performance). To put it in perspective, you would need to hold your entire scene including textures on your GPU's memory to do real-time ray tracing on the scene (to properly handle specular reflections, which could come from any object) and that isn't feasible with today's hardware (for most scenes).
I've had nightmare problems with MS-SQL from a program integration standpoint (specifically version-to-version), but your mileage will vary depending on what you do. Oracle was hellish going from version 7-8, as well, but not so much after that.
From an admin standpoint, I love MS-SQL. For that matter, from an administrative standpoint I love active directory, too. From an integration, compatibility, and to a lesser exent performance (which has improved in recent years) standpoint, I hate them both.
There's a lot of bad blood between astronomers and geologists going back to the abbreviation AU - astronomers call it an "astronomical unit," while geologists call it gold (latin: aurum). Then there's using the greek alphabet to describe light phenomina instead of radioactive decay - that's got to rub geologists the wrong way.
Astronomers also dis Disney by calling some stars "red dwarfs" - implying that they're rednecks or drunks, yellow dwarfs (commonly called yeller in the south) being cowards, and white and brown dwarfs being derogatory slang for albinos and dirt diggers (as opposed to the high classed rock diggers), respectively.
We have to band together to stop this madness by the astonomers! Sign a petition today!
the "supergroup" GTR in 1986 (containing a member of Yes and Genesis) Hit: When the Heart Rules the Mind (peaked at #14 on pop charts) Miss: everything else
the duo Sly Fox (1985) hit: "Let's go all the way" (peak at #3, I think) misses: everything else
Those were 2 of the first albums I ever bought - GTR was even, I believe, the third I ever bought. Sandwiched in between was Weird Al Yankovic's "Dare to be Stupid," which had I not bought as an album, I'd probably never have heard Yoda (to the tune of Lola) or George of the Jungle (from the 1960s cartoon).
but then you've got albums that just aren't the same unless taken as a whole, like a lot of the later Pink Floyd stuff, Radiohead (or so I hear - I don't really like them enough to buy), and pretty much any rock opera albums from the 1970s and early 80s (artists like Styx, Queen, etc).
My wife's comments, in order, in about 20 seconds after she loaded her word created document and excel created spreadsheets:
"It's not exactly like MS Office - I hate it"
"I need my macros - this is useless without my macros"
"How do I add a VB script to this spreadsheet? I need them for my pivot tables to get the right information from the database. How do I access my Access database?"
After struggling with alternates for all of the above, there was really one sticking point - Access. I'm a fair DB guy, and have even had to use MS's other junky database, MS-SQL (yes, I am biased), but I prefer Oracle, MySQL, and Postgres to anything MS makes (the latter mostly because they're free). Nothing worked enough like Access to appease her - I guess once you're hooked on crap, there's nothing like it. I gave up and dropped $400 on an OEM MS Office and a new hard drive. Anyhow, I learned my lesson - it's bad enough that you can't teach an old dog new tricks, but it's twice as bad with an old wife (she's a year my senior - actually 2 for about another week - I can call her that;) )
you're confusing "market share" with innovation - what exactly did the IBM PC bring to the table? About the only innovation was using all off the shelf parts rather than a mix of off the shelf and in-house parts. I could argue the Disk ][ was more innovative.
The IBM PC was marketed as a business machine and sold mostly to businesses because IBM was the definition of business computing at that time. It wasn't really popular as a home computer until later, so you can't even say it brought the PC to most houses - everyone I knew had a Apple, trash-80 or Vic 20 at home (and later C64). I honestly don't remember even seeing a 5150 (the first model IBM) in any house, though I did see one at my dad's office. Even my tech-nut uncle, an electrician by trade, didn't have one, and he had an Altair and several other kit computers, a PET, Vic 20, and TRS-80 - his first PC was an AT (a 286).
