id Tech 5

[xadhoom]

Here is an interesting short presentation of the id Tech 5 engine especially covering multithreading and resourcemanagement concepts used:

http://s09.idav.ucdavis.edu/talks/05-JP ... lenges.pdf

xad

[mkultra333]

Neat. Looks like a very different direction for Id, wide open outdoors instead of claustrophobic space dungeons. (Though I quite like space dungeons... hope they work a few in there.)

Id Tech 4 was great, but there was such a long gap between the first publicity and the final release that lots of other games had caught up by the time they released it. I wonder how long it'll be before we see this one in the shops.

[stoneCold]

Imo, it's a bit sad that such a major amount of their time had to be used to get a thread system going that would scale well on all the diverse CPU architectures out there (PC 2Core VS. Cell 8Core).
After all, they did a great job with their threaded texture virtualization system, and of course the rule presented in the paper "Virtual texturing + great artists = awesome environments" holds true, but the point that everyone seems to miss today is that just a small fraction of todays game studios can afford hiring a horde of topnotch artists that would actually put Virtual Texturing to full use.

Imo, they should have spent (at least some of) the time that they used to implement yet another killer quality engine that can now display X^1024 square-miles, for something more useful, like doing some groundbreaking innovation on the Procedural Content generation front.

Don't understand me wrong, I'm an id fanboy from very beginning, but it saddens me that quite any major game studio is still on the "we need photorealistic graphics, NOW" train and how they fail to see that it has become quite impossible to generate content in such graphical quality in a cost-effective manner (except for multi million $$$ titles, which cover only a relatively small fraction of the market still).

[/rant]

[wacom]

Yeah, I thought the same. I saw the presentation of Eskil Steenberg's LOVE at the Assembly '09 (via live stream) today, and it was the exact opposite to this PDF here ... Eskil creates very atmospheric worlds via simple solutions - no textures, use of only one shader, some particle effects, etc.

[stoneCold]

Wow, never heard of it before
If you've not heard of it either, have a look at this video, really impressive what this guy did on his own!

Sorry Mr. Carmack, since you missed the topic, my vote goes to Eskil this time

[nullsquared]

Wow, never heard of it before
If you've not heard of it either, have a look at this video, really impressive what this guy did on his own!

Sorry Mr. Carmack, since you missed the topic, my vote goes to Eskil this time

This guy is a genius

His tools look so futuristic and just plain ... EFFECTIVE

[xavier]

Couldn't have said it better myself

However, if you dig around in the links and the tech behind it, you find it's not just one guy (although the LOVE game might be just one guy), but that it's based on a Uni-Verse consortium initiative that ran its three-year course, and their last delivery was in March 2007:

http://www.uni-verse.org/Uni-verse-Home.72.0.html

March 15, 2007
The Uni-Verse Consortium is happy to announce its third public release of software in binary format. Quality should have been improved dramatically.

http://www.uni-verse.org/Documents.46.0.html

The Uni-verse project will run for three years, beginning in February 2004. During that period we will publish more material here, whitepapers and source code as well.

It's completely open-source and FOSS, but the original developers appear to have finished with their objectives.

[mirlix]

It looks really good, but I found some of the effect rather distracting, like this flash that comes when he selects something. Besides this few issue it seems like a extremely good tool set for doodads and so one. But I dont think it could be used for character modeling, but for this we have other tools. It like this "see your changes directly"-thingy he has .

[jjp]

I find id Tech 5 interesting, but it certainly has it's trade-offs. You can't have more than 4x anisotropic filtering, the amount of disk / hard disk space required can be huge and taking advantage of the ability to uniquely texture things might create a lot of work for artists.

[xavier]

But I dont think it could we used for character modeling

From what I could tell it's all subdiv-surface modeling, with the renderer figuring out tesselation at runtime based on hardware capabilities and poly-count metrics (supplied by the developer). But the real-time collaboration and near-zero iteration time is pretty much the "holy grail" of this type of development.

