OpenGL

Curious what you all think about it?

John Carmack seems to be a big fan. I really like the few teasers I've seen of his upcoming IDTech 5 engine/Rage game. But read this is an OpenGL engine/game.

How well do the current HD5xxx and the upcoming GF100 based GPUs run OpenGL? All I have read about these are are either their GPGPU tech (general processing shader units from what I gathered) and DX11 capabilities.

If IDTech 5 is going to be OpenGL, how come more games aren't. It would seem the transition from console to PC would be a lot easier also.

http://www.youtube.com/watch?v=XnBPqrhY3hw
- Man, Borderlands looks like a real ripoff of the Rage promo videos, a little bit shocking.

Looks fantastic considering he demonstrated this in 2008. OpenGL seems like a great open platform, why not more jumping on board?

He goes into depth about how efficient this is for a cross platform.

Anyone have an opinion or know the technicalities?

 

With Windows being the dominant OS out there for gaming using DirectX and the Xbox consoles using DX, it makes sense for them to use it. DX contains tools that make it easier to run the various functions they need. OpenGL is made for the same reason, (graphics API), but the problem is DX is more comprehensive and more efficient for gaming purposes. I think for AutoCAD OpenGL is still superior, as I'd expect it to be faster for non-gaming purposes. DX needs to run various shaders with extreme efficiency, OpenGL does not, as it's pushing extreme numbers of polygons. At least that's what I would think.

OpenGLs advantage is their is a version out their for all kinds of hardware, and OS. Mac OS X and Linux OSs use OpenGL, as I would expect Sun OS and Amiga too. It works well for them, as it's ready made. I think more OpenGL games need to be made, there needs to be more games available for the Mac and Linux. Valve developing Steam for Mac and porting the Source engine to platform is one of the best OpenGL related developments in recent history.

 

I am a long-time admirer of open, free, alternative projects, OpenGL is definitely one of them.

It would definitely make sense if OpenGL was more widely-used and investigated further in the industry and everyday use, although keeping the Open in OpenGL.

So far, OpenGL is a candidate to be a solid contender to the long-time standard in games and non-professional uses, Direct3D, using OpenGL would contribute to keep the costs lower.

Besides that, I would love to see Direct3D being shook around a bit, as the standard is monopolistic in nature, perhaps even succeeded by OpenGL in the future.

This may lead to Graphics Companies (I am looking at you, NVIDIA) to have to kick-start and become more innovative, diverse and provide better multi-platform support.

It would mean good news for Non-Windows users too, having to emulate Direct3D on, say, Linux is not the most effective solution to playing on Linux, especially with the rather poor support from both NVIDIA and ATI.

Besides that, as OpenGL is the graphics standard used by the Playstation 3, this would also boost the quality and variety of games on the console I guess.

 

Quick glance I would agree. But OpenGL seems to run just fine on Windows and porting it over to PS3 would be just as easy. No idea about OpenGL and the 360.

But watching the videos on IDTech 5 demonstrations, the whole pushing polygons on current DX11 seems to pale in comparison to IDTech 5 and OpenGL. The scenery etc just so much more immersive than anything I've see on on DX10 or DX11 yet.

 

There are a vast multitude of reasons why game developers don't jump ship back to OpenGL (it was dominant for a while back in the day). Of course, many game developers still use OpenGL (and most also write a Direct3D render as well, which is default on most Windows games). Most complete game engines (Unreal, id tech, etc) offer multiple rendering systems: OpenGL, Direct3D 9, Direct3D 10 or 11, and software. Unverifiable rumor has it that OpenGL was a faster system, but now that rendering pipelines have moved from fixed function to programmable, OpenGL has started to lose the race. OpenGL is playing catch up with Direct3D in many aspects, but it usually gets extensions (such as tessellation) well before Direct3D due to its vendor specific extension system.

