Google Labs O3D | Hacker News Reader

Google Labs O3D

code.google.com

55 pointsendergen32 comments

Google Labs' open source 3D rendering browser plugin.

Can anyone explain to me what is the benefit of having to install a plugin for a browser in order to play a 3D game that is on graphics technological level of 10 years ago. When you can install a game without a browser and get a graphics technological experience of the modern age. I could understand somewhat a point in having an in-browser 3D or 2D game without plugins (Javascript or even Flash and Silverlight which I consider defacto indluded already), I just don't understand why would I want to install a game into the browser, which you do if you install a plugin, and tolerate degraded experience in contrast to standalone installation and have a full blown experience.

I DON'T GET IT, what is the motivation? It is as if some devs hope for worldwide installation for their plugin in high hopes of being a next gen flash, which won't happen. Some experiment with the idea plugin per game (idsoftware), which brings me back to the first paragraph... is it supposed to be a VMRL comeback? Let's do everything in a browser? Browser is the new OS? What is it?

There are a lot of things you could do with it besides games. Math, other kinds of data visualization, and so on. even on games, not all games re performance oriented; 2nd Life looks stupid but a lot of people like it. I wouldn't be surprised to see Google's failed attempt to copy it revived soon.

Perhaps to experiment and develop the <canvas> element 3d context API?

The browser is designed for browsing HTML pages, NOT to do everything under the sun.

Developers need to move beyond the browser and create a more heavy-duty general-purpose Internet application -- or create a framework for installing more domain-specific Internet applications in a sandbox.

And what would be a good base for this Internet application? Right now it's looking like web browsers.

Web apps could be way more powerful if they were full-fledged applications running on the server side, with just the GUI transmitted to the client - much like what you get when you do remote desktop.

Let /all/ the code run on the server side, rather than having to do client-side scripting and having to be limited by different client capabilities, etc.

The client-side interface becomes even lighter than current browsers.

Mind you, browsers are still good for what they were originally designed for, looking at basic Web pages.

It busted once already when Larry Ellison wanted to bring that back into the market in late 90's if I remember correctly. Computer industry shifted from terminals into home/desk computers in 80's, rightfully so I think. What made it lacking back then was bandwidth, which is no so very much the case these days. I think it is a good mechanism for portable, low power machines like cell phones, portable game devices maybe or things like that, but for home/office computers, no.

There's no need to run everything server side. But some potential advantages:

-No opportunity for software piracy

-Automatically "ported" to every platform, if we're really doing a remote desktop type experience.

-Software vendors can sell subscriptions for pre-installed software which is accessible via a "one-click" strategy.

If anyone wants to work on this as a startup, please let me know. mollison@cs.unc.edu.

You are right about piracy, it seems to be primary motivator today for this server/client model.

However there are obviously two schools of thought, one already set in place and working, other on the horizon:

1. Digital Content Delivery platforms: - Consoles: Sony PSN, Microsoft Xbox Live, Nintendo whateverthenameis for retro games - iPhone App store, itunes etc. - Valve Steam

This is a pretty damn fine model, it still needs polish - but I am already using it and I like it. I get the application I wan't, it works, it is native with no overhead, cheap, models for demos and trials etc.. it is the future, for now.

2. Browser as an OS/App host this seems to be that new trend I don;t understand. We already have the OS. Sure, you can build plugins for different platforms that host the browser and push your code to that plugin - essentially either as a VM bytecode (Flash, JVM, Silverlight) which transcends browser itself and we already have trial of faith on that, or you can push code each to his own plugin. Some companies, like google here, try to make a standard but essentially, to me atleast, I don't see the difference to Flash or Silverlight or JVM - sure, there is the native 3D/driver rasterization... but that will be in those techs too, Shockwave has/had it, I'm sure Flash will too someday. I still don't see the point in browser as OS/App host. I can see browser as an app store though, or a front to many different ones, but installation and run should be done in the end on the host itself.

i think this will all become more clearer once Microsoft and/or Apple implements app store like front which we already have for phones and consoles, but this time for desktops. Funny thing is that we already have that, not in mainstream though - look at linux and other repositories and installers they have (apt with synaptics, yum etc..). Add DRM to it and you are looking at the future. Steam might be the first in mainstream world, Microsoft will follow soon, I'm damn sure about it.

Developers need to move beyond the browser and create a more heavy-duty general-purpose Internet application

Wrong, your whole operating system has to be re-designed! :D

That's not true.

I posted an example in a comment that is a cousin of this one.

How would you say it needs to be redesigned?

All the linking that you can do with the web and the multiple representations of information have to be designed into your operating system. We've also learned that tagging combined with heirarchical categorization works really really well and the filesystem should be designed around that.

Yes you can hack all that stuff on current operating systems but it isn't the same.

Well, you're talking about a slightly different issue than what I'm talking about.

You're talking about making OS's more integrated with the Web. I'm talking about changing the Web to take full advantage of existing operating systems.