Oddly enough, I remember the 5150 number because I disliked the Van Halen album by that title (the first with Sammy Hagar), not by the computer, although the naming is unrelated (VH is named after the studio address where the album was recorded).
bravo - I was going to post something much the same.
I think the only way enforcing a law like that would be to go after anybody in the US that is caught hiring offshore work for spam purposes. It would be hard to go after the pornographers unless they are the ones actually sending the spam because most of the time it's legal to create it where they are located. I seriously doubt that most porn mail originates in someplace like China or my spam box would be filled with Hot, Horny Asians just waiting for you - I'm pretty sure it's mostly outsourced from somebody in the US. I do get a few Russian, Asian, and Black e-mails like that, but 95% of them point to US sites tauting caucasian girls. Rarely do these get into my in-box (and if the filter catches them it blocks all links back to the site unless I release it to my inbox), but I sometimes lose legitimate mails like my Am-Ex bill (though I'm still messing with sensitivity settings)
Well, I knew the meaning of the word already from old RPG Harn, which is why I was curious - the first PC I created for that game started as a chandler (and it SUCKED - especially playing for 3 sessions before he even had RPG-useful skills... then he died and I played an NPC healer we rescued for the rest of the sessions, but I digress). Anyhow, the osa site was either slashdotted or incredibly slow at the time, so I couldn't really look to see it was named after Raymond Chandler (the detective novelist) and was curious to know how the project was named.
Edison also "Westinghoused" Topsy the Elephant, but he was against capital punishment, even though several of his employees went on to invent the electric chair. I guess he had no problem with cats and elephants;)
Edison didn't have a lot of inventions, but Edison mainly improved existing inventions like the light bulb and got boatloads of patents on that sort of thing. The money he made on these patents probably funded most of the research, so it wasn't without merit.
Your final point is something I was gonna say - AC was chosen mainly because it handles long run distribution of power, although the fact that DC doesn't step-down or step-up as easily also contributed (most DC power grids used multiple lines for varying voltages). Edison wanted lots of small neighborhood-level power generators since DC has a noticable drop in power in as little as a mile.
The point of using localized DC is to have one large, efficient transformer rather than hundreds of smaller and less efficient transformers.
well, join the real world - I used to do the same at work (I sometimes tested software on my Win XP and mac X.4 boxes which, for a while, we supported but had no hardware), but now my work blocks VNC, SSH, many internet radio protocols, terminal services, IRC, and a bunch of others. They don't block telnet (which they consider safer than SSH for some reason) and I can do IRC through my home CGI-IRC server. I've also modified shoutcast slightly to get by the server restriction, but I don't really use it much anymore (got enoungh disk space now for CD rips). All this was because they fear viruses and worms getting in.
Slashdot Mirror
Yeah, I agree - as far as I know Blizzard isn't using spyware to create targeted advertising aside from "you're advertising to gamers." In addition, you don't have to even play on Battle.NET to enjoy multiplayer, though if you want to avoid cheaters it helps. Blizzard pays their bills for the service thru advertising, so I understand the need to have ads there. I've heard it's profitable, too.
The main thing that bugs me, though, is that I would think Battle.NET is MUCH more expensive to run since games are hosted on Battle.NET servers. As far as I can tell, EA is providing nothing more than a matching service for multiplayer so I would expect they have trivial expenses in comparison.
I can skip BF2142 easily enough, but I hope this isn't a trend. If Midway ships UT2007 that way, I'll be much more disappointed, and if they add unlocks too (like BF), well, ew. I hate locks with a passion on that type of game - it's bad enough that skilled players already can beat the crap out of unskilled players, but then they tack on better equipment for those skilled players and give them an unfair edge, as well... yeah, well I'm not a fan. I might be ok with it if the game at least handicapped by skill (balancing the unskilled players with more numbers). In the real world, that would be like expecting a SEAL to be able to take on a dozen poorly trained militia alone - he's got better equipment and training, but a lucky shot can still take him out.
