Slashdot Mirror
GeForce3: Real-time RenderMan?
b0ris writes "This review of the NVIDIA GeForce3 at The Tech Report does a nice job explaining how the GF3 chip can create advanced graphics effects in real time. The author raises the prospect of having real-time Final Fantasy or Shrek-style animation on the desktop in a consumer graphics card. The examples from the GF3 he uses to back it up are almost convincing, even if it isn't quite there yet. Will render farms go the way of the dodo?" Well, I'm all for dreaming, but its gonna be a few years before the GeForce8 can do renderman in real time, but when we get there, Final Fantasy 21 is gonna rule.
"Most of what we perceive to be "unrealistic" lies in the modelling of animation and movement -- physical dynamics and interactions such as collisions, deformations, effectsm natural pheonomena like wind, human locomotion, etc. -- it is sitll raw CPU speed here..." Actually, its not raw CPU power, anymore. Thats what a vertex shader is. A "shader" is a device that can give every pixel its own behaviour. Grass is a great example. Right now if a developer wanted to add grass, they would have to map every situation they wanted to be in the game. For instance, if they wanted the grass to be blowing in the wind, they would have to make an animation for that and add specific times for when that should happen. Let's say that a developer wanted to make the grass bend if stepped on. The developer would have to make an animation of the grass bedning down. Then they would have some command "IF Character on grass THEN play animation.avi". The problem with this is that if I step on the grass from the right side, and the animation was made of the grass bending to the right, then it would look stupid. If I step on a patch of grass, in real life the grass would bend the opposite way (the grass would not bend into me). A developer could add another animation, one for the grass bending one way and one for the grass bending another. But this would be time consuming and would still only work for two directions - what If I walked on the grass from the north or south side? Shaders fix this problem. They give the grass its own behaviour. The GPU's pixel shaders will determine what happened to the grass, and using the commands written by the developer (since they are programmable), will make the grass bend at the exact right angle. Pixels will no longer do what they have been assigned to do before hand by developers. They will have their own behaviour. They will react uniquely to every situation. This includes "deformations, the effects natural pheonomena like wind, human locomotion, etc." That is exactly what the vertex shaders are for, and it takes a load off the CPU.
> Will render farms go the way of the dodo?
When a video card has the power of a render farm, then people will simply make a render farm using those cards.
This will always be the case, until the rendering abilities of a card become indistinguishable from reality, and can render twice as fast.
All your race are belong to Gus.
I believe that Luxo Jr. is about 20 years old, so Duff is still right. :)
[disclaimer against redundancy disclaimer]
It takes a LOT more than polygon-pushing power to make a realistic image. The Geforce 3 (and the OpenGL or D3D which drives it) cannot do motion blur (REAL distributed motion blur, not accumulation), accurate reflection or refraction, shaders of arbitrary complexity, or any scene management and geometry generation operations.
ATI gives out source code for its Radeon drivers. NVidia does not. NVidia's chips are 3 generations ahead of ATI.
Now 5 years after the hoopla, one screenshot of a 320x240 camelion that looks like a movie, and 500 layoffs later, let's all say it in unison, "who gives a fuck if NVidia doesn't release any source code!"
People always mention this.
It would take a lot of development time and coopereation from the software companies to support low bandwith render systems.
The scene files that we are working with are 10s of MB, and they reference other files that are of similar sizes. You may pull accross 100MB of proprietary scene files (which means encrypted to the users) and then the system determines what to render. It may take 30+ minutes a frame, while either creating a scad of misc files or eating up memory (such as shadow map files, motion blur files...) and then assemble all of them to make a 2-3 mb image to upload.
The average user's home machine would only be a waste to studios. The bandwidth would kill us. Legal would kill us for letting proprietary data out. Your system would be smoked while rendering... or it would take a long long time.
All the transfer time of the scene files and the textures would take longer than the render.
We keep a nice fat backbone to the renderfarm for a reason. No sense in having 200+ procs waiting on data.
We do use software that allows us to use the users desktop, but this is over a LAN and not a WAN... and that makes a big difference.
-I just work here... how am I supposed to know?
Most of what we see with "realistic rendering" on desktop boxes is OpenGL / direct3d based. This isn't realy rendering, well... its not raytraced.
Its true that they are getting close and blurring the line between rendering and desktop 3D for all practical purposes there is a difference.
I just hope rendering never goes away... I need this job!
Another difference is that game movement is not near as complex as cinematic animation. Most game movement is pre-definded movements trigered by something. A lot of secondary animation and even some primary animation is done by a complicated set of equations. It all depends on the package, but sometimes with these solvers on, you might get 1 fps when viewing the animation. Until issues like that are fixed, you will not be able to generate stuff like that on the fly.
