Google Brings 3D To Web With Open Source Plugin

maxheadroom writes "Google has released an open source browser plugin that provides a JavaScript API for displaying 3D graphics in web content. Google hopes that the project will promote experimentation and help advance a collaborative effort with the Khronos Group and Mozilla to create open standards for 3D on the web. Google's plugin offers its own retained-mode graphics API, called O3D, which takes a different approach from a similar browser plugin created by Mozilla. Google's plugin is cross-platform compatible and works with several browsers. In an interview with Ars Technica, Google product manager Henry Bridge and engineering director Matt Papakipos say that Google's API will eventually converge with Mozilla's as the technology matures. The search giant hopes to bring programs like SketchUp and Google Earth to the browser space."

Comments

vrml

[colmore]

So was there ever a single useful thing done in vrml?

I'm not trying to be snarky, I'm really curious.

Re:vrml

[stevenvi]

I once saw a 3D model of the Dopefish. It was neat. 3D graphics on my computer on the Internet! What'll they think of next?

Jumping ahead decade and a half... looks like it's just a competitor to Flash. Something else to make my computer run slower than my 486 did as I tried to execute Java applets back in '96...

Re:VRML

[daemonburrito]

Man... I thought we covered this with the last story related to the proposed Khronos 3d api. This is nothing like VRML. It is a javascript api to use graphics hardware.

Nothing to do with markup of any kind (aside from the xml in Collada, which is not necessarily part of the standard). Ugh.

Re:vrml

[Unending]

There are things I would like to see in 3D and I do think the capability to embed 3D objects is a useful step.
Off the top of my head:
-google earth in a browser.
-games are always a target for tech like this.
-any sort of 3d visualization of data that would benefit from non static viewing.

That said I disagree with how they made this, conceptually I prefer the 3D context for the canvas tag.

Re:VRML

[TheRaven64]

VRML can be accessed from JavaScript and can be 3D accelerated (it is in any relatively recent plugin). I am not completely sure I understand the difference between a 3D retained mode API, and a 3D retained mode modelling language that exposes its objects to a programming language.

Re:vrml

[mirshafie]

There are already proprietary applications for viewing 3D objects in browsers. These are useful for certain lines of work, such as displaying CAD models. I don't think the industrial companies that use these technologies will be the first to adopt open standards, but it might be a useful tool for smaller design houses.

As you wrote, online game designers will probably be all over this, and their ability to generate revenue should not be underestimated.

Re:vrml

[mikael]

Because you can embed the images in another webpage - just like youtube videos.

You don't have to bother about with unpacking zip files, rpm's, tar's, .run files, especially when you don't have admin permissions on the host machine.

Also, you won't start up your application one day, and read the message "This version is no longer supported. Please exit and upgrade to uber-version X.Y.Z".

Re:vrml

[lahvak]

So that I don't have to make my multivariable calculus students download and install new applications on their computers, so that I don't have to convince the IT folks at our school to install bunch of new applications that only a handful of students will use in the labs and classrooms, and I don't have to find an application that would run on all of my students' computers, whichever OS they use, the labs that mostly run windows, and my linux laptop.

Re:vrml

[lahvak]

Actually, there are many really cool mathematical visualizations done in VRML. The main problem with VRML was, imho, that the plugins were really clunky and never really worked that well. In fact, at the time VRML was introduced, the whole plugin architecture in most browsers was pretty simple, and even installing plugins was not easy. Finding, dowloading and installing a VRML plugin wasn't any easier than installing a standalone application for 3D, and the application probably had better functionality.

If this thing is well done, and easy to find and install, it can make quite a difference in areas such as mathematics or chemistry instruction.

Re:vrml

[dunng808]

There is no shortage of boobs posting on /.

Re:vrml

[SanityInAnarchy]

Compare Youtube videos to a native video player, the native option is much better.

Mostly because YouTube is based on Flash, and there currently aren't any major video sites using the video tag. I'd suggest that the video tag would be much better.

That's what self extracting installers are for, and you should be able to install to your home directory. If not, that's a packaging issue that's easy to deal with.

Unless they've also locked it down with something like noexec.

there's nothing stopping an app from being self updating.

True, but autoupdate is one of many things a browser / web-based application gives you "for free".

Another one is navigation. No reason a native app can't have hyperlinks back/forward buttons, and history, but why reinvent the wheel?

Another is extensibility. Without really doing much, you're probably still allowing people to write Greasemonkey scripts for your app.

