SGI: Computer Graphics

Graphics Question - Page 1

I am not a graphics expert so here goes.
I under stand that there is Ray-tracing and Scan-line Rendering.
Is there any software that would use OpenGL for animation?

Thanks.
OpenGL is mainly used for displaying the editing interface in a graphics application.
rendering is a software process that happens in non-realtime to create final output, with much higher quality than hardware typically can produce. there have been hardware-accelerated renderers, such as Gelato, but it has been discontinued.
:PI: :O2: :Indigo2IMP: :Indigo2IMP:
IRIX is very OpenGL heavy, but again as robespierre said its different from rendering. Mostly the OpenGL comes in handy for what he said, games, video editing and other graphically intensive tasks. Keep in mind, many CGI studios used tech for many, many years after it was replaced. I think Fox Animation ( Titan AE ) used either Indys or Indigo2s for the 3D of that movie. I could be wrong, I'm kinda drunk right now.
Entire collection up for sale :(
How about Machinima or the program Iclone?
Would an SGI machine accelerate any of that?
If I were to use software GL, is there any reason I could not make a high resolution image ( a single frame with say a million polygons ) with textures and save that as a jpeg?
edit=2;
AFAIK Machinima is by definition realtime, so yes, it would be GL accelerated. But Iclone isn't available for IRIX, AFAIK. Theoretically, any realtime application will be accelerated. You're going to face bottlenecks with the CPU and GPU though, especially on older hardware. My Octane2 is roughly as good as an early AMD64 desktop at doing that stuff, but I put probably $500 in this system to get it where it is.

For a hi-res image with textured triangles you'd need a pretty hefty system to do that in realtime. Pre-rendered? You'd be able to do it with most of what you're going to find.
Entire collection up for sale :(
If I were building a single frame with a lot of triangles and textures in it, I wonder if it would take minutes to build.
I am not talking about 30 fps or anything like that.
If it took one minute to 5 minutes per frame to render, I guess it should in theory be possible to make a high detailed render using OpenGL.
Then it would probably depend on the the capabilities of the hardware, ram,cpu,etc.
I believe the new Blender can animate and record the view port, If I have heard correctly.
I guess there could be CAD software that could animate an OpenGL object.
Thanks.
OpenGL is just a cross platform library of C functions that you can use to draw 3D into a window. The window comes from somewhere else, either Windows or MacOS or X, or a cross platform layer that sits between the window system and opengl such as SDL. X is both a window system and a 2D drawing system, IrisGL was both a window system and a 3D drawing system. OpenGL came from IrisGL, with the window system being deliberately removed. Animations are created from snapshots taken of OpenGL windows that are changing with time, which is to say that the data you send to the OpenGL functions is changing in a way that you program. Or use an application like Maya to program. Renderers take each snapshot of an animation and make them look realistic. Textures and light sources and such are all part of OpenGL but other than that that's pretty much all there is to it...
Project:
Temporarily lost at sea...
Plan:
World domination! Or something...
Thanks, I guess you all know that in pc land that there are some applications that use "game technology" to make animations. I guess I am not for sure why the idea did not get picked for other platforms. (Quake and I believe Unreal are some apps that used this idea)
edit:
Maybe I should have asked about CAD. If a CAD app. used opengl, I guess the limitation is some where between the app and the hardware on exactly how detailed an image can be.
A single modern high-end computer can render hundreds of millions of polygons in five minutes, I'm sure the SGI Octane III can do that, but none of SGIs MIPS-based IRIX workstations, which they stopped making 10 years ago, can do anything even remotely close to that. But if you're in the business of rendering scenes with hundreds of millions of polygons, you're doing it in big-assed render farms, often in New Zealand... :lol:
Project:
Temporarily lost at sea...
Plan:
World domination! Or something...
8-) :D
Maybe I will get a chance to see how Quake III and others will do on SGI Workstations.
Yeah Vishnu, of course you can use a modern day computer to do that, but that's no fun. And its funny because my Octane *feels* faster than my i5 school laptop for about a 25% of what I do.
Entire collection up for sale :(
Trekiej seems somewhat confused about what OpenGL does. It's not something that magically makes 3d fast, but a graphics library for displaying (2D and) 3D scenes, with the added benefit of the ability to offload the rendering from the CPU with specialized hardware that incidentally sometimes does the calculation more efficiently than a CPU with its general-purpose--and rest-of-the-system running--overhead.

What's muddled the waters is that nowadays graphics chips and drivers have all kinds of acceleration bells and whistles like video decoding acceleration, physics acceleration and (more importantly) what's called GPGPU (General Purpose GPU); where you can send the graphics chip arbitrary mathematical problems that it then uses its graphics cores and fast memory to run through. A couple of implementations for this are OpenCL and CUDA.