-I just work here... how am I supposed to know?
-sigh-
The eye can detect above 120, depending on the person. My threshold is around 80 or so, anything above that adds little to the gameplay, other than the framerate is less likely to dip below what I notice.
What makes 24-30fps acceptable in film and TV is motion bluring. Search the archvies for the arguments, as I don't feel like getting into it again.
Achieving Pixar-level animation in real-time has been an industry dream for years. With twice the performance of the GeForce 256 and per-pixel shading technology, the GeForce2 GTS is a major step toward achieving that goal.
-Jen-Hsun Huang, President of NVIDIA Corp.
Here is what Tom Duff from Pixar thinks about that:
These guys just have no idea what goes into `Pixar-level animation.' (That's not quite fair, their engineers do, they come and visit all the time. But their managers and marketing monkeys haven't a clue, or possibly just think that you don't.)
`Pixar-level animation' runs about 8 hundred thousand times slower than real-time on our renderfarm cpus. (I'm guessing. There's about 1000 cpus in the renderfarm and I guess we could produce all the frames in TS2 in about 50 days of renderfarm time. That comes to 1.2 million cpu hours for a 1.5 hour movie. That lags real time by a factor of 800,000.)
Do you really believe that their toy is a million times faster than one of the cpus on our Ultra Sparc servers? What's the chance that we wouldn't put one of these babies on every desk in the building? They cost a couple of hundred bucks, right? Why hasn't NVIDIA tried to give us a carton of these things? -- think of the publicity milage they could get out of it!
Don't forget that the scene descriptions of TS2 frames average between 500MB and 1GB. The data rate required to read the data in real time is at least 96Gb/sec. Think your AGP port can do that? Think again. 96 Gb/sec means that if they clock data in at 250 MHz, they need a bus 384 bits wide [this is typo. 384 _bytes_ wide!]. NBL!
At Moore's Law-like rates (a factor of 10 in 5 years), even if the hardware they have today is 80 times more powerful than what we use now, it will take them 20 years before they can do the frames we do today in real time. And 20 years from now, Pixar won't be even remotely interested in TS2-level images, and I'll be retired, sitting on the front porch and picking my banjo, laughing at the same press release, recycled by NVIDIA's heirs and assigns.
Vermifax
Vermifax
Logout
You're kidding, right?
How often, in every day life, do you notice diffraction and interference? I never do. Consider also that the size of objects which cause diffraction are the same order of magnatude in size as the wavelength of light (i.e. 10^-7m). Which, BTW, is far smaller than you can see. Now imagine you're going to keep track of polygons/voxels 10^-7 in size, for a room that's 10m by 10m by 3m. That's 10*10*3/(10^-7)^3 =~ 3*10^23 voxels to keep track of. Forget it. There are far better ways to simulate diffraction, if you really wanted it.
What I have seen, that's really cool, are Relativistic ray tracing. Do that Nsuck^H^H^H^Hvidia!
--Bob
1^2=1; (-1)^2=1; 1^2=(-1)^2; 1=-1; 1=0.
1) They already have enough power to render the movies, that's why they're already out in theaters.
2) The people who MAKE movies are a different group of people than those who SHOW movies.
3) Seti@home has to do a ton of redundant work, because people turn seti@home off in the middle of a block and never turn it on again, kids download block then try to upload spoofed workfiles to crank their work completed stats, and other garbage that the Studios just wouldn't tolerate well.
Consider this: Would the Seti project buy a server farm to perform this work if they could afford it? Or do you think they'd go through all of this crap, simply because they enjoy dealing with crap more than doing science?
by Mike Buddha -- Someday the mountain might get him, but the law never will.
The reason you think you've never seen photo-realistic CG is because when it's photo-realistic you can't tell that it's CG :) "Special effects" aren't the only computer graphics in movies nowadays; in a lot of movies nowadays the set that the movie is filmed on isn't actually what's seen in the movie - buildings are added (in a LOT of movies, many of which you wouldn't even think would have CG at all), people are added (for example, in the edited version of "Eyes Wide Shut", CG people were added to block out penetration and appease the US's puritan hangups. The problem was, the people were completely static and it was obvious that they weren't real), atmospheric effects are added, etc.
I don't remember what movie it was, but I read an article on the making of some movie (set in the 1800's, one of those cheesy romantic dramas, released about 2 years ago) and they showed the original filmed scene where you could see scaffolding, cameras, and lights. Then they showed the final result, which was a fully convincing 1800's-era scene. Most of the buildings and background people were created in 3D Studio MAX and rendered with Mental Ray, and you just can't tell. It was truly impressive. The buildings moved perfectly with the camera angle, the CG people walked and moved perfectly (there were no closeups of them, which removed the hardest part - facial modelling. The human eye is very good at picking up inconsistencies, especially in objects we observe every day, such as human facial emotions. It was very impressive nonetheless).