what frightens me is that I got an unlockable eVGA AGP 7800GS ~6 months ago for $4 cheaper (~$50 if you only count 4 * or better vendors like I ordered thru) than the cheapest price on pricewatch. It was $198 with express 2 day shipping (not sure if it was cheaper with slower shipping). It also proved unlockable with CoolBits, so I essentially got a 7800GT for the price of a 7800GS.
sigh - I need to upgrade soon - I'm jealous of my dual 3GHz's at work (with an extremely crappy GPU that actually does affect my work since my company is now adding GPU features into our products... sigh - if you spend $3000 on a PC, you can put in at LEAST a $50 GPU... this one listed at $27 last year when I got the machine).
heh - why not? I'll be there in a couple of days, so I can lend 'em a hand - as long as I've got a maß (beer glass) in the other. I'm college educated, so I can even drink and program at the same time (expect functional code, not legible code ;) ).
p.s. Please pay me in Euro (and yes, I will even pronounce it "oi row" just for the brownie points) - the dollar is practically toilet paper these days.
If you're a gamer, because Microsoft is pushing Vista Only (DirectX 10, exclusive to Vista by cabal) games like age of conan and Halo 2. I'm sure there will be many more that follow, and pretty soon a good chunk of the Windows gaming market will be forced into Vista through Microsoft's strong-arm tactics, whether they feel like they need the upgrade or not.
For that reason alone I will probably build a Vista/Linux dual-boot machine sometime next year. I certainly will continue to avoid WiMP for ripping, and I already planned to use my "old" PC's linux partition to become a "Media center PC" (using MythTV) if TiVO manages to kill my current Dish PVR. That 3 day thing would never work for me, as I often get weeks or months behind during work crunches. Micro$oft of all people should relate to that one.
and the SCO "screw the little guy" scholarship goes to...
DAVID BAHN!
(clap clap clap)
A state based injunction like that isn't even valid across state lines unless the business also operates there. If I were running my business in, say, Washington and had operations in Oregon, he would have to file suit in Washington, Oregon or in federal court for it to be valid. Filing a local case like that is a waste of time.
In the late '40s through late '60s the US military tested chemical weapons on its own people without their knowledge. I read about clean up efforts from some of these secret tests just a few years ago in a local rag (that's newspaper, if you don't know the idiom), so I dug around google for at least one reference.
The last time I know of that the US ASKED for volunteers, they were doing nuclear testing with citizens well within the danger zone to see the effects of nuclear weapons on people just outside the immediate blast radius (probably because PETA was protesting the use of animals - jk).
yeah, well he's right - Bush has proven without a reasonable doubt that the evidence doesn't show whether there is or is not global warming at this time and until we can prove it undoubtably, we should not put any unnecessary restrictions on campaign funde... er, I mean big businesses.
For that matter, Al Gore is an idiot*, so don't believe everything he says - I'm pretty sure he's secretly in league with law breaking PETA militants. George Bush's far superior IQ** always prevails.
* compared to a genius monkey
** compared to a dead monkey
p.s. this was intended to be funny. If you didn't find it mildly amusing, I'm sorry, and please don't grenade my house.
Ah, but in fact they bought the second publisher of D&D, BUT the first publisher of AD&D. Since the original version of D&D was discontinued and AD&D then became known as D&D, technically TSR, Inc. was the first publisher of D&D! If you were a true Sicilian, you would know that! Ah ha ha ha. Ah ha ha ha... klunk (sorry, that bit of logic was hitting Vizzini levels ;).
And to be nitpicky, it was Gygax and (Dave) Arneson that invented D&D, not Gygax and Kaye, who were the publication financers. I don't believe Kaye had much input, if any, to the material.
The idea is to have a database of signal patterns and compare this to an incoming signal, so it means that one time in 20 the signal was not identified as one of the stored signals. Since the method never gave a false identification (positive) as one of the stored signal patterns, your first assertion is true - 1 in 20 valid attempts is refused.