Another is the refresh button. Complete reboot + autoupdate all in one.

Another is extreme portability -- native players may be better than YouTube, but it's difficult finding a machine that won't play YouTube out of the box. VLC isn't a terribly big download, but it's still an inconvenience, especially on machines where such things aren't allowed.

Another is security. Trusting one plugin to add 3D support is considerably safer than trusting every single application you might want to download that might want to render 3D. The browser is necessarily a sandbox, which means you don't have to set up a more complex one (like a chroot or a virtual machine).

The list goes on. You may not like the platform, but there are advantages to having an open standard portable platform. In fact, the browser is fulfilling the promise of Java so many years ago -- compile once, run anywhere.

I would say, if you don't like doing everything in the browser, and there's a specific reason you don't like it, improve it. That's what happened here, I'm sure -- Google doesn't like doing Google Earth in the browser, because the browser has no 3D. So they've improved the browser.

Re:vrml

[drinkypoo]

OK, why would any of those be better in the browser instead of as a native application?

No install, cross-platform.

Re:vrml

[gzipped_tar]

Better integration with Google's advertisement services.

Re:vrml

[johnsie]

The web browser is the new virtual machine.

Ugh. Again.

[Gricey]

The 3d web doesn't work. What "problem" are they trying to fix? That's the main reason it keeps failing.

  -- incubus

Finally

[geekoid]

goatse will be worth looking at~

Re:Finally

[Locke2005]

"When you stare into the abyss the abyss stares back at you." -- Friedrich Nietzsche

For some reason, goatse.cx always reminds me of that quote...

Re:Ugh. Again.

[aristotle-dude]

The 3d web doesn't work. What "problem" are they trying to fix? That's the main reason it keeps failing.

-- incubus

I know that this is slashdot but did you not read the summary? This could allow for Google Earth to function in a similar way to how Microsoft virtual earth 3D does within IE without need for a fat client on the desktop. The main difference would be that it would be more open and cross platform/browser compatible.

Show some respect

[Anonymous Coward]

I give Google credit for creating open source software, but I'm personally getting tired of the half implementation for Linux. I mean here is a company who has used Linux as the foundation for their internal use and they can't even muster up a deb or rpm package for their product, let alone 64 bit Linux support. Wtf Google.

Show some respect to the community.

Re:Show some respect

[Darkness404]

Most OSS projects don't offer precompiled RPMs or DEBs. Its much easier to outsource it to the distro managers who can better add it into the next release. Think about it this way, you offer a DEB for Ubuntu, that DEB gets installed on a Debian based system that isn't Ubuntu... Unfortunately, it doesn't really resolve dependencies nicely and it requires either A) A newer version of a library B) Some obscure library that doesn't have an easy to use DEB thus killing the entire point C) Dependencies that depend on other dependencies that either have problem A or B.

VRML and X3D

[DragonWriter]

It's been done...

...And done again. The problem isn't having a way to communicate 3D content with the browser in a form which supports interaction, the problem is coming up with something worth using it for.

Re:VRML and X3D

[hack slash]

the problem is coming up with something worth using it for.

Google StreetView 3D.

Or as it will probably (hopefully) be eventually known as: Grand Theft Auto: Streets of Google

THIS HAS NOTHING TO DO WITH VRML

[daemonburrito]

ffs!

Javascript. API. OpenGL.

SVG

[SpaceToast]

Speaking as an animator and web developer, I'd rather see this effort on the part of Google and Mozilla put into 3D SVG. It would eliminate the need for yet another plugin, allow direct DOM access, and facilitate the mixing of 3d with other page elements.

Or maybe I just want Lain's web experience...

Re:SVG

[Aladrin]

Isn't this trying to open the path for that? If they can get the API down pat with the plugin, and leave it open enough to replace the plugin with built-in functionality, it'll do exactly what you want, quicker, and with cross-browser compatibility.

Re:SVG

[BZ]

I meant specifically the object-heavy DOM API required by SVG, where everything in sight is an object, including things that would be better off as strings (the .href of an , for example, just in case you want to use SMIL to animate that; same thing for .className on all SVG elements). Oh, except you have to also provide it all in string form for the Core DOM APIs.