Of course, convincing facial modelling isn't impossible - look at this picture by Asier Hernaez Laviña, which was modelled and rendered in 3D Studio MAX. Not video, but it's an amazing technical achievement and is almost indistinguishable from a photograph.
--
>>he is wrong about 99% of movies being
>>rendered with renderman.
actually i'm not. Other than Antz/Shrek from PDI (which have their own in house renderer), i'd say 95% (conservative estimate) of the feature quality CGI put out is done with PRMan.
Mental ray - yeah i'd been used for a few things (e.g. Flubber) when you absolutely HAVE to use raytracing. But other than that, no way. It can't swallow the type of scenes that PRMan can handle, the memory requirements are FAR too high, and it's motion blur is weak at best in comparision. There have been plenty of post houses that have tried to use it, but once you start to throw large scenes at it that require quality anti-aliasing, motion blur, and HUGE geometry databases it just falls apart. MR3.0 will tackle *some* of these problems, but the fact that it's a raytracer gives it some inherent limitations of what it can handle.
And this isn't from somebody who hates mental ray --- i worked on it at Softimage for 3+ years. It's a good renderer, but it's no PRMan.
>>The shader language can raytrace.
Correct. and PRMan returns BLACK whenever you call trace. Therefor PRMan doesn't raytrace - ever.
You can hook it up to another renderer (e.g. BMRT, RenderDotC) to handle the trace() calls, but that raises it's own issues.
Sarcasm mode on:
Will computers continue to get faster? Will we someday have lightbulbs in every room of the house? Will everyone who wants one be able to afford an automobile one day?
Well, it'll be a few years before we're able to play color video games on our personal computers, but when we do the arcade games will really rock!!
Sarcasm mode off:
Really? What kind of sensless 'wow-computers-are-getting-faster' is this? The article actually makes sense and is interesting. It explains how computers are getting faster. It's the silly, so-called 'editoralizing' that stoopid.
Aah, change is good. -- Rafiki
Yeah, but it ain't easy. -- Simba
I have a copy of Lightwave 3D and it supports OpenGL for realtime previews of animations. It uses its own internal renderer (or screamernet) to do the final rendering, but loading a scene in layout and hitting the realtime preview looks pretty neat on a GF2. I'd really like to see if some of the additional textures and lighting capabilities will be supported in LW6 in the future...
LOAD "SIG",8,1
LOADING...
READY.
RUN
Raytracing is only necessary in reflection and refraction -- which can be faked pretty damn good now.
Other shading methods (radiosity for proper lighting) are used elsewhere.
Real-time rendering CAN be achieved by using the proper methods and not just throwing the entire ball of wax at any scene.
The idea is SMART rendering: Z-culling (so you only render pixels that affect the scene); polygon reduction (so you don't bother with a 10,000 poly item that is so far away in the frame it is a single pixel); variable mapping (using environmental maps for reflections when appropriate (like fly-thrus where there are only "background" objects).
Think Hollywood set -- build (and shoot) only what the camera will see, nothing else.
--
Charles E. Hill
Learning HOW to think is more important than learning WHAT to think.
Well, I'm all for dreaming, but its gonna be a few years before the GeForce8 can do renderman in real time, but when we get there, Final Fantasy 21 is gonna rule.
My dream is that GeForce8 can make it unnecessary to discuss the quality of the game in the same sentence we discuss the quality of the graphics. For years now, we have seen one product after another try to top the preceding generation in terms of delivering beauty and graphic heat -- and yet it has been a long time since games have really done, IMHO, a great job of delivering fun.
This is not to say that twitch isn't fun -- or that pretty isn't interesting. Its just to say that I'm not sure that more more photorealism equates to great gaming.
Actually, the biggest problem with 60Hz is the fact you get a low frequency beat with the light output from 60Hz flourescent or incandescent lights. I've noticed that moving from Austrlia (50Hz lighting) to the USA (60Hz lighting) that a 60Hz refresh is a LOT worse here than in Australia.
Fear: When you see B8 00 4C CD 21 and know what it means
This is completely false. Nyquist doesn't apply to a synchronous transfer.
The electron gun scan rate is CONTROLLED by the video card, so the frame rate coming out of the card is constant. The RAMDAC accesses memory at a constant rate, determined entirely by this refresh rate. Frames are generated into the back buffer and flipped into the front buffer once the entire frame is generated.
Nothing is actually "sampled" in the chain from frame generation to displaying the image (unless we want to talk about pixels rather than frames).
This means you need 30fps generated by the card to get 30fps displayed on the screen - not the 120fps you are suggesting!!
