Unigine 2.1 released- OpenGL 4.0 support with Tessellation

[Sylvanas]

Let the benching begin

Nvidia users be sure to use the new 256 drivers as there are performance improvements. This thread aims to determine the difference between OpenGL and DX11 in terms of visual quality, performance and specifically Tessellation performance between the two API's- which is more efficient?

Win7 64bit 256 driver

 

[tweakboy]

What games are going to use this engine over proven stuff like valve or Unreal 3 engine. thx imo

 

[Sylvanas]

Bump for bench results between DX11 and OGL with Tess ON.

 

[Scali]

I think AMD users need these drivers:
http://support.amd.com/us/kbarticles/Pages/Catalyst-OpenGL-preview-driver.aspx

The Catalyst 10.5 release doesn't seem to incorporate OpenGL 4.0 yet.

 

[AndroidVageta]

What games are going to use this engine over proven stuff like valve or Unreal 3 engine. thx imo

None except for their benchmarks...

OK, in all seriousness...no games that youll probably be interested in playing or will ever hear about...

 

[Kalessian]

There are a bunch of things announced.

Primal Carnage and Afterfall: Insanity are probably the biggest two names.

There's a pretty neat machinima movie thing based off of the engine. Let me find it. http://www.magusexmachina.com/

 

[Lonyo]

There are a bunch of things announced.

Primal Carnage and Afterfall: Insanity are probably the biggest two names.

There's a pretty neat machinima movie thing based off of the engine. Let me find it. http://www.magusexmachina.com/

That Afterfall article says Unreal Engine 3.

 

[Kalessian]

Hmmm. Well someone has their info wrong because both wiki and the unigine site claim that afterfall is using unigine. This dates back to 2009 though so maybe they jumped ship.

 

[Lonyo]

Hmmm. Well someone has their info wrong because both wiki and the unigine site claim that afterfall is using unigine. This dates back to 2009 though so maybe they jumped ship.

Recent articles seem to indicate they did jump ship to Unreal Engine.
22nd Feb 2010 press release from Unigine says they are using that engine.
6th April 2010 press release from the European distributor says UE.
And the game website (for me at least) seems to want login details.

But if it is true that they were Unigine for like 2 months before jumping, that doesn't sound good.

 

[tweakboy]

Thanks guys great thread. Very interesting stuff.. thanks

 

[Kalessian]

In any case, Unigine has humble roots in an open source project before it decided to go commercial a while back. Looking back to when I first heard of it on phoronix, they've managed to make something more impressive than I could have expected. They're a new player with a small team and to license their engine is inexpensive compared to the big guys.

What I'm getting at is that for this round, at least, they have always had indie devs in mind. If you look at their website, they even have a sort of discount for licensing if you are a small design studio. It might be just a matter of time until unigine is a really big name.

 

[Scali]

The Catalyst 10.5 release doesn't seem to incorporate OpenGL 4.0 yet.

Scratch that, they do.

 

[taltamir]

its a new DX11 / OpenGL 4 + tessellation engine, of course nobody uses it yet... how many games are there that use DX11/OpenGL4 and tessellation at the moment?

However companies might use it in the future, as a small player it would be cheaper to license, and IF its performance is good enough it might catch on.

That being said, it might just fail... we cannot predict its future.

 

[Scali]

its a new DX11 / OpenGL 4 + tessellation engine, of course nobody uses it yet... how many games are there that use DX11/OpenGL4 and tessellation at the moment?

It's a DX9/10/11 and OpenGL engine.
It's been around for a few years (since 2005 actually), DX11 and OpenGL 4/tessellation were just recent additions.

 

[taltamir]

It's a DX9/10/11 and OpenGL engine.
It's been around for a few years (since 2005 actually), DX11 and OpenGL 4/tessellation were just recent additions.

mmm, its older then I thought then... nonetheless I consider such implementations as more of a bundle of 3 separate (but evolutionarily related) engines, DX9, DX10, and DX11... there are significant differences between those 3. The DX11 part is new and might be the big break for this company.

 

[Kalessian]

When the Heaven benchmark came out everyone started to say "whoa look at this new engine". No one had heard of it before then just because it didn't really have anything going for it before it adopted tesselation/dx11 early on.

From what I gather, the lead developer (co-founder) has pretty much designed and built the majority of this engine, and there has been no major hit to put them on the map until Heaven. He seems pretty dedicated. I say this because of the team page as of 1/2008: http://web.archive.org/web/20080111083920/http://unigine.com/company/team/ and because of the Frustum about page, where Unigine was born: http://frustum.unigine.com/about/ The guy has a drive to do this kind of stuff.

As long as he and the small core of developers are still around I don't see it failing anytime soon, I mean it hasn't failed in the years since it began why would it stop now just as it's starting to gain recognition?