Having personally worked on a browser DOM implementation and done a fair amount of black-box testing of three other browse DOM implementations, a lot of the "shittiness" is baked into the specs. The SVG example is pretty typical: instead of using strings where they make sense, objects where they make sense, and something radical like integers where _they_ make sense, the DOM APIs typically require you to keep information in multiple forms at once and sync them to each other when any of the representations is changed. Setters are typically provided for all the different representations. In practice implementations often end up violating the letter of the spec and using a canonical representation in some cases just to keep performance sane. If the representations could be generated from each other, this wouldn't even violate the spec... but they can't be.

There's nothing flawed with the concept of an object model, of course, just with the design-by-committee object models the W3C has produced.

What is this brand new techology good for?

[Locke2005]

Since the dawn of computer communications, there has always been a single valid answer to that question: porn.

We tried something like this once before..

[moogord]

Remeber - its a neat little tag that is really quite powerful in the right hands, everything supports it but internet explorer, google made a plugin for IE but still no website uses canvas because you can't ignore the fact that no IE user has it (until HTML 5 if IE stays standards complient).

I would *love* opengl ES like 3d rendering in javascript, with a fast enough javascript engine you could do some great things, at the last you could make fluid websites without the need for a flash plugin eating up cpu... but alas i feel this is doomed to the same fate as our old google canvas plugin for IE.

I might have a use for this

[HTH NE1]

I just need a good 3D model of a 1953 Martian War Machine.

Re:We can hope

[MrEricSir]

I think you meant to say a game engine that's not evil.

Re:Ugh. Again.

[Tweenk]

That plugin does not work on Linux. The point is not to implement Google Earth as a plugin. It is to implement it using Web standards (VRML, HTML, JavaScript, etc.). It's the same as using SVG + JavaScript + SMIL instead of Flash, or Google Maps which doesn't use any plugins. If they manage to pull it off, it's going to be interesting.

Re:32bit Linux only - Give me a break

[Anonymous Coward]

I'm sure an OS as advanced as Linux can run a 32-bit application even if you are using a 64-bit OS and CPU. Right?

You just fucking bitch because you just HAVE to have 64-bit, OMFG!!! They are only giving you 32??? Jesus Fucking Christ, they are robbing you blind! How dare they do that to you! What a great insult to only receive half of what you are due.

In all seriousness, ignorant crybabies like you are why people hate Linux zealots. Grow the fuck up.

Seems like people are missing the point.

[slimjim8094]

This, and the canvas/video tag (if implemented widely) and fast Javascript (V8/Spidermonkey) will kill flash.

Flat out kill it. It might take a little while, but before long it will die out as soon as comparable dev tools pop up (and they will, because it's open).

I have a feeling this will be big - not XMLHttpRequest big, but not too far off. Need proof that this will succeed? Look at the hacky ways this has been done - Javascript raytracers, animated GIFs, writing software renderers in Flash - and tell me that people won't utilize a proper alternative when it arises.

Re:Seems like people are missing the point.

[spyrral]

Why would you want to kill Flash? Flash is great:

* Large install base with very fast uptake on new versions.
* Great IDE, large ecosystem of code, developers and tools.
* Easy streaming of HD video to the browser.
* Great communication server, video chat is an example level project.
* Small file size for the plugin, support for Window, OS X and Linux.

Other then being open, what would your hodgepodge collection of technologies and tools offer over Flash?

Re:Seems like people are missing the point.

[Endymion]

Because it's performance blows on non-standard platforms? Because it's lack of support for 64-bit (well, last I checked) is a pain in the ass? Because it's a constant source of memory leaks and segfaults?

The day I can get rid of that stupid plugin will be a day I celebrate a lot.

Re:Seems like people are missing the point.

[Anonymous Coward]

Why would you want to kill Flash? Flash is great:

* Easy streaming of HD video to the browser.

Sure thing... except that one can't use Flash to play back smooth fullscreen HD video on my desktop system when it's running Windows. Moreover, decoding that HD stream uses ~95% CPU when in "windowed mode".
My system specs and software load for Windows (or Linux):
* Windows Server 2k3 (or Unstable Gentoo Linux)
* Firefox 3.0.1 (or Firefox 3.0.8)
* Flash 10.0 r22 [installed just a minute ago] (or Flash 10.0.22.87)
* AGP R400 [AKA Radeon x850] with Catalyst 8.something (or xf86-video-ati 6.12.2)
* 2GB DDR1 @ 400Mhz [AKA PC3200 RAM]
* A single Athlon XP 2800+ [@ 2.1Ghz]

