3D Graphics For Firefox, Webkit

angry tapir writes "A group of researchers plans to release a version of the Firefox browser that includes the built-in ability to view 3D graphics. They've integrated real-time ray tracing technology, called RT Fact, into Firefox and Webkit. Images are described using XML3D, and the browser can natively render the 3D scene." The browser will be released within a few weeks, the researchers say, and they are checking with the Mozilla Foundation about whether they can call it Firefox.

Comments

Clarification

[Pojut]

Images are described using XML3D, and the browser can natively render the 3D scene.

Does this mean this technology will be used strictly for 3D images/scenes, or when they say 3D are they referring to gaming?

Re:Clarification

[BRock97]

Even more confusing, is this meant to compliment WebGL or replace it? While I think it would be neat-o to have real-time ray tracing in the browser, the WebGL working group consists of some big names like Apple, Google, and Mozilla.

Re:Clarification

[BhaKi]

Does this mean this technology will be used strictly for 3D images/scenes, or when they say 3D are they referring to gaming?

Obviously and according to TFA, they're referring to 3D images/scenes. Gaming would require, amongst other things, browser-support for raw input devices, (at-least partial) server-side magic for processing interactive events. While these are definitely possible, they're not what this is about.

Re:Clarification

[Steauengeglase]

I dunno, but this is a field that desperately needs some kind of leadership. How many technologies have we had in this field over the years. Countless markup formats, plug-ins and APIs and we haven't gotten too far in all this time.

Re:Clarification

[supernatendo]

Looks like some "research group" a.k.a R&D for a company in the works with this Slusallek guy as the C.E.O.

If that is the case then you can bet that down the line you will see him try to sue WebGL for patent infringement...

Obviously he can't benefit monetarilly from an OpenSource project like WebGL so he makes his own browser based on Firefox which uses HIS proprietary 3D rendering and still wants to call it "Firefox"

A fine patent troll in the making if I may say so myself...

Re:Clarification

[Nadaka]

Patent trolls don't actually make anything, this guy did. He could turn out to be a litigious bastard anyway, perhaps not. But in any case he would not be a patent troll.

Re:Clarification

[clone53421]

Depends on what you mean by “gaming”... a lot of games can be played with the keyboard and mouse, and without connecting to a server at all. And Ajax can be used when the server is needed.

What I find more interesting, though, is that this would reveal the source code of your game to a much greater degree than, say, a Java applet does... allowing people to perhaps hook into or modify your code in order to cheat. An attempt to create anti-cheating measures would be interesting, to say the least.

Re:Clarification

[maxwell demon]

I think a simple anti-cheating measure would be to have the function names of the delivered JavaScript different each time it is delivered. Maybe even minor structural changes in your code.
After all, who says your JavaScript cannot be generated by a cgi script?

Re:Clarification

[Mad Merlin]

What I find more interesting, though, is that this would reveal the source code of your game to a much greater degree than, say, a Java applet does... allowing people to perhaps hook into or modify your code in order to cheat. An attempt to create anti-cheating measures would be interesting, to say the least.

Not really, you just leave the client as a glorified renderer and do all of the core game logic on the server side, like in Game!.

Re:Clarification

[clone53421]

You’d still have the possibility that players could bot.

You’re trusting that the clicks and keystrokes that the client says it received were actually performed by a player, not a bot. When the bot can directly hook into your Javascript functions and the objects in your 3D environment, it’s significantly easier to build than it would be if it had to capture screenshots and scrape the information from those. So you’d naturally want to detect and/or avoid these sort of hooks. (Randomizing names, as suggested by maxwell demon, is probably a fairly viable idea.)

Re:Clarification

[icebraining]

still wants to call it "Firefox"

If they allow it, they're fucking hypocrites, after what happened with Debian.

Re:Clarification

[Anonymous Coward]

welcome to the bullshit that is html5, dhtml, vrml & all the rest. why bother....?

meanwhile, back in the real world, flash 3d goes from strength to strength. with gpu acceleration in the pipeline (no pun intended) its a clear winner.

http://en.wikipedia.org/wiki/Papervision#Projects

Re:Clarification

[boxwood]

Maybe they should just make games that require more skills than a bot is capable of. It is really difficult to build a "bot" that can beat a good chess player, and chess doesn't have very many variables to deal with.

Also not all games are competitive. Many building games (sim city, transport tycoon, etc) are not competitive, but would be more awesome if you could cooperate with your friends on building stuff. And these types of games are very popular.

Get ready for farmville in 3d hahaha.

Re:Clarification

[the8thbit]

Gaming would require, amongst other things, browser-support for raw input devices

Fun fact: HTML6 (or what will essentially be HTML6) will have a [device] tag, that will be able to, in theory, comunicate with ANY device. (Controller, flightstick, camera, mic, HDD/SSD, rom drive, etc...)

(Replace the brackets with less than/greater than signs.)

Re:Clarification

[clone53421]

Interesting.

Lots of security issues, to be sure: I assume you’ll have to explicitly specify which devices a page can access. (Like Flash, which prompts you before accessing your webcam or microphone.)

Re:Clarification

[Xelios]

The implementation shown at CeBIT is not based on WebGL, it uses RTFact which was developed in house at the University of Saarland. However the WebGL implementation is in development right now, and will be released later on to make the XML3D format more compatible with mainstream browsers (those without RTFact built in).

Re:Clarification

[the8thbit]

Yes, I believe security was the primary reason it isn't included in HTML5. That said, I'm not sure exactly how it would function- it would be very bad if HTML was able to give a website liberal access to your HDD/SSD, cam, and mic without prior notification.

Re:Clarification

[clone53421]

it would be very bad if HTML was able to give a website liberal access to your HDD/SSD, cam, and mic without prior notification

Agreed, and the same approach that Flash takes would be an adequate solution, I think.

One thing that’s frustrated me is the fact that Javascript is completely incapable of accessing a file on your hard disk. Whereas a server-side script (PHP) has a well-defined and secure way to get the file: the file upload form element. The file upload box is already a secure element: Javascript cannot select the file. The user is completely in control of which file is chosen.

However, I’d love to see a Javascript interface to get the contents of the chosen file, exactly the way the server can, and none exists. It’s impossible. The only way to do this would be to send the file to the server and then send it back to the client via Ajax, which is an awful workaround: slow, and puts unnecessary burden on your server to do something that should be possible to do entirely client-side.

Re:Clarification

[Anonymous Coward]

http://www.youtube.com/watch?v=wKQkO7gk1Tc
video

Re:Clarification

[petermgreen]

transport tycoon
You ever played transport tycoon multiplayer, I can assure you it is competitive or at least can be played that way ;)

The only problem is there are too many dirty tricks possible (e.g. buying up huge swathes of land arround where the other player is trying to construct) so unless you have an agreement with the other player(s) to play nice it quickly devolves into shit ;)

oh great.

[Anonymous Coward]

so this means that in the near future ill have to have quad sli pci-e cards with 1tb of ram and a few extra powersupplies to render all of the popup/under/over/through ads.

but really, someone educate me... why would anyone find 3d rendering in a browser useful? its almost certainly not going to be able to compete, quality wise, with any recent high end graphics renderings (lightwave/maya, etc)--- and with modern compression schemes and encoding formats and everyone having broadband, why wouldnt someone just embed a higher quality video into their site instead of rendering 3d inside of the browser?

i cant just imagine firefox now, instead of consuming 500mb of ram playing some simple facebook games consuming 2gb loading 3d models instead of 2d sprites.

Re:oh great.

[dingen]

so this means that in the near future ill have to have quad sli pci-e cards with 1tb of ram and a few extra powersupplies to render all of the popup/under/over/through ads.

Not if you stick to Internet Explorer. Then you just need the TB of RAM to render the Flash elements sent to you as an alternative for the native XML3D/HTML5/Javascript ads.

Re:oh great.

[Joce640k]

It could be interactive, respond to mouse clicks....etc.

Re:oh great.

[quadelirus]

In the article they said they demoed a version of the wikipedia page for Venice and a user could walk around a scene of one of Venice's cathedrals in browser. That is pretty cool, but I agree it will go through the same thing that flash and midi music went through when they got added to the browser: there will be overuse everywhere for 5-10 years until web designers finally take back control from their clients and get back to designing good looking useful interfaces that only use 3D scenes when it really makes sense.

Re:oh great.

[mr crypto]

I'm in medical software development and we desperately want a good 3D browser solution. Not sure if this one will fill the bill though in terms of all of the other facets to software tools like standardization, reliability, stability, etc.

Re:oh great.

[timeOday]

why wouldnt someone just embed a higher quality video into their site instead of rendering 3d inside of the browser?

Sheesh people, it's not like we're speculating about something totally new here. There are lots of popular 3d web apps already (such as this game, which should make the point of "why render in the browser" obvious). This is just a new language for doing it.

Re:oh great.

[b4dc0d3r]

It's a nice operating system, with serious hardware requirements, it just needs a web browser.

Re:oh great.

[Anonymous Coward]

Lol, anti-M$ post. Classic dingen!

Re:oh great.

[Anonymous Coward]

Simple,If one look at the history of gaming industry in digital media it goes like this. They started with text-based(60s) followed by 2D games uses 2D sprites, then few isometric view games/ 3D perspective view(http://en.wikipedia.org/wiki/Video_games_with_isometric_graphics), then we had full 3d games. And in full 3d games also, the trend is at first 3D models has very less detail ( basic contours ) later 3D models with textures, then 3D games with Non photorealistic rendering (NPR), then now the games we are playing are photorealisitic rendering, in future we going to have real time photorealistic rendering .