Fear: When you see B8 00 4C CD 21 and know what it means
Disclaimer: I work for DotC.
RenderDotC doesn't raytrace either. You might be thinking of Mirage-3D, the author of which, the great Timm Dapper, also works for DotC.
sub f{($f)=@_;print"$f(q{$f});";}f(q{sub f{($f)=@_;print"$f(q{$f});";}f});
The eye can't even detect anything above 30 FPS or so.
The cake is a pie
Cells in the retina have a recovery time of ~30 milliseconds. Do the math.
(If "80 FPS" seems choppy to you, it is because this is an average. The framerate only has to dip below 30 or so for a hundred milliseconds or so to be detectable.)
Here's a link from google
The cake is a pie
> But I think we're still a ways off from being able to do away with the repetitive textures that dominate roads or brick walls in videogames.
Exactly. Why? Because we use textures as a form of compression. Computers just don't enough memory and bandwidth to allocate an unique texture for EVERY surface. (Light maps push this boundry though, as can be noted in Quake with it's light map cache.)
The reason textures even "work" to begin with, is that from a distance, a surface looks pretty much "flat". But at the microscopic level (atoms) the "surface" is extremely hilly. In the real world, *ALL* those micro details ADD UP when that object is light. And that is why the textures in any game stand out like a sore thumb. It's not the "textures" themselves that are the problem. It's the surface roughness and lighting that we are CRUDELY approximating (for real-time rendering.) Bringing this back on topic, thats why off-line rendering farms can look SO much better and realistic. They have the time to do all the expensive math calcs needed for realistic lighting (i.e. ray-tracing)
> It didn't appear flat, but rather bumpy
That's why bump-mapping is so badly needed in today's games. It fakes the atomic "roughness" of a surface.
I'll dig up a link to that Quake 1 client (with source) that added bump-mapping later today. The cool part was that you could adjust the level of bumpiness. A textured brick with a little bit of bump-mapping looked WAY better and started to look like a real brick (with indents.)
>> . Pixar renders their frames at a color depth of something like 16 bits per channel.
> The fill rate on the Geforce series is reasonably high. The color depth is 32 bits
For the GeForce, 32 bits per pixel is only 8-bits per channel, and can leave bad banding and mach artifacts with overlays.
16 bits per channel is 64 bits per pixel (ARGB). Unfortunately it will be a while before consumer cards even start thinking of supporting it.
The GeForce 3 demo at MacWorld was Luxo junior rendered in real time, so Pixar quality animation is possible, for a sufficiently early value of Pixar...
Here, even the most advanced renderer won't help much if you're talking about real-time interactive stuff -- it is sitll raw CPU speed here...
There's 10 types of people in this world, those who understand binary and those who don't.
The theory goes: The fact that none of the situations described in the song are ironic, is itself ironic. :)
"Destroy science and religion. Science would re-emerge exactly the same; but not religion." - Penn Jillette, paraphrased
Its a 10? minute long realtime rendered video.
www.theproduct.de
It's really amazing, and it would seem that what they were describing in the article is already here, but maybe im not quite clear on what they meant.
Malcolm solves his problems with a chainsaw,
Malcolm solves his problems with a chainsaw,
And he never has the same problem twice.
1) Tom Duff sounds on the money with regards to the technical misconceptions...but an even bigger ever elusive problem: 2) "Pixar-level" animation in the end is not about polygon count, it's about COUNTLESS man-hours spent modelling, lighting, and animating....no card can ever replace that.
You are wrong. The shader language can raytrace.
Using BMRT together with PRMan, it can ray trace, and many people use it. Like in Hollow Man, for instance.
Here is a gallery, which includes Hollow Man. The call looks like this :
color trace (point from, vector dir)
Traces a ray from position from in the direction of vector dir. The return value is the incoming light from that direction.
Source
Education is the silver bullet.
What's so frightening about terminating a program?
Education is the silver bullet.
BMRT Raytracing Howto
You people are amazing. You don't even bother to look at my link, and you tell me I'm wrong.
Education is the silver bullet.
If I'm in Word, and I add an Excel spreadsheet to it - and then I print it - is that Word printing a spreadsheet? By my definition, yes - by your definition, no.
By my definition, Word can use Excel as a spreadsheet renderer. By your definition, apparently, just because Excel is not built in to Word, it means that Word is incapable of printing spreadsheets.
They're telling me, "There's no way for Word to print a spreadsheet," and I'm saying they're wrong. You're also saying I'm wrong. But I'm not, I'm right - and the page I pointed to shows how it can be done. It's not easy, and there are problems, but it can be done. The actual facts are on my side.
Your definition might be more technically correct (since "we don't say that program 1 is performing the specific task"), but mine is certainly more useful. Since my point was that program 1 is able to perform a specific task, by commincating with another program. Many programs are incapable of communicating with other programs in such a manner, and that makes PRMan pretty cool, in my book.