From my personal experience (I've fiddled with OpenGL, primarily on Windows but also on Android and some minor DirectX)... OpenGL is not as developer friendly. I personally love its style, but it doesn't have as comprehensive set of tools. It also doesn't have as many useful utility functions (loading models et cetera). Also, OpenGL is not a complete suite of tools as DirectX is. OpenGL only handles graphics, and that is it. Its utility library GLUT which is not open source offers very basic functionality where as DirectX is a complete set of tools: audio, input, graphics, and even more. OpenGL also has a very fun problem... On Linux the default driver is Mesa, which is actually a software implementation (and thus very slow) of OpenGL, but at least it includes almost all features. On Windows, Microsoft only offers the library for OpenGL 1.1 (OpenGL is now on 4.0). This means that any program has to manually load extensions (and also check if they are available to use) on the current graphics card with the current driver. Essentially in order to use modern features, you have to depend on the graphics card, not Windows to guaranty that it exists. Also, Intel graphics (the most common graphics solution) terrible support for compliant OpenGL code, to the extent that many perfectly fine sections of code show up entirely wrong. Each operating system has a different input system, a different way to obtain a rendering context, unique extension libraries (like WGL) and usually have different quirks in dealing with OpenGL.

Onto the good news... OpenGL ES has taken the mobile world by storm (it is standard on the iPhone and Android systems) primarily due to its simplicity and streamlined nature. It also embraces the future by only allowing a programmable pipeline (in 2.0) rather than having fixed functionality. OpenGL ES is very similar (it is a subset) of OpenGL and thus if it becomes popular on phones and mobiles it may spread back to the PC sphere. Also good news is the increasing popularity of Linux and Mac OS, which are both much more dependent on OpenGL. As it gains in market share, the more the graphics vendors will support OpenGL with feature rich and optimized drivers.

Personally, I love OpenGL's syntax (and terminology since I learned it first) but I regret to say that Microsoft, Intel, and Khronos group have driven it into a rather precarious position. Thankfully the open source community has come up with some great libraries that make developing with OpenGL much more practical. It has hope, but OpenGL needs to be radically enhanced. $.02

 

OpenGL has been around long before DirectX and was the standard for a while. The problem with standards such as OpenGL is they strive for portability and robustness; While silicon graphics and other Special-interest groups that contributed to OpenGL had trouble pushing the libraries forward as they had to contend with keeping portability, Microsoft blazed ahead without any care to standard or portability (hey we'll simply force hardware vendors to comply with us right?) and was able to attract more developers with they feature-rich libraries while OpenGL was trapped by their own standards.

Because of it's robust API and simplicity it inherently has the potential to be better performing and more efficient than DX - that is if you know what you're doing like John Carmack who is a god among game devs. Not to mention it is inherently portable, all that is needed is an OpenGL driver supplied by the vendor.

tl;dr - OpenGL is better , DirectX is easier and more functional. Most game devs are lazy, and are simply looking to push a profit - hence they will use whatever's easier (DX). Combined with Mcirosoft and hardware vendors pushing DX harder then ever, don't look for OpenGL to make a "comeback" any time soon.

 

if you are making "just a windows" game, DX has better support out of the box in windows than openGL, (just because DX is built in). it is true that directx covers a large spectrum of gaming technologies, whereas openGL is specifically a graphics library. Of course, you can use openAL for your 3d audio library, and there are other free, portable tools available to meet other functions.

there was another thread about this, but basically, just comparing the renderer of DX to openGL, this is what is up:

- the performance isn't significantly different these days, because the overhead of the API is small compared to the actual computational tasks required of rendering complexity (shadow and lighting, complex geometry, other effects occur frequently in modern games)

- OpenGL can work on any operating system (and a variety of hardware devices), because it is portable, cross platform, and open source. DirectX works on x86/x86-64 processors running windows, and the xbox 360. dx is proprietary microsoft technology.

- The design of the two rendering APIs is very different, but they ultimately support the same functions. DX calls can be translated to OpenGL calls dynamically.

- openGL can take advantage of the latest DX 10/10.1/11 features in graphics cards that support those standards, and also backports some DX11 features to DX10 hardware.

 

The main reason John Carmack insists on OpenGL is for the portability then?

Also wonder if John Carmack's plan with ray casting and voxel tree will be able to compete with DX11 tessellation. At least my understanding that's his goal. If he can do this, wouldn't this make the OpenGL platform more interesting since tessellation won't be such a big feature for DX11?

 

Carmack liked the design mentality of Open GL over DirectX. the fact that it is portable and open source probably didn't hurt either. But AFAIK he tested both APIs kind of early and picked openGL and stuck to his choice.

 

Now Intel Larrabbee was about pushing the point cloud, voxel tree, ray casting rather than polygons right?

But DX11 seems to be all about polygons. Does that mean Intel is planning to push for a new set of API for 3D Gaming to compete with Direct3D or are they going to latch onto OpenGL?

 

dx11 tesselation is already in opengl 4.0