Now, browser based games the history goes like this, they started with text-based followed by 2d games, then few isometric view games (farmville).. Following April 2010, with 3D graphics in browsers, we can except to have Non Photorealisitc rendering games.

Dont you get it ? We are re-writing an operating system to fit into a browser.

Re:oh great.

[Steve Max]

Sounds very futuristic, really cool. What would be the next step?

I know: 3D chat rooms! Or even better, chat rooms are sooo 20th century: let's make a 3D social network! You would create your own avatar, purchase a house, meet with friends... It would be like a second life, but online!

Removing my tongue from my cheek for a second, if that's the usage it will get, I can't see how it would succeed when VRML failed doing the same (albeit slower) 15 years ago.

Re:oh great.

[tepples]

Dont you get it ? We are re-writing an operating system to fit into a browser.

True, it's an inner platform, but advantages of an inner platform include portability and privilege separation.

Re:oh great.

[berashith]

but really, someone educate me... why would anyone find 3d rendering in a browser useful?

porn

Re:oh great.

[PitaBred]

How about just a 3D product model when you're looking at a new widget? A 3D map of where a place is? Hell, just simple games in more than two dimensions? There are lots of places where 3D can really enhance information. It doesn't have to take over the entire interface like with Second Life and how VRML was positioned.

Re:oh great.

[maxwell demon]

Now imagine rendering it in 3D!

Re:oh great.

[Hatta]

Why would you want to play a 3d game in your browser when you could run a native app instead? Why should we use the web for anything other than hypertext, when a native app will always work better? All this is doing is turning your web browser into an OS, but you've already got an OS. More layers of abstraction means more bloat and more bugs. No thanks.

Re:oh great.

[clone53421]

Because I don’t want to run a .exe that I downloaded off the web from some site that I don’t trust.

I trust my browser to put a layer of abstraction between the code and my computer that will prevent it from doing anything malicious. This is not always completely safe, but it’s safer than running the .exe.

Re:oh great.

[Anonymous Coward]

OK...thats not 3d, or at least not what this article is talking about. Thats a 2D game with simulated 3D. its called VECTOR racers. Duh???

Re:oh great.

[clone53421]

Um, yes, it is. You can have 3D “vectors” just like you can have 2D ones. And even if they’re doing a perspective projection in software, it’s still 3D.

Re:oh great.

[Hatta]

Your OS should provide privilege separation. A wise choice of language and libraries will provide portability.

No love for VRML

[BadAnalogyGuy]

We've had 3D graphics for YEARS in browsers. It is called VRML and it is a standard that has been with us since the early days of graphical browsers.

But the real question is who in their right mind will develop anything as ephemeral as a web page with this complicated technology? The time investment involved to come out with even the simplest of models is enormous. Maybe not John Pinette enormous, something smaller like Louie Anderson enormous.

Re:No love for VRML

[bbbaldie]

I have "fond" memories of vrml sites, a 14,400 modem, and a 486 slc machine. One thing about it, though, it was peppier than 1995 Java!

Re:No love for VRML

[dingen]

Just like Second Life, the 3D web is not something people actually want, but more something which makes sense to old fashioned journalists who write for old fashioned media.

They think it sounds great. Looking at pretty things instead of reading boring stuff is in their eyes the ultimate evolution of computing. That's why you keep reading this sort of stuff all the time. But it will never stick, because in reality, it's just not very useful.

Re:No love for VRML

[Joce640k]

The way I see it: If there was real use/demand for it, it would be here already...

Re:No love for VRML

[Jaydee23]

Who will use it? I'm sure that there will be somebody who develops the next big thing with it and hypes it to the stratosphere. I'm constantly amazed by what catches on on the web (Twitter? FourSquare? Go figure) Needless to say I never invest in web companies.

Re:No love for VRML

[davechen]

VRML always sucked. In particular that plugins that were supposed to do VRML sucked.

Who knows if it will be any different this time around.

As for who would like 3D graphics on the web, well I would. And there are tons of 3d models out on the internet, so throwing together a simple scene shouldn't be too difficult.

Re:No love for VRML

[Junta]

I don't see this as a technology to dominate the web experience, but rather to enable things like running a first-person-shooter or other games, or perhaps other special-purpose applications, but games would be the broader case.

Re:No love for VRML

[Conspiracy_Of_Doves]

I don't think that the point of 3D graphics in a browser is to build entire websites as 3D environments, but rather to have specialized sections of websites where they are applicable. For instance, fully rotatable views of items that you might purchase. Aside from being cumbersome to program, VRML wasn't nearly good enough to do something like that. This might be, however. I think that this technology, especially combined with the canvas tag, has the potential to do a lot of good for the web.

Re:No love for VRML

[PhilHibbs]

That's like saying "Why do we need XML when we already have HTML?" - better to ask "why not use the official successor to VRML, i.e. X3D?"

Re:No love for VRML

[vadim_t]

Just like Second Life, the 3D web is not something people actually want

The 50K people logged in right now would seem to disagree. Right now it's a fairly low activity time, should go up later. And from the inside it seems to be still getting larger.

They think it sounds great. Looking at pretty things instead of reading boring stuff is in their eyes the ultimate evolution of computing. That's why you keep reading this sort of stuff all the time. But it will never stick, because in reality, it's just not very useful.

I see it in a different way. Not everything has to be a revolution. Back when there was a lot of news about SL there was a lot of hype for sure, but there must be some use to it, since it didn't die when it stopped getting talked about so much. Some people see no point in SL, that's perfectly fine. I see no point WoW either, but that doesn't make it a failure just because it fails to appeal to every person on the planet.

I think this will be in the same way. Uses will be found for it. It won't be a revolution that will change every website everywhere. Not everybody has an espresso machine, and not everybody is going to have 3D on their website, but that doesn't mean those aren't useful things.

Re:No love for VRML

[maxwell demon]

That's like saying "Why do we need XML when we already have HTML?"

No.
Either: "Why do we need XML when we already have SGML?"
Or: "Why do we need XHTML when we already have HTML?"

Re:No love for VRML

[twoshortplanks]

Google Earth in your browser

Re:No love for VRML

[Anonymous Coward]

Two words:

pr0n!

Re:No love for VRML

[Anonymous Coward]

The way I see it: If there was real use/demand for it, it would be here already...

My instinct is to agree, because I never happened to have ever seen or imagined any way that 3D in a browser would be useful.

However, when you put it the way you just did, I get these flashbacks of infamous old quotations about how every good idea has already been invented, there's a world market for 6 computers, etc.

I think the way this should play out, is that someone needs to think of an application where 3D is useful. And then it'll be time to think about how to put 3D into the browser. Don't count out the first thing from ever happening, but it's certainly correct to say that no one has done it (and told the world) yet.

Re:No love for VRML

[PitaBred]

Why does the whole web have to be 3D or not? Why can't we just make the parts of it 3D that make sense to make 3D? It's not a hard damn concept.

Re:No love for VRML

[TheRealMindChild]

Like flying cars?

Re:No love for VRML

[Adm.Wiggin]

Goodbye FlashBlock, hello Block3D.

Re:No love for VRML

[Anonymous Coward]

The 50K people logged in [secondlife.com] right now would seem to disagree.

Nah, the real humans amongst that number are too busy engaged in "adult" activities to disagree right now. The rest are bots.

Re:No love for VRML

[clone53421]

It’s trivially easy. Just add ##canvas rules to Adblock Plus.

Less trivially easy is then making exception for sites that use the canvas element for legitimate content, but it should still be doable.

Re:No love for VRML

[Adm.Wiggin]

Indeed, so a "click-to-3D" solution like FlashBlock would be great. In fact, FlashBlock's approach could possibly work, although I'm not sure if replacing the canvas element would make the JavaScript blow up. That'd be fun to test. :)

Re:No love for VRML

[clone53421]

I'm not sure if replacing the canvas element would make the JavaScript blow up. That'd be fun to test.

Depends on the Javascript. ;)

I’m currently working on (well, on and off) a somewhat large personal project (currently at 100+ KB of Javascript and counting)... and my code does this:

canvas = document.getElementById("canvas");
try {
    ctx = canvas.getContext("2d");
} catch (e) { // using internet explorer, hack something that kinda still works
}

So basically, if you replaced my canvas element, my code would (more or less gracefully) hopefully fall back to an ugly contrived mode that mostly works. (If the canvas variable comes back null, it doesn’t work. As long as something exists with that ID, it works.)

If you want a sample of what I’m doing, here’s a snippet from multiply(n1, n2) (I’m using arrays of 32-bit integers to hold larger data... sig1 and sig2 are 56-bit integers broken into 24 and 32 bits.):

// multiply sig1h:sig1l x sig2h:sig2l (result has 112 significant bits in a 128-bit space)... just trust me, it works
var p1, p2, p3, c, sig_h, sig_l;
p1 = bitwise_mul(sig1l, sig2l);
[p2, c] = bitwise_add(bitwise_mul(sig1h, sig2l), bitwise_mul(sig1l, sig2h));
p3 = bitwise_mul(sig1h, sig2h);
[p1, c] = bitwise_add([c, p2[0], p1[0], p1[1]], [p3[0], p3[1], p2[1], 0]);
// shift the result to the right 56 times and round based on the 55th bit shifted out (result has 56 significant bits in a 64-bit space)
[[sig_h, sig_l], c] = bitwise_add([(p1[0] >> 24), (p1[1] >> 8)], (p1[2] >> 7) & 1)
exponent += 56;

bitwise_add adds two 32-bit unsigned numbers and returns Array(sum, carry). If you give it two arrays, it adds the corresponding elements from last to first and propagates the carry, returning Array(Array(sum0, ... sumn), carry).

bitwise_mul multiplies two 32-bit unsigned numbers and returns Array(product0, product1).