By the way, it's "English," not "english."
Education is the silver bullet.
After executing your commands, I am left with the following statement from you :
prman can't do radiosity via an SL trace without BMRT.
Now, I will apply the English language suggestion for good writing, "Don't never use double negatives," after which, your statement becomes :
prman can do radiosity via an SL trace with BMRT.
This is shockingly like my original statment :
PRMAN CAN RAYTRACE USING BMRT AS A TRACER.
And I guess I agree with myself. So, then I can only laugh, when I read your insulting statement :
If you still think that's prman doing raytracing then I suggest you take a remedial english class.
Because I just proved that you "think that's prman doing raytracing"! So, why exactly did you need to insult me twice in your post, in order to agree with me?
Education is the silver bullet.
Thanks - having the last word is kind of fun.
You might consider getting some help for that persecution complex.
If this is a back-handed apology, I accept. If it's merely another insult, you might want to consider taking some agression management classes. I think it's somewhat childish of you to move an argument of fact into a namecalling bout, suggesting that I don't live in the real world, don't know how to use the English language, and now that I need therapy, and have a "widdle head."
Semantic debates are sometimes enthralling, because you can twist words to make facts lie - but they don't really further understanding. Your definitions of "program" and "call" are well stated, and I believe I understand them, but I don't believe that they reall help you.
Word can call Excel to do spreadsheets, but that doesn't make it a spreadsheet program. PRman can call BMRT to do raytracing, but that doesn't make it a raytracing program. Eudora can call PGP to do encryption, but that doesn't make it an encryption program. The JVM can call methods in a class file to give you the long-distance phone calling capabilities in DialPad, but that doesn't make the JVM a long-distance phone calling program. Quake III : Team Arena can call jpeg library functions to load jpeg images, but that doesn't make it a jpeg image loading program. Internet Explorer can call Hotmail to send email, but that doesn't make Internet Explorer an email program. PRMan can call the shader language to do shading, but that doesn't make it a shading program.
Utility is an interesting thing. I can use a butter knife to turn screws, and I will agree with you that my ability to use a butter knife in that manner doesn't somehow turn it into a "screwdriver," in the traditional sense. But, if your definition of a screwdriver is merely that it is an implement with which one may turn a screw, it becomes a pretty hazy line. If you ask me for a screwdriver, and I hand you a butter knife, I'll laugh at myself, right along with you. It's silly, it's not what you asked for, but it'll do the job. You have to agree that if a demolitions expert is trying to defuse a nuclear bomb without his tools, the clock says 26 seconds, and he asks me for a screwdriver, if I hand him a butter knife - I've saved the day!
Most effects houses have a hard time staying in the black, and most use PRMan. If an effects house is down to the wire, and their client demands that a certain effect needs to look more real, and the only way to pull it off is by having PRMan call BMRT to do raytracing, they'd rather use my definitions of "program" and "call" than yours.
I love your last paragraph - I think you should use it, then next time you're on Jerry Springer.
Education is the silver bullet.
Stores around here are kinda slow as well. But, pricewatch always has the new stuff as soon as it comes out. You have to pay for shippig ans all that, but if you want the latest, there you are.
No replies made to AC posts. Please log in.
I'm using Win2K SP1, and the driver for the GEForce 3 works quite well. All the graphics features work. The chameleon demo is indeed impressive. Considering that NVidia just started shipping the Win2K driver, I'm quite impressed.
I have more fans in my systems than most users go for. Nothing is overclocked. GeForce 3 boards have heat sinks on the RAM and a fan on the graphics chip. This board is pushing the limits of what's possible with current semiconductor technology. Power and cooling should be sized accordingly. Just shoving this into some low-end PC with a minimal power supply and fan may not work.
However, I'd say we ARE about advanced enough to do crap like this in realtime... ;) (No goat links, I swear!!)
However, there is some hope... I remember reading in a great book about 3D games (Black Art of Macintosh Game Programming) that raytrace-quality realtime games would be (according to the author's math) about 20 years away. Interestingly, that's exactly what the Pixar guy predicted, and that book was printed in 1996. My observations: today, Pixar does far more than simple raytracing. It's radiosity up the wazoo, for example (I assume ;). So to me, this suggests that ~20 years from when the book was published, we will be able to have realtime raytracing of 1996 quality. Still not too shabby. BUT. There are gazillions of optimizations you can make in realtime games that you can't make in raytracing. Here's how I see it: We can improve the algorithms a few powers of ten, efficiency-wise. (Don't say we can't, you'd be very wrong.) We can speed up our processors a few powers of ten. I think we're getting there faster than these guys are suggesting, just as long as we don't aim for the moving target of Today's Pixar Production$. (As he points out, there will never be a day when the realtime graphics are as good as the prerendered ones, simply because the big companies have the cash to throw at it to make it look better.) Anyawy. Sorry this was so long. Great stuff ahead, though. :)
The streets shall flow with the blood of the Guberminky.