SGI machines can't do this (well, AFAIK, it's not something I've looked into) because back then it wasn't something you did. If you had a lot of calculations, you added a rack of C bricks to your Origin :-)

Anyway, I am not an expert, see signature.
:Octane: halo , oct ane Image knightrider , d i g i t a l AlphaPC164, pond , soekris net6501, misc cool stuff in a rack
N.B.: I tend to talk out of my ass. Do not take it too seriously.
Also, SGI made both workstations (like the Octane) and servers (like the Origin). If you had a heavy rendering task, you would run it on a server with many processors. Workstations like the Octane, with OpenGL graphics, were used for the interactive part of the task.
:PI: :O2: :Indigo2IMP: :Indigo2IMP:
TeamBlackFox wrote: And its funny because my Octane *feels* faster than my i5 school laptop for about a 25% of what I do.

I love "interface responsiveness." Hit the return key or click on something on a reasonable IRIX machine, and the machine feels like it immediately starts doing what you asked it to do. On a current laptop, there is an ever-so-slightly perceptible moment where it feels like the machine's mind is somewhere else, probably looking for some kewl visual effect to show you or figuring out a way to send all your personal data to some botnet controller somewhere.
Trekiej wrote: 8-) :D
Maybe I will get a chance to see how Quake III and others will do on SGI Workstations.


Since you asked, Q3 plays relatively well on my 800MHz Fuel with a V12, certainly enough to be playable. There are some graphical glitches in the menu, however.

That said, my quad G5 with Quadro FX 4500 mops the floor with it.
smit happens.

:Fuel: bigred , 900MHz R16K, 4GB RAM, V12 DCD, 6.5.30
:Indy: indy , 150MHz R4400SC, 256MB RAM, XL24, 6.5.10
:Indigo2IMP: purplehaze , R10000, Solid IMPACT
probably posted from Image bruce , Quad 2.5GHz PowerPC 970MP, 16GB RAM, Mac OS X 10.4.11
plus IBM POWER6 p520 * Apple Network Server 500 * HP C8000 * BeBox * Solbourne S3000 * Commodore 128 * many more...
josehill wrote: I love "interface responsiveness." Hit the return key or click on something on a reasonable IRIX machine, and the machine feels like it immediately starts doing what you asked it to do. On a current laptop, there is an ever-so-slightly perceptible moment where it feels like the machine's mind is somewhere else, probably looking for some kewl visual effect to show you or figuring out a way to send all your personal data to some botnet controller somewhere.


Indeed this feels the case on GaNoU/Leenux and used to be on FreeBSD, until I switched to i3wm on FreeBSD. I still use i3's stacking mode a lot and I want to see if I can somehow enable the 4DWM/CDE mouse over window = active behavior, although i3's vi keybindings are workable for sure. I'm just happy a few pieces of GPL are gone off those boxes.
Entire collection up for sale :(
If I write an application that uses OpenGL, do I not have the choice, depending on what machine it is, to use a software render or hardware accelerated render? I wished I had followed the DirectX and OpenGL on pc video cards back in the 90's to see what actually made them directx or opengl compatible.
The way to force opengl to render in software is to uninstall your video card's opengl capable drivers! :lol:

But no, there is no way you can force it to happen programatically, which is to say, in the opengl source code that you write...
Project:
Temporarily lost at sea...
Plan:
World domination! Or something...
Thanks. :D
TeamBlackFox wrote:
josehill wrote: I love "interface responsiveness." Hit the return key or click on something on a reasonable IRIX machine, and the machine feels like it immediately starts doing what you asked it to do.


Indeed this feels the case on GaNoU/Leenux and used to be on FreeBSD, until I switched to i3wm on FreeBSD.

After extensive research, I've decided that this is because Unix and the Yewnix wannabes on the desktop are braindead crap. Yes, this includes the newly-fashionable bsd's.

4Dwm is possibly responsive due to older software being lighter and, maybe in Irix case, written by smarter people. It is not as responsive on an smp box as I expected, however. In fact, it's disappointing.

There are a couple contributing factors ; the software is often shit written by imbeciles. Both BeOS and OS/2 have well-understood thoroughly documented methods for keeping the interface responsive. That was twenty effing years ago. You would think that knowledge would propagate. You'd be mistaken.

At the operating system level, it seems unbelievable that so few people would understand the difference between a server and a desktop. But what the hell, three hundred and fifty million butt-slapping grunting morons happily threw away four trillion dollars and the soul of the country handing the best friend the US had in the middle east over to our worst enemy, all based on lies a third-grader could see through. So I guess it shouldn't be a surprise.

Factoid : Unix handles threading very very poorly. Threads are not preemptive. Priorities are not kicked up, they are downgraded (eventually) if a thread doesn't give way all by itself. This is not that much better than cooperative multitasking. It works about as well, too.

The scheduler sucks. You can verify that by watching the cpu monitor. Processors are not evenly loaded. This is a good indication of a craapy design. 'Throw another processor at it ! That'll fix it !' Unh, yeah.

In short, pthreads are crap.

Loonix COULD have rectified these problems to create a high quality responsive desktop. Then they could have really had The Year of Loonix on the Desktop.

Instead. they created gtk3.

Gag me with a spoon.
wrestle poodles and win ! ...