Earlier revisions of the engine were not really ready yet. At least, looking at some of the project dev blogs suggest that not everything was all there. On top of this, if you've ever tried to run the old tech demos then you'd see that the graphics were about even with some of the open sourced engines based on the old iD tech stuff. This (2.1) might be be the first real worthwhile version, and it looks to be targeted at indie devs.

We could be looking at a story about two guys who worked on an open source engine as a hobby, then decided to go big with what they know and then they actually succeed. The classic build it in your basement/garage scenario.

/end opinionated rant

 

[Scali]

mmm, its older then I thought then... nonetheless I consider such implementations as more of a bundle of 3 separate (but evolutionarily related) engines, DX9, DX10, and DX11... there are significant differences between those 3. The DX11 part is new and might be the big break for this company.

No, it's a single engine.
You can select all 4 APIs in the benchmark.

 

[Scali]

Earlier revisions of the engine were not really ready yet. At least, looking at some of the project dev blogs suggest that not everything was all there. On top of this, if you've ever tried to run the old tech demos then you'd see that the graphics were about even with some of the open sourced engines based on the old iD tech stuff. This (2.1) might be be the first real worthwhile version, and it looks to be targeted at indie devs.

I speak from personal experience when I say that making a convincing demo is the most difficult part of it all.
You can have the greatest engine in the world, but if you don't have the content to show it off, you won't be able to convince people.
Artists play a very important role in it, and it's very difficult especially for a hobby/free/open source project to attract good artists and really show off the technology to its full potential.

 

[taltamir]

No, it's a single engine.
You can select all 4 APIs in the benchmark.

Yes you can select all of them in the benchmark, yes they are bundled together, yes they can be used in the same game, and yes they are compatible with each other.
No I would not say that makes them the same engine, they are evolutionary and incremental improvement to the engine. Things changed a lot between DX9 and DX10, and even between DX10 and DX11. just because a game can work in DX9, DX10, and DX11 doesn't mean its one single "thing" doing all 3.

 

[Scali]

Yes you can select all of them in the benchmark, yes they are bundled together, yes they can be used in the same game, and yes they are compatible with each other.
No I would not say that makes them the same engine, they are evolutionary and incremental improvement to the engine. Things changed a lot between DX9 and DX10, and even between DX10 and DX11. just because a game can work in DX9, DX10, and DX11 doesn't mean its one single "thing" doing all 3.

It's a single "thing" because you can create a single codepath that can run on multiple APIs.
And you can use some kind of extension mechanism to use special functionality that is only available in some APIs.
That's how they can run the same benchmark on 4 APIs. That's the strength of Unigine. It is API-independent (and also platform-independent, as it can also run on linux, and probably also OS X and other similar OSes, but they haven't released any benchmarks for those).
The whole point of an engine like this is to abstract those API differences away. I should know, I built one myself.

 

[taltamir]

interesting. Just to clarify, are you saying that this is something unique to Unigine? or that all engines work that way?

 

[Scali]

interesting. Just to clarify, are you saying that this is something unique to Unigine? or that all engines work that way?

No, it's not unique to Unigine. With DX10 having Vista as minimum requirement, more and more engines became multi-API. They had to, if they wanted to support XP and DX9 hardware under Vista.
And then there's engines that also support consoles (eg Unreal Engine 3, CryEngine 3).
But not ALL engines work this way. Valve's Source for example, was DX9-only and Windows-only for a long time. They have recently added OpenGL support though (but not for Windows), and support OS X now, and possibly linux in the future.

But Unigine is one of the few engines that offers both DX and OpenGL on Windows (with all the latest features that is), and can run on various OSes.
The lack of console support will make it less attractive for game studios though.

 

[Scali]

Back on the topic of performance...
I haven't done a full benchmark, because I didn't think there was much of a point.
On my Radeon 5770, the first part ran quite choppy in OpenGL 4.0 with Cat 10.5, averaging about 9.2 fps.
So I thought "Humm... it wasn't THAT slow in the DX11 version, was it?", and I quit and reran it in DX11 mode. And indeed, it ran with an average of about 20 fps in the first part.

Not much of a point in running a full benchmark... apparently the OpenGL path is way WAY slower... at least on my Radeon 5770. I wonder what nVidia's cards do, they always seem to perform better in OpenGL anyway.

 

[tweakboy]

Wow, nice thread. Informational that's great.

 

[Sylvanas]

It's a single "thing" because you can create a single codepath that can run on multiple APIs.
And you can use some kind of extension mechanism to use special functionality that is only available in some APIs.
That's how they can run the same benchmark on 4 APIs. That's the strength of Unigine. It is API-independent (and also platform-independent, as it can also run on linux, and probably also OS X and other similar OSes, but they haven't released any benchmarks for those).
The whole point of an engine like this is to abstract those API differences away. I should know, I built one myself.

It's very much in existence for Linux. See here. (or the download page ) I think you can run either tropics or Heaven on OSX too via the Phoronix test suite.