Fun.

Short answer

[Anonymous Coward]

The browser will be released within a few weeks, the researchers say, and they are checking with the Mozilla Foundation about whether they can call it Firefox.

No.

CPU hungry

[sshock]

Do I really want my CPU to overload while navigating the web?

Re:CPU hungry

[quadelirus]

RT Fact is a GPU ray tracer, I think.

Re:CPU hungry

[maxwell demon]

Expect to see amazing new exploits using the GPU.

Re:CPU hungry

[Anonymous Coward]

he is using an intel gpu. its overloading his cpu because intel can't build gpus to save their lives

Re:CPU hungry

[PerfectionLost]

My CPU already overloads from browsing the web. I'm begining to think I need to ditch firefox anyways...

Re:CPU hungry

[quadelirus]

That may or may not be true. On the one hand the GPU doesn't have an OS so there isn't any protection of memory from any program which seems like exploitation might be pretty trivial, but GPU code doesn't have access to the disk drives or to CPU memory, so it would be hard to craft an exploit that did anything other than crash the machine. IMO.

Re:CPU hungry

[Anonymous Coward]

Meh... Don't expect it to ever be implemented in IE.

Re:CPU hungry

[anss123]

Get Flashblock. I use it to good effect on my netbook.

Two word

[Anonymous Coward]

Porn ography.

Re:Two word

[Anonymous Coward]

Oh. Yeah. OK, now I get it. Up until this moment I couldn't see any possible reason why anyone would want to do this other than to burn some VC money.

Now I realize this is the next big thing.

However it won't take off until some lame non-porn apps use it somehow, so we can justify having it installed on our machines. Once that happens, people will use it like crazy monkeys.

The parent is marked "funny", but I'm completely serious.

Re:Two word

[Shotgun]

Oh. Yeah. OK, now I get it. Up until this moment I couldn't see any possible reason why anyone would want to do this other than to burn some VC money.

Now I realize this is the next big thing.

However it won't take off until some lame non-porn apps use it somehow, so we can justify having it installed on our machines. Once that happens, people will use it like fornicating monkeys.

The parent is marked "funny", but I'm completely serious.

There fixed that for you.

My wife is a personal trainer. She has her clients do this exercise called "fornicating monkey". Do a deep knee bend, and then grab the back of your ankles. Now repeatedly squat lower.

No. I don't have any idea what this has to do with the subject at hand. Just thought it was funny.

Re:Two word

[FragHARD]

Porn 3Dography. would be a better term ;)

Calling it Firefox

[rwv]

The name of the Debian version of Firefox is Iceweasel. Based on that, I'm assuming that the Mozilla Corporation is going to exercise their trademark and kindly request that these researchers think of a better name for their fork of Firefox that incorporates XML3D.

If successful, it wouldn't surprise me to see the Mozilla folks include this feature in a future release of Firefox.

Re:Calling it Firefox

[phooka.de]

If successful, it wouldn't surprise me to see the Mozilla folks include this feature in a future release of Firefox.

Heaven forbid, please no!

We don't need a rendering engine for every arcane formalt ever developed incorparated into a browser that's deployed on millions of desktops. Just remember, each supported protocol adds new complexety, new errors and with this new secutiry-issues that'll lead to exploits, bad press, compromised machines and painful bugfixing.

Stuff like this should never be part of the browser, it should be an addon.

Re:Calling it Firefox

[boxwood]

Actually I don't think Mozilla actually requested debian to change the name, debian changed it themselves because firefox was trademarked and debian want to be free of intellectual property issue.

Mozilla trademarked firefox because if they didn't, anyone could redistribute it filled with malware and destroy the firefox brand. And there wouldn't be anything mozilla could do about it since FF is open source. With the trademark, if someone tries to distribute a crappy version, mozilla can make them change the name.

Whether Mozilla allows them to use the firefox name largely depends on the quality of their changes. if this add-on makes firefox buggy or slow then they'll ask them to change the name.

Re:Calling it Firefox

[Muerte2]

2001 called, they want their opinion back! People said the same thing about audio/video back in the day, now look at HTML5.

I'm *NOT* advocating that we just add every format into a browser, but should a good format emerge I don't see any reason not to add it to a browser. Better in the core browser than in an addon: speed and reliability are always better in the core browser.

Re:Calling it Firefox

[petermgreen]

Actually I don't think Mozilla actually requested debian to change the name, debian changed it themselves because firefox was trademarked and debian want to be free of intellectual property issue.
Then you didn't follow the issue.

Debian seem to think trademarks are ok but copyrighted logos under a non-free license aren't. I can understand where this position comes from though i'm not sure I agree with it.

Debian claim they were given permission by mozilla a long time ago to use the firefox name on thier unofficial builds of firefox despite the fact they didn't use the branded logos (the firefox source has always included a set of "unbranded" logos, you see them on nightlies and alpha releases among other places).

Later someone from mozilla came along (seems they got new people in charge of such things) and basically said that if debian wanted to use the firefox name they had to both accept the non-free logo and give mozilla veto power over any patches debian applied. Neither of these conditions was acceptable to debian.

They could have done what the nightlies and alphas do and used the release codename as the name of the browser but a browser whose name changed with every release would be pretty confusing for users. So they switched to using thier own name. Soon afterwards they also switched from using the mozilla supplied unbranded logos to using thier own logos (though I don't think this was caused by a licensing issue but just because they looked better than the unbranded mozilla logo and also fit with the new name).

Call it Cerberus

[rwv]

I'd propose "Cerberus" as the name for their forked version of Firefox that has XML3D rendering capability. Cerberus is is three-headed dog that guards the gates of Hades. After all, Hades has lots of fire and the connection between foxes and dogs is tangible (they are both canines, AFAIK).

Re:Call it Cerberus

[maxwell demon]

Especially given that they have a link "coding hell" in the navigation column ...

Re:Call it Cerberus

[Asclepius99]

I'm voting for PerspectivePanda.

Re:Call it Cerberus

[lena_10326]

How about GraphGiraffe? PixelPup? RenderRabbit? TraceTripe?

Re:Call it Cerberus

[Anonymous Coward]

I am so down with GraphGiraffe.

Re:Call it Cerberus

[mfnickster]

The name "Cerberus" is already in use for an FTP server: http://www.cerberusftp.com/

Dunno if they have a trademark or service mark on it, though.

It also might get confused with Kerberos by people who don't know better! :)

Re:Call it Cerberus

[Anonymous Coward]

Firefox is not a fox though..

Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

We realized that 3D graphics in the browser were stupid and useless back in 1995, when the VRML hype was much like the HTML5 hype is today.

It's one of those things that sounds fantastic, but in reality there are very few useful applications.

This is just the 15-20 year cycle we typically see with computing technology. Many of the Firefox developers were born after 1990, so they aren't even aware of the browser experiments and failures before about 2005. Not knowing history, they're doomed to repeat the mistakes of the true innovators.

Re:Doesn't matter. 3D in the browser is stupid.

[Junta]

WebGL/RT/HTML5 are not fundamentally stupid. VRML hype mistakenly centered around a 3D navigation model for most of the web replacing 2D textual interaction with some image content, which was stupid.

However, richer multimedia content is a fact of life now with increased bandwidth. If it were not, then flash wouldn't persist (overuse of flash was a fad that has abated a bit in favor of javascript/css mechanisms, but flash persists for video and games without viable alternatives). Various video streaming sites that are relegated to flash today for games and videos would be freed from Adobe's whims as the embedded video, canvas, and 3d capabilities are expressed in industry standard terms.

Re:Doesn't matter. 3D in the browser is stupid.

[JBMcB]

We realized that 3D graphics in the browser were stupid and useless back in 1995, when the VRML hype was much like the HTML5 hype is today.

There are a few differences.

VRML was never really an industry standard, it evolved from an SGI project and was adopted by a few other companies. There were competing technologies that seemed better, but were mostly closed. In any case, they required browser plugins that were large, clunky, and crashy.

At the height of VRML's popularity, there really weren't any standards for desktop 3D acceleration. Getting decent performance from a VRML browser required a pretty fast machine, and the graphics were very crude even then.

Now we have an industry standard backed by the group in charge of HTML, ridiculously fast 3d hardware on even low end desktops, and, with the modded FireFox and Webkit backends, integration with the codebase.

This might end up working.

Re:Doesn't matter. 3D in the browser is stupid.

[BRock97]

We realized that 3D graphics in the browser were stupid and useless back in 1995...

...and slow! I was there when VRML was landing (just finished high school) and all I saw it used for were virtual rooms were avatars would talk in a 3D IRC like environment. Only big problem back then was we didn't have 3D acceleration and the interface was clunky and painfully slideshow like in speed.