Ah - I see what you are getting at - you are saying that a ray tracer can find each pixel on the screen by the sort alone. I was giving worst case scenario estimates (big-Oh), which favor the rasterizer with z-buffering because it doesn't require a sort, but average case scenario estimates may well indeed favor ray tracing. Going back to the original topic, it seems to me ray tracing had a better big Oh than the old raster standard the painter's algorithm (like nlogn vs n^2) and had better average case analysis as n approached infinity. In school zbuffering was really a future thing, and I don't recall doing any analysis on it (I think it was even new in the book we used - Computer Graphics: Principals and Practices [2nd edition?] that year). For that matter, I was still writing code in PHIGS and GL, at that time, not even OpenGL, though I picked up some OGL by my second graphics class the next spring.
There are occlusion sort algorithms on the polygon side that may give it a good run for its money, like cPLP (since standard PLP is lossy) or hierarchal z-buffers, though there isn't really a "one shoe fits all feet" algorithm like you're talking about.
I've now read that even polygon data can be sorted into KD-trees, so it's not exclusive to ray tracing (as I stated earlier, it's been a long time since I've used them if at all...), so I don't think your statement is correct. I also don't think my first ray tracer had any sorting at all, though obviously sorting has some speed benefits.
The LOD culling that is most common in raster games lately is probably a chunked level of detail algorithm such as GeoMipMapping (sometimes combined with GeoMorphing for smoothness), which has a horrible problem with popping because most programmers intentionally choose to set the error values to noticable levels rather than have framerates suffer. The old standard was ROAM, but that was CPU based and GeoMipMapping is optimized for use with a GPU, making it much more appealing on today's hardware. A ray tracer won't help with LOD unless you reduce the shape count (as opposed to polygon count). Your best hope is probably to have less shapes to form your model to begin with, because you have more powerful primitives (cube, sphere, etc) and you could also make much more efficient use of constructive solid geometry to make simpler objects without polygons (yes, you can do CSG with polygons, too, but most of the time that's used to create a polygon based subsurface). Still, if you have too many objects in the scene you either need to distance cull them or simplify them, so the solution is non-trivial.
Another, newer form of raster detail ugliness can be seen in games like Oblivion, which I believe uses a technique called distance based detail mapping (look at the ground and move - you should see a sphereical detail section around your avatar - it's most obvious on the paved roads in towns) and distance based object culling (like trees). I ask, though, how Ray Tracing would fix that? Anything I can think of will still cause popping or repetitive detail. Terrain simplification will probably be the same with ray-tracers as it is with polygons because height maps are so much more disk efficient than other methods and therefore will still have the same sort of culling issues. If anything, I would expect to see programmers adding splines for smoothness before moving to a more disk intensive model.
I still don't see, but perhaps its because a KD-tree isn't "grid-locked" like a quad or oct tree (I vaguely remember them from classes, but it's been a long time). The raster code I've been working on does trivial adds to the view volume of blocks of data, then it does trivial rejects to the actual polygons after that, and I believe that has a similar effect to that of a KD-tree. The scene is potentially infinite since it uses terrain paging (in reality, it's a 4GB data file), so trivial rejection of everything would potentially take infinite time. The potential terrain that is in view is found using a surface projected cone, since it's mathematically much simpler than a frustum at the expense of accuracy (we leave the accuracy to hardware). In practice, this has proven extremely fast.
In ASCII-art below, O is the observer (camera), and X and Y are potential view volume. * is cached data that is out of the view volume that is paged in background. O is added by default. Up to 3 additional "blocks" of code can be added to the data that needs to be parsed based on camera direction - the Ys show what should be the worst case scenario (viewer looking toward a corner with parts of two other areas in view).