It is slightly ironic that the same people who one day were saying digital art is still art are the next day saying that animation on the level of a movie that took thousands of man-hours to create can be generated by a computer. Thus stripping away the art value of the movie (or at least the animation in the movie).
It does piss me off when people misuse words, especially words that are very nuanced and clever. But there you go. People are stupid.
BTW, good examples.
"He's more machine now than man, twisted and evil."
Not only that, but it's not like "slow" renderers are going to be standing still while nVidia advances their technology. There is a heckuva lot of room for advancement before we can generate photorealistic images of any arbitrary environment. Wake me up when there's a GeForce that can do stuff like this in real-time (and don't forget the motion blur!), and then we can compare state of the art again.
Sure there is incredible potential in the geforce 3 and programmable stuff like it, but computer graphics in movies has too many extra things going on for it to be done in hardware realistically. If there was specialized hardware for all the nuances like raytacing, inverse kinematics, procedural textures, volumetrics, radiosity, etc. then maybe in could be done in realtime, but I don't think we're going to be there any time soon. Rendering is very slow because for the most part it can't take advantage of special hardware because the quality just isn't there. Anti-aliasing, motion blur, and good depth of field take time as do physics simulations, heavy subdivisions, and complex shaders. It's coming, but not for at least 4 more years, and that can only happen with the high end rendering tools like renderman and mental ray being written to take advantage of a specific card which isn't going to happen unless there are standards in place.
This Wiki Feeds You TV and Anime - vidwiki.org
Renderman doesn't do radiosity. Different techniques are used other than raytracing for realtime games and probably will continue to be for some time.
This Wiki Feeds You TV and Anime - vidwiki.org
After all, that orange tree would have taken years to grow.
How can we continue to believe in a just universe and freedom to eat crackers if we have no ale?
I never liked FF. Big Dragonball style hair, people riding these weird chickens... silly big swords... that's all I ever saw. Well, a friend of mine was playing FF8 and his character had to dress up like a woman to go do something. I guess that's more interesting than a big chicken. I'd rather watch Record of Lodoss War or some other old classic.
The FF movie looks nothing like the video games I have seen, thank goodness. I hope the characters in the movie aren't breeding those giant chicken things...
- Someone Confused by the FF Hype
"10,000 spoons when all you want is a knife", how is that ironic? It would only be ironic if later you discovered that a spoon would have done just as well for say, opening a can of paint.
A journey of a thousand miles starts with a brutal anal raping at airport security
If you read the RenderMan Interface Spec. (V 3.2) the entry for trace() reads as follows:
So, you can call trace() in prman, but it's not going to do you any good.
That said, it is possible to write a ray tracer in the shading language! This has been done, in fact, by an insane person named Katsuaki Hiramitsu. This shader, however, does not use the trace() call. The trick lies in actually defining the objects you're going to do ray tracing on in the shader along with your own version of trace(), which is, by necessity, intimately bound to the type of object you've defined.
So, saying the shading language can ray trace is like saying you can keep yourself alive for a while by eating selected portions of your own body. It's possible, but certainly pessimal.
I still play Ms. Pac Man, but I hardly ever play games from just five years ago.
Graphics are cool and all, but they're essentially just pornographic. Not in the sexual sense, but pretty graphics just sit there vacuously to amuse your eyes. As has been said long and loud, game developers should strive to focus at least as much on gameplay as they do on making their graphics cutting edge. Give the user an elegant interface, something fun to interact with, something new, and something challenging.
You see? You see? Your stupid minds! Stupid! Stupid!
Remember Freedom is slavery, war is piece, and Ignorance is strength!
Now I need to stop goofing around here on the slashdot insoc-msdn party news and go back to work studying the 11th edition of NewSpeak by MSN expedia. I keep hearing people here on slashdot speaking in oldspeak.
You people all need to learn how to excell(smart tag link to Microsoft office homepage)on what you do to learn and explore (smart tag link to internet explorer site)your newspeak party langauge. With free(link to how free you are with hailstorm/.net)enterprise and innovation to lead the market, great Microsoft can actively access(link to ms access)all the information we need. We need an active innovatorto actively explore, and actively leadthe market, and they ask that we all support the revoluton by your activation subscription.