As for uses, I could think of a few and have already started coding them. Instead of loading a PNG or GIF, it is pretty nice to be able to download a float array, be able to display it, and allow a user to interrogate it. Giving a user that kind of capability in the browser while not requiring them to download an application or a browser plugin is pretty darn nice. And while not mainstream, scientific fields could greatly benefit from something like that.

Re:Doesn't matter. 3D in the browser is stupid.

[AmigaMMC]

Agreed. I'd be much happier if they built a Firefox that could perform better with Flash websites. I have to use some pretty Flash intensive websites and my Firefox crashes too often. I don't have that problem with Chrome. Have been using FF since pre-1.0 version but I feel like it's getting slower and slower no matter what they claim and 2.x didn't crash as much as 3.x does (although 3.x seems to be a bit better, just a bit, at handling memory use).

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

How is HTML5 any more of a "standard" that VRML was? It has basically came out of Google, with mixed adoption by a few other companies/organizations (Apple, Opera, Mozilla).

Microsoft, who like it or not still produces the browsers that represent over 75% of the browser market, really hasn't shown any interest in adopting it. Maybe the W3C will bless HTML5, but that doesn't change the fact that it'll work for only 25% of Web users.

Regardless, what you've said still doesn't change the fact that this is generally a useless technology. The best we'll get out of it are some games, and these will run like shit anyways because they'll be written using JavaScript. Maybe another virtual world or two (which we already have with Second Life and WoW). Most video operations don't benefit at all from 3D graphics; they inherently need 2D acceleration.

Like we found in the 1990s with VRML, 3D rendering in the browser is of limited use. It's a novelty that's bound to resurface every 15 years or so, as people forget the experiments and failures of the previous generation of developers.

Re:Doesn't matter. 3D in the browser is stupid.

[tepples]

flash persists for video and games without viable alternatives).

The viable alternative is for video is HTML 5 video, and the viable alternative for games is JavaScript and CSS sprites. Or what am I missing?

Re:Doesn't matter. 3D in the browser is stupid.

[clone53421]

Or what am I missing?

3D games.

CSS sprites are fine... for sprite-based games.

Re:Doesn't matter. 3D in the browser is stupid.

[dotHectate]

Honestly I think that there is only one possible benefit to this. Describing the source (via XML3D or Flash or some form of a vector format image) has the potential to reduce the size of images. Need a 1600x1200 image of Earth? That's 46,080,000 bits of data required to represent it. The same thing generated in 3D could be as simple as the number of characters in this comment (sphere plus texture). Even with the texture file needed it's still more efficient. But I don't want VRML again either.

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

For 3D we have WebGL
http://en.wikipedia.org/wiki/WebGL

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

*cough*
http://www.web3d.org/x3d/specifications/vrml/

still it's the idea that the web should deliver *also* the presentation of the documents that is basically flawed, the web should only do it's job of marking up text and other resources in a way that makes easy to the browser to display them while still leaving the freedom to the user on how to process the content, without forcing the presentation style. That's why flash is so bad, in the end, and the market space for web plugins is narrowing each day with the fifth cometh of html.

yeah, 3d plugin could push the forcing of the presentation up so far, but in the end there is no need to deliver 3d data to the client, so there is no future for this kind of effort. If we wanted to watch 3d object on the web, vrml would have been in wide use by now. However, I can potentially see how this could be used for fully 3d entertainment, but real 3d films, films in which you can navigate trough the scenes, won't be coming for a loooong time and probably never. For everything else, youtube + html5 already have support for 3d display, so meh.

Re:Doesn't matter. 3D in the browser is stupid.

[sosume]

> Many of the Firefox developers were born after 1990

Are you saying that one of the currently most popular (and relied on) pieces of software
is being written and maintained by a bunch of high-school kids? I call shenanigans.

Re:Doesn't matter. 3D in the browser is stupid.

[clone53421]

For 3D, maybe you have WebGL...

Component returned failure code: 0x80070057 (NS_ERROR_ILLEGAL_VALUE) [nsIDOMHTMLCanvasElement.getContext]

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

well what else matters but his uniquely functional configuration, you insensitive clod?

how can you explain that to the 98% of web users that are very happy with flash and the steady advancements its made - as3, 3d, video, rias, multiplayer browser games....?

its a big ask!!!!

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

This might end up working.

Except ... instead of specifying a rasterization and shading policy which is supported by all these ridiculously fast 3D cards, they are specifying ray-tracing.

Anyway, did VRML have a "height of popularity"? Even back then it paled in speed and looks to even the most poorly-coded 3D games. As, it seems, will this new approach.

Re:Doesn't matter. 3D in the browser is stupid.

[JBMcB]

How is HTML5 any more of a "standard" that VRML was?

Who cares? X3D is an ISO standard developed in an open process. http://www.web3d.org/x3d/

Regardless, what you've said still doesn't change the fact that this is generally a useless technology.

That's what people said about 3d desktops, now every major OS has some 3d elements in the UI. I don't see an immersive 3d environment in a browser, but I do see 3d elements being useful for the display of some types of data, and navigating through certain types of datasets.

Most video operations don't benefit at all from 3D graphics; they inherently need 2D acceleration.

If you can put a video on a 3d surface, you can pipe it through the GPU to speed up deinterlacing, interpolation, scaling, etc... All modern video acceleration schemes use the GPU anyways, having a separate processor do all the heavy video processing doesn't make sense these days.

It's a novelty that's bound to resurface every 15 years or so, as people forget the experiments and failures of the previous generation of developers.

I don't think the X3D working group has completely forgotten the past failures of in-browser 3d, which is why X3D is not VRML, though it maintains backwards compatibility.

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

Need a 1600x1200 image of Earth? That's 46,080,000 bits of data required to represent it. The same thing generated in 3D could be as simple as the number of characters in this comment (sphere plus texture). Even with the texture file needed it's still more efficient.

Uh, then what's the point of having a texture map less than 1600x1200 ? You will get less detail than the image you're trying to avoid downloading.

Re:Doesn't matter. 3D in the browser is stupid.

[clone53421]

Water... land... mountains... rainforests... deserts...

Texture maps applied to vector shapes. You don’t need a full-sized texture; they tile.

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

CSS sprites? Really? Yeah, that'll perform...

Try Canvas.

Re:Doesn't matter. 3D in the browser is stupid.

[h3x87]

Flash problem. But 2 days ago Firefox 3.7 Alpha 2 was released with out-of-process plugins. Flash can no longer crash FF.

Re:Doesn't matter. 3D in the browser is stupid.

[billcopc]

Didn't VRML look and behave kind of like Second Life ? I still think both are equally stupid, but obviously some people thought it was worth taking semi-seriously. SL brought much better, dumbed-down tools to build this 3D web, but it's more or less the same thing.

Re:Doesn't matter. 3D in the browser is stupid.

[tepples]

Anonymous Coward wrote:

Try Canvas.

I did, and 60 percent of the time, the browser just rendered the "Get Firefox, SeaMonkey, Safari, Chrome, or Opera" text inside the Canvas element. But for those 40 percent of PC web browsers that aren't IE <= 8, allow me to rephrase:

If IE <= 8 market share ever declines, the viable alternative for games is JavaScript and 2D Canvas. Or what am I missing?

Re:Doesn't matter. 3D in the browser is stupid.

[clone53421]

I did, and 60 percent of the time, the browser just rendered the "Get Firefox, SeaMonkey, Safari, Chrome, or Opera" text inside the Canvas element.

That’s odd... I guess you couldn’t take a hint? :p

If IE <= 8 market share ever declines, the viable alternative for games is JavaScript and 2D (3D) Canvas.

For 2D (3D) games, yes... and the IE <= 8 market share will decline when viable games require Javascript and 2D (3D) canvas to play.

Or when porn requires it. More likely when porn requires it...

Re:Doesn't matter. 3D in the browser is stupid.

[ivucica]

Maybe because Flash, as a plugin, will always require extra time to load itself? Have we learned nothing from the Java applet lesson?

Re:Doesn't matter. 3D in the browser is stupid.

[Thinboy00]

Maybe because Flash, as a plugin, will always require extra time to load itself? Have we learned nothing from the Java applet lesson?

Re Java: Funny, Sun went to the trouble of competently implementing cross-platform support, unlike Adobe... (and don't tell me that it's not Sun anymore: Oracle didn't invent Java, they bought/are buying it)

Re:Doesn't matter. 3D in the browser is stupid.

[ivucica]

I like the idea of Java, and I like Sun. (The old one, don't know what Oracle will do.) And I don't like Flash.

But having Java block my entire browser for 10-20 seconds because some jerk put an applet on his page is completely unacceptable. Being thankful for crossplatform support has nothing to do with dissatisfaction with 10-20 sec blockade of my browser to display a small button that says "Click me".

I haven't had a random encounter with an applet for a few years now, and "praiz teh lord" for that. But that also makes me unaware if there have been any changes in that field. If the Java applets now appear instantly (at least their login screen) and don't block the entire browser while the plugin (not the applet!) is loading, well, that might change my opinion.

If I dislike Flash blocking the browser for 2-3 sec on my various machines and OSes, why should I praise Sun just because it "went to the trouble of competently [disputable] implementing cross-platform support". I like Sun, and I like the idea of Java. I like the benchmarks which say that Java is hyper-fast. But...