*****
*XYY*
*XOY*
*XXX*
*****
Although the code I'm working on only does this in 2 directions, it can easily be expanded into 3 - you could even use the projected cone method and just project it onto another side. Border crossing polygons can be split, if necessary, but I've never had to do anything like that in practice (for static objects that overlap the terrain I just add vertices on the border in a pre-processing step). So far I haven't had to do any simplification of dynamic objects outside of distance based detail reduction (a sqrt on each actor), but that could become an issue later.
Anyhow, I'm not sure if that is the equivalent of what you were talking about (I think it is, since ray traced objects are more difficult to split, thus the KD-tree instead of oct-tree), but it is a way to reduce an infinite dataset to a manageable one using data structures.
I've written a ray tracer and I don't know what you're talking about. A ray tracer needs to test each ray against every object in the scene. You can use bounding boxes or spheres to trivially reject objects, but that's true of either model (or you can use pre-sort data structures such as quad-tree, oct-tree, BSP tree, etc). For specular objects the ray is also checked against every object in the scene again for what it reflects, which can increase time significantly; however, ray tracing is faster in the respect that it only checks one point per ray on all objects rather than all points on the rasterizer that contain the triangle, which should be faster in a large scene (but in practice takes little time since it's handled by graphics hardware - I haven't had to write a software blitter for that sort of thing in ages).
I'm pretty sure they both rasterization using a depth buffer and ray tracing would take linear time (O(n)), while the painter's algorithm was O(nlogn) because of the required logn sort.
I certainly didn't mean ray tracing was any better than rasterization on diffuse - they're both bad. I just wanted to point out that ray tracing has flaws and shouldn't be considered the be-all end-all solution for computer graphics - there are still issues that need to be resolved. I have mixed feelings about ambient occlusion as the diffuse technique as presented by the paper, but I have the same feelings about it being used in polygons (it's fast, but not realistic). I certainly don't mean to bad-mouth ray tracing, because I think it's a great technique and I even predicted that it would become the dominant technique starting around 2010 or 11 (prediction was from 1996 and based on Moore's law), because that's when I believed single processor machines would hit the 17GHz mark which I had calculated was required to maintain 640x480x30FPS. Yes, that prediction was a bit short-sighted, as resolution expectations, CPU technology improvements, GPUs (not that current GPUs would help much because they're not meant to handle scenes), and multiprocessing were not considered.
you are correct - pixel/fragment shading is not free - both use the same default shading model in OpenGL and probably DirectX.
the default shader is, I believe, Lambert (a close relative to Phong - if not, it's Phong) for OpenGL and probably DirectX as well. Programs in the shader can change this to whatever you want it to be (e.g. a cel shader) and you would need to do that in either a ray tracer or rasterizer.
there's a lot of things I like about ray tracing, but it's not without flaw - it handles specular highlights fantastically, but doesn't handle diffuse well at all, so you have to bolt on other techniques. Most people (including Intel) use ambient occlusion since it's a quick technique (also commonly used in polygon based graphics), but it tends to make muddy shadows (see the wikipedia entry). Radiosity is more realistic, but the patch computations are incredibly expensive (but parallel-able). photon mapping is another method that could be used, but I haven't used it myself. In college I wrote (with a team) a simple ray tracer and shortly after that class wrote a radiosity engine, so I'm familiar with both techniques. I never did really understand how to combine them, but I remember seeing POVRAY do it in the mid 90s and really wanted to figure out how they did it (but I graduated and was putting in startup hours, so that never happened).
Oh, and waves on a lake are non-trivial - to be completely realistic, you need to deal with subsurface diffusion (or estimation of), foam and caustics (if you can see through the semi-transparent water surface). The specular mirror effect would be nice, but I don't see true caustics from either a raytracer or a rasterizer (you'd need to use ray bars or cones, probably).
I learned the same thing as you, but I believe the geometric complexity towards infinity favoring raytracing was only true for painter's algorithm. Actually, I think the magic number was actually somewhere around 100000 polygons for painter's algorithm, but the occlusion characteristic of the z-buffer negated the advantage.