[mirlix]

Im not an Artist so maybe Im wrong, but personally I think it would be extremely difficult to get all the details for say a big monster with 4 arms and a stony skin with just drawing lines/polys and subdividing, of course with enough time it might be possible. But using alpha brushes like ZBrush/Mudbox to deform a mesh seems a lot faster to me. Of course one can combine alpha maps with the subdiv approach shown in the video, but I havnt seen it in the so I just assume it isnt possible, yet. But if this feature is integrated it would be pure awesomeness . I will ask my artist what he thinks about this.

But besides this minor point the project is great and the GUI is quite unusual while looking great and easy to use..

[jjp]

Generally you could model about anything with subdivision surfaces, they work great for characters. But the tool obviously is not as sophisticated as any of the major 3d dcc applications.

[nullsquared]

I think it would be extremely difficult to get all the details for say a big monster with 4 arms and a stony skin with just drawing lines/polys and subdividing.

Judging by the look of the video/screenshots, a big monster with 4 arms and a stony skin doesn't quite look like the art style he (they?) are going for

[mkultra333]

The Love editor looked slick, but it didn't strike me as really making it any easier to make good models, just easier for non-artists to make bad models. The artwork on that game looks like a giant leap backwards to me. Maybe it's just in the "programmer art" stage or something.

As of yet, I've not been impressed by any of the screen shots or graphics... with the exception of the Asteroids game, which looked great.

The thing that sounded most interesting, from the web page, was the game mechanics involving the camps controlling doors and such by transmitting trigger words at various "radio frequencies," and trying to hack enemy systems by learning the code words and frequencies.

[jjp]

The Love editor looked slick, but it didn't strike me as really making it any easier to make good models, just easier for non-artists to make bad models.

The editor certainly is not better than any existing tools. What makes it outstanding really is the fact that changes happen synchronously in the level editor and everyone working there can see them. I have doubts that this really is practical or even desirable. And after working with this tool for a few hours I would develop a craving to exterminate the stupid moron who thought it would be cool to have flashing light effects in a modelling application.

[wacom]

The editor certainly is not better than any existing tools.

I also don't think that editor is "better". It's just developed by him so it's more suited for his project, that's all.

What makes it outstanding really is the fact that changes happen synchronously in the level editor and everyone working there can see them.

That's just an implementation of Verse. I tried Verse with Blender and 3dsmax, and it worked more or less (TBH, with 3dsmax 8 and Blender 2.3x).
I also like the idea of "having no files". Make a model in your editor, *have* it instantly in the game. I think that's just cool. There's no simpler pipeline.

[mkultra333]

I thought the graphics in the gui looked better than the graphics in the game. Which seems like a bit of a problem...

Then again, there's not really a clear distinction between gui and game, so if the game itself is about building things and using the gui, it could work. Although here I see it falling into a no man's land, too complex for the online gamer and too simple for the professional artist. If he finds a way to make a cut down, simple mode (perhaps with even more tinsel and stroboscopic candy!) to attract the Second Life crowd, at the same time as making an advanced mode with more austere, powerful controls for pros, it could work well.

And really, would it kill him to have texture maps?

I think I prefer my modeller to have simple, bland controls. I don't want to be distracted by magical flashing, vista-like swirling and orbiting spaceships. And I don't want to be left with the false impression that my models are better than they really are just because the gui itself looked so cool. I think that's what triggers some scepticism with the Love project, if you remove the pretty gui and just examine the models, they look pretty poor.

[DanielSefton]

I think the way it works is just a proof of concept; it's a nice idea, but to be used practically it really should be done in a more productive manner (i.e. not without any proper GUI). Collaborative building isn't new, but it's certainly the way forward, especially for online games. I'm attempting something similar (not at the object manipulation level though lol) Personally this impresses me more: http://www.heroengine.com/world-building

By the way, that guy isn't such a genius as you might think, most of it is verse -- when you realise that verse (which is done by more than one person) is actually most of what he shows in the video.

Try it out for yourself (run the server and hey presto): http://www.quelsolaar.com/files/verse_apps.zip

On topic... ID Tech 5 was announced 2 years ago, I remember seeing the video of the WWDC conference here:

http://www.youtube.com/watch?v=HvuTtrkVtns

So when they were developing this technology, that's over 2 years ago, it's starting to look dated already (especially with the likes of CryEngine 3).