Oh, and I also slightly dislike the fact that many Java apps don't feel native anywhere, either because of the theme or due to ... just ... some feeling. And don't get me started on my experiences with J2ME (a small hello-world piece of code behaves completely differently on 4 test devices). Let's leave it at that.

Re:Doesn't matter. 3D in the browser is stupid.

[Junta]

I mentioned HTML5 as one of the technologies that was meaningful to get out of this (parent poster had described HTML5 as VRML-like hype along with WebGL). Unfortunately, it will be a tad longer before HTML5 is viable for a large audience website can adopt exclusively (without flash) given the subset of browsers that currently implement HTML5 as well as how slow the general populace is to update browsers.

HTML5 also provides powerful primitives to build 2D games, but 3D is missing. I personally appreciate the sensibilities of 2D games, but there are some experiences that are worthwhile that come with 3D.

Re:Doesn't matter. 3D in the browser is stupid.

[petermgreen]

Presumablly this also means they won't be able to blame flash whenever a flash user has a firefox problem.......

Re:Doesn't matter. 3D in the browser is stupid.

[petermgreen]

The viable alternative is for video is HTML 5 video
It's not really viable to completely dump flash in favour of HTML5 video at this point due to the large number of users whose browsers don't support html5 video.

And afaict sites that want to support firefox's implementation of html5 video AND support flash (for all those IE users out there) need to encode and store all thier video TWICE.

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

Oh, sorry... I assumed he meant an *accurate* 1600x1200 image of Earth.

Re:Doesn't matter. 3D in the browser is stupid.

[clone53421]

It would be accurate, to a certain degree... it’s one way of lossily compressing the image. The surface of the earth, viewed from space, isn’t really that varied, and it could be compressed quite a bit using appropriate methods. It’s mostly texture, with various shades of blue, green, and brown.

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

> It would be accurate, to a certain degree... it's one way of lossily compressing the image.

But there really is no way to make it as accurate as the 1600x1200 photo and make it smaller - any compression you can apply to a texture map, you can apply to the original photo as well.

There is no bandwidth savings, all you gain is the ability to view it in 3D. You can't cheat information entropy!

Re:Doesn't matter. 3D in the browser is stupid.

[clone53421]

any compression you can apply to a texture map, you can apply to the original photo as well

No; you’d be adding regular JPEG compression.

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

> No; you'd be adding regular JPEG compression.

And if you use lossy compression, you will lose detail - thus making it not as accurate as the photo. You can't make up in redundancy what you lose in information!

Re:Doesn't matter. 3D in the browser is stupid.

[clone53421]

Different forms of lossy encoding can encode different things better. Depending on what you’re encoding, one form of lossy encoding might give a huge reduction in quality where another would give very little reduction.

The only form of compression you can use on a JPEG image is regular JPEG compression. Building a mesh of varying textures may well give a quality, for its small file size, that is far above the blurry, low-detail JPEG that you’d get if you compressed it to the same size.

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

You're getting off track. The point is that what dotHectate is suggesting is not possible.

"Even with the texture file needed it's still more efficient" - no, it is not, because you still have to transfer all the bits (compressed or not) to represent that level of detail. Tiling textures is exploiting redundancy, which is what you do when you're compressing an image anyway.

If he's talking about something like Google Earth, then the "savings" in bandwidth is illusory because you're only downloading the high-res image data for a small section you're focusing on. If you were to download the entire thing it would take far more than the 46,080,000 bits needed to draw the 1600x1200 representation.

Re:Doesn't matter. 3D in the browser is stupid.

[clone53421]

"Even with the texture file needed it's still more efficient" - no, it is not, because you still have to transfer all the bits (compressed or not) to represent that level of detail. Tiling textures is exploiting redundancy, which is what you do when you're compressing an image anyway.

My point is, you might be able to utilize a more efficient compression than JPEG.

Re:Doesn't matter. 3D in the browser is stupid.

[Anonymous Coward]

> My point is, you might be able to utilize a more efficient compression than JPEG.

Of course you can, but you can do that with the original 1600x1200 image as well.

There is nothing inherently more efficient about representing something in 3D, in fact it requires MORE data to be downloaded because of geometry, UV mappings, bump maps etc.

dotHectate's suggestion is equivalent to saying "Need a 1600x1200 image of Earth? ...The same thing generated in 2D could be as simple as the number of characters in this comment (rectangle plus texture)."

Mapping a texture map onto a figure doesn't gain you anything, only compression does--and you can do that just as well in 2D as in 3D. Is this so hard to understand?

No thanks

[__aardcx5948]

Not interested. What I want: rendering accelerated by the graphics card, in some way, better than (if) it is now. No more slow scrolling pages full of graphics.

Ahh, the day that comes...

Re:No thanks

[Briareos]

Ahh, the day that comes...

Believe it or not, it's already landed on trunk - at least for Firefox running on Windows 7.

np: Autechre.ws Webcast (02.03.2010)

Re:No thanks

[__aardcx5948]

Indeed, but what about us who run real OSes? ;)

Might be worth trying out on the W7 machine though!

Re:No thanks

[Rutulian]

Looks like they are working on a Cairo backend as well.

Re:No thanks

[BZ]

Might help if the "real OSes" provided more useful 3d apis to build on instead of having to do things from scratch, no? ;)

Re:No thanks

[Anonymous Coward]

*headshot*

Re:No thanks

[chris-chittleborough]

Briareos is talking about Direct2D.

Firefox does not support RT Fact (RT = Ray Trace, BTW) or XML3D.

The HTML5 designers, following Mozilla's lead, are specifyng WebGL, which is a JavaScript API for OpenGL ES 2.0, the current version of OpenGL for Embeeded Systems.

Gallery?

[master_p]

I've searched the web but I can't find any picture of an image rendered with RT Fact. The news are repeated in various news sites, as always, but none of them has a single image of the 3d engine output.

Re:Gallery?

[cheesybagel]

Try the PC World article. FWIW RT Fact is a 3D ray tracing engine. I guess they integrated it into the browser somehow. But saying it does not require a plugin? Eh seems strange to me. I thought RT Fact was written in C++ with x86 assembler intrinsics.

Re:Gallery?

[maxwell demon]

Ok, it's a video, not an image ...
Found, of course, with Google.
Searching directly on YouTube gives two further results.

Re:Gallery?

[Rutulian]

Ok, is it just me, or does that demo look really slow. Are they using any hardware acceleration for this stuff?

Re:Gallery?

[Anonymous Coward]

> Ok, is it just me, or does that demo look really slow. Are they using any hardware acceleration for this stuff?

Um...I'm not aware of any commercial graphics cards that do ray-tracing acceleration. That's still in the experimental stage.

Re:Gallery?

[Rutulian]

No, just the 3d rendering acceleration. I know POV-Ray runs a lot faster on my computer than this demo appears to run.

Tech for the future

[muyla]

Maybe this tech will be big when 3d monitors are out... just imagine the pop ups really poping out of your screen :)

Re:Tech for the future

[Datamonstar]

Is that you, Master Shake?

Re:Tech for the future

[bcmm]

Maybe this tech will be big when 3d monitors are out... just imagine the pop ups really poping out of your screen :)

Just imagine the Goatse...

Ray tracing vs. Rasterization

[quadelirus]

Why would they choose real time ray tracing over rasterization methods? Rasterization is still much faster and you can achieve all kinds of ray tracing like effects if you want to.

Re:Ray tracing vs. Rasterization

[Anonymous Coward]

Because to rasterise beforehand you need to know the attributes (dimensions, resolution, color gammut) of the output device. These vary considerably between connecting machines. Precomputing the rasterisation is faster, but is of less quality. It depends what you use the web for if speed or quality is preferred.

Re:Ray tracing vs. Rasterization

[tepples]

I think quadelirus was talking about doing the rasterization on the client using triangle-mesh-based methods like WebGL, speculating that WebGL would be faster than even GPU-accelerated ray tracing.

Re:Ray tracing vs. Rasterization

[Animaether]

You posted as anon - I do hope you bookmarked and come back to check for replies, though.

I think you misunderstood the meaning of the word 'rasterisation' here.

In overly simplistic terms... rasterisation is what your gaming card / console does whenever you play a game of Modern Warfare 2, HALO, etc. This is unrelated to pre-rendered content.
http://en.wikipedia.org/wiki/Rasterisation

Re:Ray tracing vs. Rasterization

[quadelirus]

Yes, Rasterization vs. Ray Tracing presents a trade-off between speed and quality. Obviously truly ray traced scenes with nicely modeled materials and light sources will produce better images than rasterization. In this case, however, we are not talking strictly about ray tracing vs. rasterization. We are talking about real time ray tracing vs. rasterization. Real time, meaning that we are assuming a lower bound of, say, 15 fps (this is pretty poor, but just to pick one). At that speed (or faster) rasterization beats real time ray tracers in terms of quality, especially when it comes to renderers that can't be tuned for a specific app (i.e. a browser based viewer that needs to be able to render a general, user created scene vs. a game that knows what its scenes will look like and can tune things so that the features used are fast in deference to other features).

On top of that you can do some tricks with the rasterization to fake lots of ray traced effects like caustics, global illumination, glossy-glossy reflections, etc.

Maybe, the thought, just to give a potential answer to my question above, is that the 3D window in the browser will be resolution limited enough (say 400x400 pixels) that ray tracing will begin to make more sense. But in this sort of case the differences between rasterization with some of the tricks built in and ray tracing will probably not be noticable.