The main advantages of ray tracing is physically correct lighting, shadows, and reflections, as well as the ability to use any geometric shape that can be mathematically defined (e.g. a sphere). The downside of parallelizing this is each additional parallel ray comes with a concurrent increase in memory requirements. Ray tracing also tends to do specular well, but isn't as good at diffuse, so many ray tracing programs hybrid-ize with something like radiosity or ambient occlusion.
clipping is less of an issue now than it was pre z-buffer (or depth buffer, if you prefer) and artifacting from it should be no worse than ray tracing. The painter's algorithm had the three triangle problem (same as the three polygons on wikipedia) that caused odd artifacting, but I don't see how zbuffer is any worse than ray tracing, outside of the physical shape of round objects - both traditionally keep the closest pixel.
To answer the parent GPU question, the short answer is it's not easily possible with this generation of GPUs. The current generation of GPUs are designed to handle meshes and texture sets fed in one at a time, not en-masse (e.g. a scene), though actions on these meshes are done in massive parallel fashion. That's not to say that ray tracing isn't done on GPU - it is - just that it's done on individual meshes. For some examples of ray tracing on GPU, check out Displacement mapping and relief texture mapping (though both use pseudo-ray tracing in practice for increased performance). To put it in perspective, you would need to hold your entire scene including textures on your GPU's memory to do real-time ray tracing on the scene (to properly handle specular reflections, which could come from any object) and that isn't feasible with today's hardware (for most scenes).
I've had nightmare problems with MS-SQL from a program integration standpoint (specifically version-to-version), but your mileage will vary depending on what you do. Oracle was hellish going from version 7-8, as well, but not so much after that.
From an admin standpoint, I love MS-SQL. For that matter, from an administrative standpoint I love active directory, too. From an integration, compatibility, and to a lesser exent performance (which has improved in recent years) standpoint, I hate them both.
There's a lot of bad blood between astronomers and geologists going back to the abbreviation AU - astronomers call it an "astronomical unit," while geologists call it gold (latin: aurum). Then there's using the greek alphabet to describe light phenomina instead of radioactive decay - that's got to rub geologists the wrong way.
Astronomers also dis Disney by calling some stars "red dwarfs" - implying that they're rednecks or drunks, yellow dwarfs (commonly called yeller in the south) being cowards, and white and brown dwarfs being derogatory slang for albinos and dirt diggers (as opposed to the high classed rock diggers), respectively.
We have to band together to stop this madness by the astonomers! Sign a petition today!
sometimes even bad albums contain A great song.
1980s pop hits for cases in point
the "supergroup" GTR in 1986 (containing a member of Yes and Genesis)
Hit: When the Heart Rules the Mind (peaked at #14 on pop charts)
Miss: everything else
the duo Sly Fox (1985)
hit: "Let's go all the way" (peak at #3, I think)
misses: everything else
Those were 2 of the first albums I ever bought - GTR was even, I believe, the third I ever bought. Sandwiched in between was Weird Al Yankovic's "Dare to be Stupid," which had I not bought as an album, I'd probably never have heard Yoda (to the tune of Lola) or George of the Jungle (from the 1960s cartoon).
but then you've got albums that just aren't the same unless taken as a whole, like a lot of the later Pink Floyd stuff, Radiohead (or so I hear - I don't really like them enough to buy), and pretty much any rock opera albums from the 1970s and early 80s (artists like Styx, Queen, etc).
My wife's comments, in order, in about 20 seconds after she loaded her word created document and excel created spreadsheets:
;) )
"It's not exactly like MS Office - I hate it"
"I need my macros - this is useless without my macros"
"How do I add a VB script to this spreadsheet? I need them for my pivot tables to get the right information from the database. How do I access my Access database?"