See that wasn't hard. You need to all speak newsspeak and only use these adjectives innovation, lead, explore, access, active, word, excell. This will make thought crime impossible. less is better. For something double-pluss-un-innovative like linux you should not use the word bad or sucks. You all are required to use the following words above with the -un extension. If something is really innovative you need to put doublepluss innovative or really bad its doubleplus uninovative. Everything non Big brother is just plus uninnovative. So remember its not GNU-linux but doubleplussuninnovative gnu/linux. Now lets here you all respect big brother now and after my newspeak lesson I will play some video games and render linus doing a double-plus-un-innovative things to scare people so that Bill Gates can actively explore my record so that I can be considered loyal member who doesn't doublethink.
http://saveie6.com/
uh, since when has nvidia had unstable drivers? They're usin' unified driver architecture, meaning that the same drivers that they made for the original Geforce 2 years ago now can be used with the geforce3 (although the drivers might need to be tweaked a little for the differentiating architecture). This gives Nvidia over 2 years to tweak their drivers for maximum stability and performance. the only reason why you're complaining, im assuming, is because you are using the leaked beta drivers that Nvidia never authorized. well guess what? they arent supposed to be out! So if you care about stability that much, just get yourself the official drivers and stop bashing the company for something it has done right :P
Well, I had to throw my TNT2 card into the trash because it had unstable drivers. I recently built a low end machine for my wife to play everquest, and since eq doesn't need a high end graphics card I purchased a TNT2. About 50% of the time that she zoned or started the game her comp would get a fatal exception 0E. In fact it would get this fatal exception:
A fatal exception 0E has occurred at 0028:C0006EB2 in VXD VMM (01)+ 00005EB2
After days of playing with it.. buying new memory, trying everything I could I finally find out this is a problem that the TNT2 card has been having for years. I would say the drivers are realitivly unstable. I bought a Voodoo5 card for the machine and it hasn't crashed since. If you goto this link you'll see that the only workaround I could find (the AGP apeture size and disable video caching did not work) was to TURN OF HARDWARE ACCELERATION. ROFL! Why by a 3d card at all if you have to turn off the hardware acceleration to get it to work properly? Can't even run EQ with the acceleration turned off. And just to prove that the problem is not eq's.. I had the same problem when running Half-life on the card.
It saddens me that Nvidia is quickly gaining a monopoly on the graphics industry, because I truely do not want to purchase another card from this company after they knowingly allow the old TNT2 cards to have driver problems without fixing it.
I thought someone said there was going to be free beer!
really? I have a tnt2 (Diamond tnt2...I forget the revision number...) in my machine...I know its old, I'm waiting for creative to release a geforce3. But, I've had my Diamond Viper 770 forever (and so have all of my friends) and I (them too :-) have never had _any_ repeat _ANY_ problems with them. Be it using the drivers that came with the card, or the latest detenator drives I downloaded from Nvidia (for Linux or win32). Either your board was defective (did you try to exchange it?) or you shoved it into a pci slot some how. Was that voodoo5 you bought pci? If it was, have you ever tried another card in your AGP slot? That could be fried. Also, check your BIOS, do you only have 64 megs of ram, and have your AGP apeture size set to 64 megs? The TNT2 chipset was one of the best chipsets in my mind. I'm still running a Viper 770 on my windows boxen at home, and my linux box here at work.
:)
Heheheh.. Yeah the voodoo 5 is AGP. I don't think it's the TNT2 chipset but the way the specific card vendor (IOMAGIC) integrated the chipset into the card didn't work with the drivers, i.e. the TNT2 chipset works fine.. but the IOMAGIC card has problems with the drivers. I agree that it is probably a problem with the IOMAGIC card, and not Nvidia, but the people who purchased IOMAGIC cards need support too!
If you see my name up there, I develop device drivers for PCI cards, so uhh.. I think I can tell the difference between a AGP slot and a PCI slot.. heheheheh.. funny to think about trying to fit a keyed agp card into a PCI slot.. I'm certain the pin locations are different, so if you did manage to get it in I don't think you would see ANY graphics..
The machine had 256 megs of ram, and I tried all the various AGP apeture size settings with no success.. the only way I could get the card to work reliably is to turn off hardware acceleration in windows all together, which wasn't an acceptable fix for me.
As for the card, I've seen several iomagic cards have the same problems (3 or 4 different cards that people have had) in some 3d games (it seems EQ and Half-life is the worst). Again, I think the chipset is probably okay, but the drivers didn't seem to work with iomagic's specific implimentation of the card.
As to the person who mentioned that it was funny that I got a Voodoo5.. Well I had only built the machine for 1 purpose, and that was to play Everquest. I purchased the Voodoo card AFTER Nvidia had purchased 3dfx, but I got it because I knew that card had very few issues with EQ, compared to the number of issues I saw people were having with the GeForce2 cards at the time.
I thought someone said there was going to be free beer!