Anyway, I still want the question answered. As far as I know (and I'm a graphics researcher--though a relatively new one, so I'm still learning) real time ray tracing is really not nearly as good as rasterization in terms of quality for a real time frame rate on reasonably fast consumer hardware. If I'm wrong, please post links to academic papers detailing the high quality fast ray tracers, because I'd like to add them to my "to read" queue.

Re:Ray tracing vs. Rasterization

[Rockoon]

Because rasterization 'with all kinds of ray tracing like effects' is a bitch.

Shadows alone are extremely complicated in a rasterizer, with special cases for self shadowing, for when the camera is within a shadow or not, when something reflective is being rendered, when something refractive is being rendered, and so on and on.

Essentially nobody has EVER made general purpose rasterizer that flawlessly supports shadows in concert with all the other 'ray tracing like effects' and it is likely that nobody ever will, because the problem is more than just non-trivial. There is always another edge case. Games get away with it because they impose restrictions (explicit or implicit) which avoid most of the edge cases that the renderer can't handle.

Even highly developed engines such as Valve's Source Engine still have problems with incorrect shadowing of their own (non-arbitrary) content, and thats in scenes without reflections or refractions complicating the problem. Now factor in that a renderer such as this is supposed to render arbitrary content, and you see the main problem with rasterizers as general purpose photo-realistic renderers is that nobody can do it, in spite of decades of effort.

The reason to use a raytracer is because all the photo-realistic behaviors of light fall right out of it by definition. Adding yet another behavior of light is simple. Shadows, reflection, refraction, global illumination.. its all SIMPLE (tho certainly less efficient.) The problematic "quality" issues raytracers have are trivial in comparison, with the hardest probably being the inherent aliasing of sub-pixel features.

Re:Ray tracing vs. Rasterization

[quadelirus]

I understand that ray tracers are better than rasterizers in terms of simply modeling real light transport, but I don't think that realtime ray tracers even come close (I could be very wrong, and if so I'd like paper refs because I'm studying this area at the moment). Some error in the rasterizer is ok as long as it looks good, it doesn't have to be physically plausible. As soon as you start adding ray traced inter-reflection, refraction, global illumination and such you are well outside the realm of a realtime ray tracer (I believe, I'm willing to bet I'm wrong, and I believe that one day we will have realtime ray tracers, I just don't think we're there yet). I don't see how this RT Fact can incorporate interesting effects and still be realtime. If this is the case, it seems we might as well just use a rasterizer.

Re:Ray tracing vs. Rasterization

[clone53421]

Anyway, I still want the question answered. As far as I know (and I'm a graphics researcher--though a relatively new one, so I'm still learning) real time ray tracing is really not nearly as good as rasterization in terms of quality for a real time frame rate on reasonably fast consumer hardware.

From TFA:

There are two ways to generate a three-dimensional image for a two-dimensional screen: rasterization and real-time ray tracing. What the researchers have done is developed faster software for performing real-time ray tracing, which is also aided by today’s faster processors, Slusallek said.

They’ve integrated that real-time ray tracing technology, called RT Fact, into Firefox and Webkit, the rendering engine for browsers such as Safari and Chrome, Slusallek said.

Re:Ray tracing vs. Rasterization

[zeroRenegade]

Hey Man,

RTFact is based on a research paper from the University of Saarland. The paper describes the implementation of a generic real time ray tracing framework with source code in C++. The goal or objective is an interactive real-time ray tracer.

From the different implementations I researched (Manta from MIT, OpenRT, Arauna, RTSL, plus many more), RTFact is by far the most legitimate implementation. There are a million papers out there on interactive ray tracing, but only a few really take into consideration some of the major problems. I played around with OpenRT, but the amount of artifacts and aliasing really take away from the interactive experience. I have not played around with it recently, so maybe they now have an improved adaptive anti-aliasing solution, so my comments may be outdated.

The base code for RTFact is supposed go open source, but I have been waiting around for a long time without even a remote tidbit of information until this post. They actually even went backwards as they removed the paper from public distribution. Whenever it does go open source, it will be posted here.
http://www.rtfact.org/

Now the generic ray tracing api/framework is RTFact, but from the sounds of the article posted above, they are actually integrating the scene graph RTSG into WebKit, which has also been developed by the university of Saarland. This is only speculation and I could be completely wrong.

If you want some info on RTFact, check out:
http://tiny.cc/gHMrW

For info on RTSG, check out:
http://tiny.cc/3ezO8

If you want the original paper, the only link I could find from Google seems to be broken, but it may be due to the servers being overloaded by downloads after the announcement. I have the paper somewhere here on one of my drives, but it would take me a while to find, so if you want me to spend the time looking for it, you would need to give me some incentive by proving to me that you are in fact doing research.

In regards to your question, without a doubt, rasterization will eventually be replaced by ray tracing. Just look at Pixars evolution into photorealism. When the frame rates improve with better hardware for the general public, the framework will begin to be used in game engines, and not just scene graphs. The reasoning is aesthetic as much as it is technical. Ray tracing is truer to the physics of light than rasterization, so even though you can "fake" effects, the graphics will always be more appealing being rendered backwards than forwards. I do numerical simulation (with a background in CS from UW, where my heavy graphics knowledge comes from, plus a few years in real-time simulaton), and the true physics of the problem always gives a better solution than assumptions, approximations, correlations, and correction factors. It is a comment that my prof continually reiterates. For example, caustics will never look as good rastered as they do ray traced, since the ray tracer will map the full motion of the photons.

I could go on for hours, but I will leave it at this....

Re:Ray tracing vs. Rasterization

[cheesybagel]

Actually you can do raytracing with shadows in real time. On some scenes you can do reflections in real time as well. Refraction slows things down, as when you have transparency you cannot do a lot of optimizations in your code. There are some fake so called instant global illumination methods and ambient occlusion that are nearly real time, not very realistic, but ok.

Re:Ray tracing vs. Rasterization

[quadelirus]

I read that, actually, thanks. Still doesn't tell me what its quality is compared to good rasterization techniques. The next post after yours is much more helpful, however.

Re:Ray tracing vs. Rasterization

[quadelirus]

Do you have the paper title and/or do you know what journal or conference it was published in?

Also, the big thing that makes me somewhat skeptical about realtime ray tracing is that it doesn't lend itself to good GPU algorithms since it is inherently divergent. Anyway, thanks for the good info.

I definitely agree that ray tracing allows us to model light transport physics in ways that will never be available in rasterization.

Of course there are other methods, like light fields...

Re:Ray tracing vs. Rasterization

[Rockoon]

Why do you impose the realtime requirement on it? Watching the video, it looked like they were getting about 1 fps or so, not anything resembling the so called "realtime" or "interactive" level that is required by games.

In short, soft confusing game requirements with web requirements.

Re:Ray tracing vs. Rasterization

[quadelirus]

First, I impose the real time requirement because the announcement says that it is a real time ray tracer, which means they are claiming real time. If what you say is true, and it is only getting 1fps then it isn't even really interactive time (which is slower than real time). Second, why would you want an embedded ray tracer that wasn't real time? Wouldn't it be better to simply render the images offline somewhere else and then serve the rendered images? What advantage do you get from rendering on the fly if you can't actually interact with the scene in a meaningful way in real time?

Re:Ray tracing vs. Rasterization

[clone53421]

Yeah, TFA was light on details about the quality and speed. However, the video in the PC World article looked pretty decent, although the framerate was quite choppy — nowhere near what you would want from a game, but more than adequate for a walkthrough.

But basically your question was “I thought ray-tracing was too slow to do in real-time”, and the answer as given in TFA was “they’ve designed a faster ray-tracing algorithm”.

Other poster’s comment was pretty informative, yeah... although I can’t figure out why he used tiny.cc urls.

Re:Ray tracing vs. Rasterization

[Rockoon]

Because the user can control the camera and lighting, as well as other scene properties such as surface materials and objects.

You dont get that with offline rendering. Didnt take more than an instant to think of this, so why didnt you think of it? Oh yeah.. its because you are ignorant about what rendering techniques can and cannot do but insist that your opinion be heard anyways.

Re:Ray tracing vs. Rasterization

[quadelirus]

Ok, Rockoon. Let's lay it out. I write ray tracers as part of my job. I am studying this stuff very in depth. I understand what a ray tracer is, how to design one, how to write one. I also know how to use rasterization techniques like imperfect shadow maps. I know the methods behind SSAO, SSDO, etc. My original question was not meant to be an argument, it was meant to be a discussion topic. You clearly did not read the announcement. The announcement said that this is a "REAL TIME" ray tracer. Now I am pretty sure, given the current state of the art, that real time ray tracers don't give you the same quality as rasterization methods. And by quality here I mean it in a subjective sense, since a rasterizer generates a gross approximation. The quality is only how good that approximation looks to a viewer.

They will in the future, but they don't now. So there are two questions. (1) If you care about real time, why not implement a rasterization method for now and worry about ray tracing in a few years when it is better (now I'm not saying that you should do this, I'm asking for good and not asshole reasons for doing it)? (2) If you don't care about real time, why not render it on a server and send the result? A user most definitely COULD change scene properties and still have it rendered on a server. In fact, some game companies are working on systems to do just that--everything rendered remotely and only the results are sent across the network.