After struggling with alternates for all of the above, there was really one sticking point - Access. I'm a fair DB guy, and have even had to use MS's other junky database, MS-SQL (yes, I am biased), but I prefer Oracle, MySQL, and Postgres to anything MS makes (the latter mostly because they're free). Nothing worked enough like Access to appease her - I guess once you're hooked on crap, there's nothing like it. I gave up and dropped $400 on an OEM MS Office and a new hard drive. Anyhow, I learned my lesson - it's bad enough that you can't teach an old dog new tricks, but it's twice as bad with an old wife (she's a year my senior - actually 2 for about another week - I can call her that
you're confusing "market share" with innovation - what exactly did the IBM PC bring to the table? About the only innovation was using all off the shelf parts rather than a mix of off the shelf and in-house parts. I could argue the Disk ][ was more innovative.
The IBM PC was marketed as a business machine and sold mostly to businesses because IBM was the definition of business computing at that time. It wasn't really popular as a home computer until later, so you can't even say it brought the PC to most houses - everyone I knew had a Apple, trash-80 or Vic 20 at home (and later C64). I honestly don't remember even seeing a 5150 (the first model IBM) in any house, though I did see one at my dad's office. Even my tech-nut uncle, an electrician by trade, didn't have one, and he had an Altair and several other kit computers, a PET, Vic 20, and TRS-80 - his first PC was an AT (a 286).
Oddly enough, I remember the 5150 number because I disliked the Van Halen album by that title (the first with Sammy Hagar), not by the computer, although the naming is unrelated (VH is named after the studio address where the album was recorded).
bravo - I was going to post something much the same.
I think the only way enforcing a law like that would be to go after anybody in the US that is caught hiring offshore work for spam purposes. It would be hard to go after the pornographers unless they are the ones actually sending the spam because most of the time it's legal to create it where they are located. I seriously doubt that most porn mail originates in someplace like China or my spam box would be filled with Hot, Horny Asians just waiting for you - I'm pretty sure it's mostly outsourced from somebody in the US. I do get a few Russian, Asian, and Black e-mails like that, but 95% of them point to US sites tauting caucasian girls. Rarely do these get into my in-box (and if the filter catches them it blocks all links back to the site unless I release it to my inbox), but I sometimes lose legitimate mails like my Am-Ex bill (though I'm still messing with sensitivity settings)
Well, I knew the meaning of the word already from old RPG Harn, which is why I was curious - the first PC I created for that game started as a chandler (and it SUCKED - especially playing for 3 sessions before he even had RPG-useful skills... then he died and I played an NPC healer we rescued for the rest of the sessions, but I digress). Anyhow, the osa site was either slashdotted or incredibly slow at the time, so I couldn't really look to see it was named after Raymond Chandler (the detective novelist) and was curious to know how the project was named.
Edison also "Westinghoused" Topsy the Elephant, but he was against capital punishment, even though several of his employees went on to invent the electric chair. I guess he had no problem with cats and elephants ;)
Edison didn't have a lot of inventions, but Edison mainly improved existing inventions like the light bulb and got boatloads of patents on that sort of thing. The money he made on these patents probably funded most of the research, so it wasn't without merit.
Your final point is something I was gonna say - AC was chosen mainly because it handles long run distribution of power, although the fact that DC doesn't step-down or step-up as easily also contributed (most DC power grids used multiple lines for varying voltages). Edison wanted lots of small neighborhood-level power generators since DC has a noticable drop in power in as little as a mile.
The point of using localized DC is to have one large, efficient transformer rather than hundreds of smaller and less efficient transformers.
well, join the real world - I used to do the same at work (I sometimes tested software on my Win XP and mac X.4 boxes which, for a while, we supported but had no hardware), but now my work blocks VNC, SSH, many internet radio protocols, terminal services, IRC, and a bunch of others. They don't block telnet (which they consider safer than SSH for some reason) and I can do IRC through my home CGI-IRC server. I've also modified shoutcast slightly to get by the server restriction, but I don't really use it much anymore (got enoungh disk space now for CD rips). All this was because they fear viruses and worms getting in.