FWIW, The Linux Flash 10 plugin WILL NOT play the HD stream in windowed mode on this hardware. You can fucking forget about playing it in fullscreen... it's skip city. :/

I ripped the stream (a 1280x720 MP4 clocking in at ~2Mbit/s) quite some time ago.
My CPU usage when playing it with SMplayer 0.6.6 (using either the ATI OGL or DirectX renderers and postprocessing disabled) under Windows is ~65%.
When playing under Linux with SMplayer 0.6.7 (with the Xvideo renderer and a postprocessing level of 3) it is ~55%.
Hell... with the X11 renderer, a PP level of 3 and the video fullscreened, the CPU usage is only ~75%. (In case you don't know what that means... unaccelerated, postprocessed, fullscreen HD video playback through mplayer on Linux is faster than windowed HD playback through Flash on Windows.)

Here's the stream that I tested:
http://www.youtube.com/watch?v=4pXfHLUlZf4
Be sure to press the "HD" button.

and in another news..

[miknix]

the comet CHAIR-R51 is in route collision with planet earth!

Linux SketchUp Already?

[Doc Ruby]

Google's main 3D project is SketchUp, an easy 3D modeling studio. But it's not available for Linux. And it runs crappy, if at all, in WINE. It's also nearly the only way (other than a really tricky multi-app process with Blender) to import 3D buildings into Google Earth. Which means that without a Linux SketchUp, it's nearly impossible to get Google Earth to place the buildings properly (it requires IPC which doesn't work with SketchUp running within WINE).

So if Google is going to spend programmer hours bringing 3D to the masses, how about finishing bringing SketchUp to Linux already?

What about Metastream?...

[thekm]

Unsure how well this will go, maybe it'll work just because it's google. But there was an *awesome* 3D plugin ages ago called Metastream. It was by the group that made Kai's Power Tools (the first set of photoshop plugins that really got the plugins thing moving along). What made it awesome was that you could model the one model with as much detail as you wanted and then export it for Metastream. In the webpage you could just call the server and say that you wanted a little low-res version to show as a thumbnail, but if you wanted a product detail you simply call for the same thing but with more detail. The Metastream plugin changed the geometry detail and image mapping to whatever was needed to get it done... just like progressive images, but better and more complete (would be the exact same if you could tell a progressive image you just wanted it really small).

Anyways... Metastream didn't take off, but it was certainly an example of it done really well with a lot of possibilities. Because it was so good, it makes me doubt as to whether it'll be cool when google does it. Metastream was awesome.

Re:32bit Linux only - Give me a break

[BitZtream]

OMG
I MUST HAVE 64BIT EVEN THOUGH I HAVE NO FUCKING IDEA WHY!@$!%@#^%!@#%

You my friend are a fucking idiot for so many reasons.

Please name one practical reason why this has to be 64 bit. And please also specify if you mean 64 bit addressing or 64 bit words, and please explain specifically why it must be 64 bit.

Please do so quickly so I can tear you a new asshole before I go to sleep.

3D Web Applications

[megaheda]

If you build it, they will come... real 3D web applications, that is. They're heeerree, or at least they will be.

People claim that there are no real applications for web 3D. Humbug. Here's just a tiny subset without even trying. No, 3D isn't the solution for everything, but it's the solution for enough things.

Phase 1: Niche 3D apps move to the web
It'll start with the niche applications that are already 3D moving onto the web - CAD, architectural walkthroughs, collaborative design etc. A light version of your CAD or design software will be a web app (starting with SketchUp).

Phase 2: Mainstream web apps add 3D
Next, existing non-3D web apps will start to add 3D capabilities. Product configurators will be visual, your driving directions will have a 3D mode, your customer service applications will let you pull up a model of a piece of equipment and engage in real-time collaboration. These apps will function like regular web apps, but better.

Phase 3: Entirely new 3D web apps
Finally, over time, 3D will find its way into nearly every current web application that even touches on visual or spatial data - search, maps, 3D medical atlases, you name it. You'll be able to use 3D models to index databases. Text and visual information will co-exist like images and text but better because 3D models are structured. Virtually whatever you see can be 3D.

Oh yes, it will happen. Maybe not right away. The first batch of 3D web apps will almost certainly suck. There will be lots of failed projects until people figure out how to build 3D web apps. O3D may or may not be the ticket, but something else similar to it will be. It will take years and years, but it will happen.

Be kind to everyone you meet, for they are fighting a difficult battle. - plato