The real question, though, is not what the difference is between ray tracing and rasterization. That is pitifully obvious to anyone even slightly versed in the field (you seem to know enough to be dangerous yourself). The question is why do I want a real time ray tracer in my browser over a rasterizer. The reason I'm asking this is not that I hate ray tracers. The reason I'm asking this is not because I don't understand them. The reason is that given the current state of the art it appears that ray tracers don't render the quality at real time speeds that rasterizers do, but this might be wrong. So I asked the question in order to find out if there are papers out there for GPU ray tracing algorithms that produced high quality results (since I haven't seen any). I was hoping that people who knew would helpfully post some papers so I could read them. You unhelpfully decided to bash me when you don't even know who I am or what I study and you clearly misinterpreted my questions.

So you could helpfully post some academic papers detailing real time ray tracers that perform well and do more than can be done in a rasterization based solution, or go home.

Re:Ray tracing vs. Rasterization

[quadelirus]

Furthermore, if it weren't real time, what web application would set up a scene, let users move lights and stuff around, and then tie up their computer to slowly render a result. The motivation for this is not to be a 3d modeler within the browser. The motivation is to embed interactive 3d graphics in websites. For instance, in TFA they explain that they demoed an application where a viewer of the venice wikipedia page could "visit" a cathedral in venice and look around the scene. The point is definitely NOT to allow users to change lighting or material properties, it is to allow rich 3D content to be delivered via a web page which requires real time. That is why they say they are using a real time renderer. Or did you miss that?

Re:Ray tracing vs. Rasterization

[quadelirus]

Yeah, the problem is that a lot of academic papers have been published recently with "real time" ray tracers. But by "real time" they tend to mean 10-15 fps on the latest and greatest NVIDIA graphics hardware (usually over-clocked) with the ray tracer written in CUDA, which means
It may simply be that the paper is out for review, which would explain why I can't find it. The SIGGRAPH 2010 reviews come in on Monday (which means a busy day for me :)).

The other reason is the gut instinct that ray tracing is not an algorithm that is well suited for the GPU the way rasterization is, and so doesn't benefit nearly as much from advances in the GPU. I believe there is some work to design ray tracing specific hardware (it won't look much like a GPU) but I don't know how far they've come. Now they may have figured out some sort of clever acceleration structure, but I really want to see some video of the thing in action, and I want to read the paper on how they do it.

Re:Ray tracing vs. Rasterization

[zeroRenegade]

Here is the citation:
http://ieeexplore.ieee.org/Xplore/login.jsp?url=http://ieeexplore.ieee.org/iel5/4625376/4634600/04634631.pdf%3Farnumber%3D4634631&authDecision=-203

I can look for you, but it would be on one of a half dozen drives sitting on my shelf.

I was surprised they took it out of public circulation. Now you have to pay for it, or be associated with an institute that has access of the above IEEE digital library.

You should check if you have a document delivery service at your college or university, since they tend to be able to find most anything.

In regards to your comment, divergence is only a problem for ray tracing due to branching with SIMD. But the upcoming GTX 300 is going to have a MIMD hardware architecture, and we will then enter the world of photorealism. There may be some issues for CUDA at first, but I gather OpenCL is more than sufficient for the task at hand.

Here are some good papers on the subject by NVidia:
http://www.tml.tkk.fi/~timo/publications/aila2009hpg_paper.pdf
http://portal.acm.org/citation.cfm?id=1413968

The second link requires access through a university network or personal account.

In regards to your last comment, I believe light fields are more significant for volumetric displays. The day I get to play with an interactive volumetric display is the day I die a happy man.

If you are interested in light fields, check out:
http://gl.ict.usc.edu/Research/3DDisplay/
http://scripts.mit.edu/~raskar/lightfields

I like it

[Chrigi]

I think easy to program and easily accessible 3D is good for the web. There are quite some cases, where it makes a lot of sense. The Wikipedia example they showed is actually kinda neat. Some things just work better in 3D than 2D (e.g. molecular structures). From Text to Pictures to Animations to Video to 3D; I think it's just the next step of content display and doesn't replace everything on the web but adds to it and opens the web to a whole new range of content. Of course there are things one can see as negatives but that doesn't make it bad in principle. CPU/GPU/RAM hunger might not be as we wish and choosing ray tracing might be a rather strange choice. It's not like you'll browse the web as you would play Second Life.

Player 2 uses what?

[tepples]

a lot of games can be played with the keyboard and mouse

Player 1 uses a keyboard and mouse. Player 2 uses what? Per the same origin policy, applications running in web browsers do not connect peer-to-peer, so communications among the separate PCs, one for each player, will have to be bounced off a server through AJAX or Web Sockets as in an MMORPG. Latency probably won't be good enough for a twitch-fest first-person shooter.

An attempt to create anti-cheating measures would be interesting, to say the least.

It would involve running a copy of the game simulation on the server and synchronizing the two instances through AJAX or Web Sockets. Then the server can tell when the player does something he's obviously not supposed to.

Verdict: the architecture of web games will likely discourage twitch-fests that a simple bot can play better than a human.

Re:Player 2 uses what?

[clone53421]

Player 2 uses what?

A different computer. And yeah, you’re going to have to use a server to do this, which means you’ll have latency, and it probably won’t work well for certain types of games, which is pretty much what I said before.

It would involve running a copy of the game simulation on the server and synchronizing the two instances through AJAX or Web Sockets. Then the server can tell when the player does something he's obviously not supposed to.

Many forms of cheating are undetectable by the server. Wallhacks, aimbots, and full-scale botting... these are not easily detectable on the server. (Wallhacks are easily preventable by not sending the client the information about the other side of the wall... but preventability is a slightly different question from detectability.)

Verdict: the architecture of web games will likely discourage twitch-fests that a simple bot can play better than a human.

Probably. But like I said... if someone tried to make one, their anti-cheating measures would be interesting.

Re:Player 2 uses what?

[icebraining]

will have to be bounced off a server through AJAX or Web Sockets as in an MMORPG. Latency probably won't be good enough for a twitch-fest first-person shooter.

"As in MMORPGs"? That's what happens in all games, except a handful which actually use P2P. All commercial FPS need some kind of server, even if it's one of the players' PC (as in LAN games).

Re:Player 2 uses what?

[tepples]

All commercial FPS need some kind of server, even if it's one of the players' PC (as in LAN games).

I'm not aware of anything in the Web Socket API that allows an end user to open a web socket for incoming connections. The server has to be a separately installed application, and if you're requiring the end user to install that, you might as well have a separate app for the game too.

Re:Player 2 uses what?

[icebraining]

But that's only required if you want to play LAN games (thus needing to run your own server). You don't need to accept incoming connections if you're playing online using the devs' server, which is how these games are going to be distributed anyway.

And even "standalone" games are starting to require internet connections even to play in LAN (see Starcraft 2).

Re:Player 2 uses what?

[Anonymous Coward]

You seem hellbent on assuming that any games made with this will necessarily be multiplayer. I don't see why.

Re:Player 2 uses what?

[tepples]

What's the point of requiring an Internet connection to play a game if not for the purpose of playing in an instance with other players?

Re:Player 2 uses what?

[clone53421]

There are plenty of online single-player games. Ask the people who made and play them.

Sometimes the internet is a way to compare hiscores. Other times, it’s just a way to download the game so that you can play it. Many of them will play just fine if you disconnect from the internet after the game downloads. You may even be able to tell your browser to work offline and later revisit the game, loading it from the cache with no internet connection at all.

It’s no different from any other game that you’d download, install, and then play offline, connecting to the internet each time it starts up to download the most recent updates, and occasionally to upload your scores.

Re:Player 2 uses what?

[phoenix321]

The same reason why Google Earth is online-only: you could have 20 Terabytes of content, as long as only a small chunk of it is visible at any given frame.

3d browser markup extensions since 1990s

[peter303]

I remember playing with a few 15 years ago. They wrapped OpenGL as I recall. They did not perform very well in the pre-broadband era.

I hope you can disable this.

[JustNiz]

I hope there's an option to disable this in the browser.
I can already imagine that the only place where this tech will get used will be in advertising banners etc.

Re:I hope you can disable this.

[Anonymous Coward]

Nope, you won't be able to. Ever. In fact, as soon as you install it, it'll be burned on to the motherboard itself to refuse any and all ways to disable or uninstall this in your browser. It'll fucking ETCH ITSELF ONTO YOUR BRAIN to prevent you from doing anything against it. Oh, and it'll force you to LOVE it and NEVER WANT TO DO ANYTHING WITHOUT IT EVER AGAIN. And then THEY'LL ALL SPY ON YOU AND STEAL YOUR SOUL.

Yes, of COURSE you'll be able to turn it off. Seesh, THIS coming from the Pavlovian-response-like "FlashBlock is teh answar to EVERYTHING!!!!!" Slashdot crowd? Fnord fnord fnord fnord.

Re:I hope you can disable this.

[clone53421]

Oh, I’m sure.

Actually you can already disable it, and they haven’t even finished developing it yet: ##canvas

(It’s an Adblock Plus filter, in case you didn’t figure that out.)

X3DOM based on WebGL

[bernd_semmel]

AFAIK there is already a working HTML5 conform 3D implementation based on WebGL and the X3D standard. It's called X3DOM (http://x3dom.org).