I agree, real gamers will download their games every time to get the latest tech.

Having 3D in a browser(that runs fast, unlike existing solutions) is a huge advantage for creating richer web-pages with embedded 3D visualizations.

IMHO generalized 3D rendering solutions tend to be poor compromises. They definitely don't make for cool non-homogenized looking games. But having some standard win is nice for academic and learning purposes as there ends up being more commonality between code samples.

On another note: it was a smart move on Google's part to build a slightly higher level system than say OpenGL, as doing collision detection for instance in Javascript is going to be dog-slow for awhile(At least across the board until the newer javascript engines are adopted more broadly).

I agree that for big studio games this might not be a good match - but for smaller more casual games, it may allow us to write them in JS instead of Flash.

Also, anyone that's tried to install a game on Windows knows why consoles still exist..

Why do you think the technological level of the 3D is 10 years behind? The plugin seem to use a unifying new shader language running on DirectX and OpenGL. I haven't had yet a chance to test it, but that sounds rather like something that would even be useful for desktop applications.

Also right now it is really hard doing 3D on the web. The only larger application that got it working somehow which I know of is Runescape (and even that has had a lot of troubles on different systems over the years). Which is btw. one of the most successful roleplaying games despite looking (and arguably also playing) a lot worse than a lot of free mmorpg's which require installation on the desktop. That is some hint that there might be a big market.

fusionfall! unity3d

Well, yes - I have seen Unity 3D and as it exist already a little longer it certainly has already a few nice examples. But what I'm interested is more - why should O3D be outdated? Is there anything in their way of technology that would prevent doing 3D graphics as seen in Unity?

Also they seem to aim at rather different targets. O3D is about creating an open webstandard. Unity3D is a proprietary 3D engine that can run as plugin on some platforms (windows+mac). O3D has to be at a low level to succeed - basically offering a way for web-programmers to access the 3D capabilities of graphic cards with tools which they know (javascript). Unity 3D is an excellent engine with a real nice price, but it doesn't look like a tool for web programmers but like a tool for desktop programmers which want to have their desktop applications also running in a browser.

Dear Google,

Thank you, it's nice to write OpenGL code in Javascript, but if you're going to write a plugin that lets JavaScript access OpenGL then please don't rewrite the entire OpenGL API.

From what I see from the documenation this is not an OpenGL wrapper. First it wraps OpenGL and DirectX. Then it also seems to be shader based, so it seems to be less a rewrite of OpenGL, but more a rewrite of GLSL. They describe in the FAQ why they didn't use GLSL itself.

"Most content on the web today is in 2D, but a lot of information is more fun and useful in 3D."

Wait wait... where did I hear that before?

Oh yeah. VRML.

The performance of this plugin destroys Flash in every way. It also does vsync (wow yay we're past 1998 now, unlike Flash) and you can put the scripts right in the page. Adobe's inability to make Flash not blow is stunning – they have an insanely huge install base and tons of money, but still can't make Flash perform even close to what might be called acceptable. Come on! You're getting the chance to run NATIVE CODE in just about every desktop computer out there, and you're blowing it!

Anyway, Adobe makes me angry. I hope this helps make them die off even faster.

Well if it's some kind of high level api to deal with 3d and shaders, and it's fun to code with and hasn't too shitty performance, it will probably work well.. What will be really needed is some good fun and simple game to create a buzz around it.

Wow, this could be massive for indie game developers but I'm not exactly up to snuff on browser-based 3D tech. Would love to collab and build a web game if anyone is interested, we're working on a game but for the iPhone / iPod Touch.

It appears my computer is too slow judging by the performance of the samples. :(

I see this:

O3D is an open-source web API for creating rich, interactive 3D applications in the browser.

And think, "Oh cool! It must be some funky 3D javascript library or something. This is pretty neat."

Then I scroll down and.. oh. Yet ANOTHER plugin. How is this different from say.. flash or silverlight or..

Silverlight and Flash do not yet support 3D accelerated graphics. A better comparison would be Unity... but Unity is not free to develop for. (And, unlike O3D, none of these other plugins are open-source.)

I am currently working professionally with Unity3d. Unity has taken care of a lot of the annoying legwork wrt animations, asset workflow, physics, state synchronization, c# (mono) extensibility, and more. Unfortunately, the primary language is a custom variant of ECMAScript. Another contender in this space is Torque3d, which offers the ability for you to have your own plugin and native code run on people's machines, but the licensing is not as straightforward as with Unity.

Today it's a plugin. Tomorrow it's probably built in to Chrome, Safari, and Firefox.

AAHAHAHAHAAHAH That's what they said about VRML!

On the other hand, I see the title, and think: "Oh dear, another funky 3d javascript library which is completely impractical and useless for any real application..."

Then I look at the article and see:

O3D is an open-source web API for creating rich, interactive 3D applications in the browser.

Perfect! Someone big finally taking initiative and creating a plugin which might actually eventually have a large userbase.

Reply on news.ycombinator.com