[nullsquared]

but to be used practically it really should be done in a more productive manner (i.e. not without any proper GUI)

Define "proper GUI"? The GUI he used look completely proper for his tools - very slick and to-the-point, not like traditional GUIs with thousands of menus and tabs and buttons you have to worry about.

[DanielSefton]

Define "proper GUI"? The GUI he used look completely proper for his tools - very slick and to-the-point, not like traditional GUIs with thousands of menus and tabs and buttons you have to worry about.

What I mean is, his GUI (or not his GUI -- verse's GUI), to me, is just gimmickry. Sure it looks cool, and indeed it very much suits his requirements. But a practical end-all solution for everyone? I don't think so.

It's a bit like the argument with Blender's interface. You generally work in Blender with one hand on the keyboard and one hand on the mouse, and that's it, you can do everything. But it's not to everyone's taste. People whine with non-standard UIs. Many people still prefer the 3ds max/maya bog standard hundreds of buttons.

Traditional GUIs are like they are because of the OS - unless there's some major shift, it's going to be static interfaces with hundereds of buttons to worry about for a long time to come. Tell Microsoft to develop Windows 8 with a UI like that, and software will follow (along with uproar)

I'll be interested when we all have these in our offices: http://www.youtube.com/watch?v=dyMVZqJk8s4 (Cool? Yes. Practical? No.)

[jjp]

The GUI he used look completely proper for his tools - very slick and to-the-point, not like traditional GUIs with thousands of menus and tabs and buttons you have to worry about.

I guess the GUI is alright to use if you are the one who programmed it and know where what is

It's a bit like the argument with Blender's interface. You generally work in Blender with one hand on the keyboard and one hand on the mouse, and that's it, you can do everything. But it's not to everyone's taste. People whine with non-standard UIs. Many people still prefer the 3ds max/maya bog standard hundreds of buttons.

Traditional GUIs are like they are because of the OS - unless there's some major shift, it's going to be static interfaces with hundereds of buttons to worry about for a long time to come. Tell Microsoft to develop Windows 8 with a UI like that, and software will follow (along with uproar)

A non-standard GUI can be great, if done for a good reason and if one properly communicates how to use it. When I was forced to use Blender some time ago my main problem was not the interface itself but it's poor documentation. ZBrush manages to get away with a very non-standard GUI.

[nullsquared]

Define "proper GUI"? The GUI he used look completely proper for his tools - very slick and to-the-point, not like traditional GUIs with thousands of menus and tabs and buttons you have to worry about.

What I mean is, his GUI (or not his GUI -- verse's GUI),

Verse is a networking protocol, not an all-in-one-solution.

to me, is just gimmickry. Sure it looks cool, and indeed it very much suits his requirements. But a practical end-all solution for everyone? I don't think so.

Well... that's kind of the point. It's not an end-all solution for everyone because it's an end-all solution for him and his project, which is what matters.

I'll be interested when we all have these in our offices: http://www.youtube.com/watch?v=dyMVZqJk8s (Cool? Yes. Practical? No.)

Youtube reports a malformed URL

[DanielSefton]

Youtube reports a malformed URL

Sorry, yeah, fixed. http://www.youtube.com/watch?v=dyMVZqJk8s4

Verse is a networking protocol, not an all-in-one-solution.

http://www.quelsolaar.com/files/verse_apps.zip - that's verse, basically most of what you see in the vid

I guess I'm just pessimistic. I mean, I'd love there to be a complete generic toolset that worked that way. But for now it's "open verse, play with it, be impressed, get confused, close it, open up 3ds max"

Edit: Whoa,

A non-standard GUI can be great, if done for a good reason and if one properly communicates how to use it. When I was forced to use Blender some time ago my main problem was not the interface itself but it's poor documentation. ZBrush manages to get away with a very non-standard GUI.

I completely agree. Blender's interface has a steep learning curve, but once you get used to it, you're suddenly working much faster.

[wacom]

http://www.quelsolaar.com/files/verse_apps.zip - that's verse, basically most of what you see in the vid

No, Verse really is only a network protocol. All apps in this zip USE verse, and are developed by Eskil Steenberg (while he's one of the main developers of Verse).