There's no need for a separate build of Firefox when an HTML5 conform implementation will work out of the box in Firefox 3.7 via WebGL. There's already an official bug by Sam Ruby (http://intertwingly.net/blog/2009/11/05/Web3D) in the W3C bug tracker (http://www.w3.org/Bugs/Public/show_bug.cgi?id=8238).

Real-estate

[Anonymous Coward]

Could have its usefulness in real-estate marketing, construction, and such.

Just what Firefox needs... More bloat !

[Anonymous Coward]

Just what Firefox needs... More bloat !

Secondlife is a Platform

[RobertLTux]

as the tool chains improve and more stuff becomes stable the uses of SL will become more and more common

Linden Labs just released a public beta of the new SL 2.0 and i has a "Shared Media" feature that very much blurs the lines between Real Life and SecondLife.

You want to show a presentation inworld?? rez up a screen and go for it

(note for SL residents before you login using 2.0 grab the tweaks and have them applied)

No love for 3D.

[Anonymous Coward]

"Just like Second Life, the 3D web is not something people actually want, but more something which makes sense to old fashioned journalists who write for old fashioned media."

I disagree but then I've seen what things like Curl and Contona3D can do.

Advantages of 3D games in the first place?

[tepples]

CSS sprites are fine... for sprite-based games.

I can think of one advantage of a 3D game over a 2D game: greater draw distance than one screen-width, which in many games is about 20 m. But apart from that, what scenarios do only 3D graphics allow for?

Re:Advantages of 3D games in the first place?

[Anonymous Coward]

greater draw distance than one screen-width, which in many games is about 20 m

Huh? What does that even mean?

what scenarios do only 3D graphics allow for?

They look more realistic, and they allow you to simulate physics more realistically.

Re:Advantages of 3D games in the first place?

[tepples]

Huh? What does that even mean?

"Draw distance" is the distance of the farthest object from the camera before it disappears, in game world units. For example, a platformer might put the leading edge of the scrolling view 384 pixels in front of the character, where 32 pixels represent 1 m. The draw distance in this case is 384 pixels or 12 m. First-person shooters need a long draw distance, but there are stylized FPS games with 2D graphics, such as Nintendo's Duck Hunt and I-Mockery's Kill the Dog from Duck Hunt. But the only way to get long draw distances with a free-roaming camera is a 3D engine.

They look more realistic

Possibly, but prerendered sprites as seen in Donkey Kong Country do a decent job of showing depth on what is ultimately a 2D plane. And stylized graphics still have their uses; for example, the character models in the movie Up are anything but realistic.

and they allow you to simulate physics more realistically.

Not all video game designs need a realistic physics simulation. Play 99 Bricks to see what would happen if, say, Tetris had more realistic physics. Besides, a physical simulation can be done in 2D: see Crayon Physics and all its clones.

Re:Advantages of 3D games in the first place?

[clone53421]

But the only way to get long draw distances with a free-roaming camera is a 3D engine.

Ok. Yeah, that’s an advantage of true 3D.

Not all video game designs need a realistic physics simulation.

Some do. Having it when you need it is an advantage.

You asked for another scenario where 3D is an advantage. I gave you another scenario where 3D is an advantage. Of course it won’t apply to “all video game designs”.

The Most Obvious Use

[Foolomon]

Great! Now we can see the Fail Whale in 3D!

Why not use standards?

[Anonymous Coward]

This will never take off. Has anybody ever heard of XML3D? Where's the spec? Will this ever become a standard when there are already zillions of 3D markup formats around?
And it needs a custom browser build. Remember how 3D that required browser plugins failed back in the 90s? Now it's not even a plugin, they want you to install their browser. To view nonexistant content in a format nobody has ever heard of.
What's wrong with existing standards? WebGL, X3D. Inline X3D is the way to go according to the HTML5 spec. It already works in all WebGL-enabled browsers (alpha builds of Firefox, Webkit and Chrome). The difference is that this approach is entirely standards-based, X3D is an established standard, and WebGL is being implemented in most browsers right now.

Working right now in Shockwave

[Animats]

Shockwave already has full 3D capabilities. Here's s reasonably good 3D scene, a haunted house. Shockwave in current use; both Disney and Dreamworks have 3D promotional games for upcoming movies, and Porsche has a car configurator. It's possible to do a reasonably decent game in Shockwave. Unfortunately, Shockwave seems to be associated with crap sites full of ad-heavy low-end games that keep trying to download additional plugins.

A big downside of Shockwave is that, unlike Flash, the whole file has to load before it starts. It lacks instant gratification. Did the XML 3D crowd deal with that issue?

Incidentally, this is not the only XML-based 3D system. There's Web3D, which is simply VRML 97 with XML delimiters. VRML itself works quite well today. When it came out in 1997, few people had enough graphics power to run it, and it got a bad reputation. Now, everybody does.

Re:Working right now in Shockwave

[clone53421]

It’s a plugin.

It’s associated with crap sites full of ad-heavy low-end games that keep trying to download additional plugins.

The whole file has to load before it starts.

You spent a significant portion of your post explaining and excusing the very flaws that make Shockwave an undesirable solution.

A big downside of Shockwave is that, unlike Flash, the whole file has to load before it starts. It lacks instant gratification. Did the XML 3D crowd deal with that issue?

It could be done in the same way that web applications are currently handling the issue.

Re:Working right now in Shockwave

[muridae]

And XML3D won't be associated with "Yet Another 3-D IRC"? From Blaxxun to Second Life to some Flash site or another, if it has 3-D someone will create an avatar based chat room. Someone will use it for obnoxious advertisements, and someone will create a 100 MB file that has to load completely before starting.

As for it being a plugin, that seems like a better solution. You do not get the benefit of what ever efficiency is gained by being native, but you do not start subjecting everyone to the "Next big thing in advertisement, a 3D picture of a hamburger! With bites being taken out of it at random! Just think of the marketing potential!"

Interesting

[Anonymous Coward]

This could take internet porn to a whole new level...

Just like Quake Online.

[Anonymous Coward]

Download the addon for your browser and you can play (essentially) Quake 3 Arena for free. Decent framerate, decent enough graphics, that'll do for me.

Eheh

[SmallFurryCreature]

50k? Oh wow. For a "site" that is available around the world, that is pathetic. A dutch only site already does 10k easily. So 50k for the entire world is nothing.

And how many of the people logged in are bots?

Re:Eheh

[vadim_t]

50k? Oh wow. For a "site" that is available around the world, that is pathetic. A dutch only site already does 10k easily. So 50k for the entire world is nothing.

It's up to 57K now. This is not WoW, there are no instances or independent realms. Everybody is in the same world and has their stuff on a central server.

BTW, that seems to be around what Eve Online should have, but nobody seems to be making posts about that Eve is dying.

And how many of the people logged in are bots?

Not a whole lot, they recently clamped down on the useless ones. So there ones that exist should be mostly land management and such, and those exist because there are a lot of people to manage.

Re:Eheh

[Anonymous Coward]

50k? Oh wow. For a "site" that is available around the world, that is pathetic. A dutch only site already does 10k easily. So 50k for the entire world is nothing.

Since you didn't bother to mention which Dutch-only site: the "site" you are referring to is a generic Second Life (or Blue Mars, or PlayStation Home, etc) like 3D "world"? Or are you comparing apples to oranges here?

Google O3D?

[AmigaHeretic]

Is this suppose to be like http://arstechnica.com/software/news/2009/04/google-releases-3d-graphics-plugin-for-browsers.ars

http://code.google.com/apis/o3d/

Correction: pseudo-3D (Peter Pedant post)

[Zoxed]

3D desktop, 3D graphics, now 3D web. None of it is 3D, it is all *pseudo* 3D.

(Call me back when it is real 3D: i.e. when moving your head revealed more information round the side.)

(Bah, humbug ?)

Re:Correction: pseudo-3D (Peter Pedant post)

[clone53421]

It’s 3D. It’s just 3D with only one eye open.

As soon as we get cheap devices to display true 3D, switching from what you call “pseudo 3D” into true 3D is just as easy as opening your other eye. You just render it from a 2nd viewpoint.

Sprite-based pseudo 3D, on the other hand, is not 3D, and you can’t “just render it from a 2nd viewpoint”. There is no true 3D data from which to create the 2nd view.

Good luck getting your language installed

[tepples]

Your OS should provide privilege separation.

Should != does. To my knowledge, neither Windows nor any popular distribution of GNU/Linux comes with a GUI wizard to make a jail for a given native app.

A wise choice of language and libraries will provide portability.

Good luck getting the end user to convince a PC's administrator to install the runtime environment for your pet language and libraries. Much of this inner-platform work involves the web because the user is expected to already have a web browser installed.

Double the failure: Accessibility and Mobile Web

[narcc]

I see we've found yet another way to screw blind and low-vision users out of content. [This is worse than those youtube videos that are no more than a text only slide show set to crappy music. (To anyone who makes those: I know you're ashamed of your voice, but anything is better than no voice at all and 100x better than your taste in music.)]

I guarantee that this new 3D garbage will not be used in an accessible way. It's bad enough that no one bothers to use the ALT tag properly.

For you mobile web users remember: accessibility issues affect you! This is why most of the sites you visit look like shit, are hard to use, and don't render in a manner best suited to your device.

In addition, even your fancy iPhone isn't going to do real-time ray tracing. This content will not be available to you. Sorry.

Not that it matters -- I predict that like VRML, this won't take off.