It sounds to me like you had somthing else going on, or you had a bad TNT2, I've played quite a lot of EQ over the past year on a TNT2 without a single crash, before that I was using a PCI TNT without a crash as well.
I would guess you either had a heating problem, old drivers lying around in the windows directory, or some motherboard/card interaction. nVidia has probably the best driver support out of all the consumer card companies, compared to say 3dfx who couldn't even be bothered supporting my Voodoo2 properly a little over a year after it was out and never did get a real OpenGL implementation going.
Well.. I'm 99% sure that it was a driver problem because if you follow my link lots of people are having the problem, and I have seen the problem happen on several different cards on different machines. All of the cards were from one vendor tho, IOMAGIC. Now, that aside, you guys have definately made me MUCH less worried about my decision to purchase a GeForce3 card for the new machine I am building. I have honestly been very worried about purchasing any Nvidia chipsets since I saw the problem. I know that lots of peope HAVEN'T had the problem, but the problem definately does exist...
I thought someone said there was going to be free beer!
It seems to me that we will never get to the point of realistic rendering in games because by the time one reaches pixar-level animation, there is so much art and detail going on before the rendering stage.
If we do get game companies trying to produce games with "realistically rendered" graphics, won't they need budgets of 100 million for each game to develop all of the data (detail, world, etc.) that the hardware will operate on? Then we'll be walking into the software store laying down $5,000 for a game instead of $50.
STOP . AMERICA . NOW
Most of this stuff can continue to use Newtonian optics. Even Bragg diffraction, now that I think about it. So you're right that degenerate modelling will do us for quite some time, and most of the canon.
But the first person who wants to model proper rainbows, sun-dogs, or coatings...
The really hard part about QED isn't the iterations. It's defining the integration regime in the first place. I haven't looked in a couple of years, but I bet even the best Feynman diagram tools still can't work without heuristic input.
--Blair
"Luxo, Jr. always wanted to grow up to be an electron microscope."
Damn cool, but it's not QED yet.
It's all based on waves (classical theory). And it seems to be a surface phenomenon only, and dependent only on the geometric surface description.
Real QED would include interactions of photons with the subatomic particles of the atoms within the body of the material.
Diffraction and thin-foil effects are too-simple examples, with classical analogues. Phase-conjugate mirrors or simulacral holograms; now there you have to have QED.
This isn't to take away from Stam's work. It's gorgeous. The idea of walking into a bar with a double-barrelled shotgun and blowing away the pseudo-retro Wurlitzer with the wave-rendered CDs rotating on top, wave-rendered shards of CD spinning through space...
The idea of finding a secret because of its slight change in lustre vs its surroundings when the overhead lights dim and an accent spotlight becomes dominant...
The idea of being able to tell painted plastic from painted metal and painted wood, or black dirt from gunpowder and incinerated-demon charcoal...
Someone get nVidia on the horn.
--Blair
Is anyone working on a Quantum Electrodynamic model of raytracing? Diffraction gratings would be cool. It would improve other things. Like hair, thin films, etc.
--Blair
I think it's a little ironic that today we talk about bringing Shrek and Final Fantasy to the desktop when just yesterday a slew of 4's and 5's affirmed that, beneath raw power, there is art in computer graphics.
Believe me, there is a lot of artistic skill that goes into making animation like that, from storyboarding to complicated modeling and animating to directorial talent and writing ability.
Just because Avid-style editing has been brought to the desktop, doesn't mean what you see on iFilm is as good as what you see in theaters. Most of the time it isn't. It's all about the talent, not the tools.
Case in point: Robert Rodriguez, who scraped together only $7,000 to become one of Hollywood's hot young directors. For those who don't know about him, his latest film was the hit Spy Kids.
The problem is that even if we can have 100k polys per frame, it's going to suck to see your marines run straight to gunfire, or just stand and watch while their friends get slaughtered because their fuzzy logic is too fuzzy to realize they could help. Those bastards are always out of the reaction range, even if it's just a few meters away. Frankly, AI has been too long neglected. I know AI can be pain in the ass to program, but I'd like someone to make a licensable AI engine, just like with GFX engines these days...
--- Hajotkaa siihen, kapitalistit!
I am not sure about real time rendering for flicks like shrek and final fantasy. But real time rendering for stuff that is less detailed is definitely here. I've worked with Maya quite a bit, and I remember hating the way I had to wait ages for it to render my movies ( 8 hours meant that I would leave it to render over night). Now what I was working on was not cutting edge and my use of lighting and textures was quite simple. I'm pretty sure that the GeForce 3 would render something like that in real time with no problems at all. Real time rendering is here and I'm happy that I wont have to wait a day before I can see the result of my work. This means a massive improvement in productivity and ease of use.. atleast for me.
- Tempestdata