mafoo Wrote:Alexander Smith Wrote:I don't think going higher and higher in system requirements in the way to go with dim3. There's almost NO games released for dim3 right now, and what we need is better basic support.
Unreal Tournament 2004, Halo, and many other FPS games run on G4s and 10.3.9. I find it hard to understand how upping the system requirements is going to help anyone.
They also get about 10fps.
The fact is that no one is really going to really use dim3 while it looks like Java3D, and runs about the same speed.
The editor and the skeletal animation combined with the ease of the javascript for the coding are the major selling points.
Maybe it would be worth looking into attaching the renderer from another opensource project to the dim3 core.
For reference here are the sys requirements for World of Warcraft, which you should get about 10fps at with mid range settings:
933 MHz or higher G4, or G5, or Intel processor
512 MB RAM or higher; DDR RAM recommended
ATI or NVIDIA® video hardware with 32 MB VRAM or more
-------
On the other hand, you could do the more sensible thing, and leapfrog the others engines. Make shaders a requirement and have the lighting system written in GLSL.
I have a 933mhz Power Mac G4 with 512MB RAM and a GeForce4ti 4600. Halo and UT2k4 run quite well on my computer, much faster then 10fps. and that's with fairly high settings.
But that's beside the point. If dim3's system requirements have to be raised, the improvement is going to have to be big, with support for top level specular, glow and cube map reflectivity, as well as support for things like bloom. I don't want to have the 10.3.9 users unsupported unless there's a very good reason for doing so.
Let's not start a war on this, the beta is an evolving product and let's see what I can get. Reflections; real specular; etc, are ALL targets, without GLSL and still using 10.3.9. I'm always going to support 2 OS's back. When 10.6 ships, 10.4 will be required, etc.
[>] Brian
Quake 3 (Team) Arena seems to provide a lot more features then dim3 does, still it runs OK with a 1 GHz G4 and a GF2.
ggadwa Wrote:I'm still trying as hard as I can, but holding down system requirements is getting to be a real problem for me. I end up doing some effects in very cheap ways and they don't work out right.
Funny you mention that now. I recently showed my dim3 stuff to a buddy of mine who is in computer science and who is a Java and SQL programmer. He said dim3 would look like a Java game and would run at the same speed while running on 100 times faster hardware.
On the positive side he said XML can be OK and SQL would be overkill. He disliked JS to be the scripting language but approved its use under accessibility aspects. He also liked the licence free idea, that might be a selling point.
mafoo Wrote:The fact is that no one is really going to really use dim3 while it looks like Java3D, and runs about the same speed.
Note bah, i wrote a long reply, but I got wiped in the process of writing.
Here is the rough jist:
Alex, most G4 towers shipped with a Geeforce 4 MX which is just a rebranded Geeforce 2. GeForce4ti 4600 is incomparable to the MX.
The G4 tower probably also has 2 processors. Don't forget the altivec optimisations in those games as well which is why G4's have lasted well for quite a long time than G3s.
Compared to an iBookG4 of the same specs on paper, the standard G4 tower has about 4x the power.
-----------
On the subject of the engine, I'd just try and put in as much as you can and see what the overall speed hit is.
Its hard to write an lighting engine that looks good in diffrent settings. So maybe focus on getting one style homed down. (Doom 3 engine for example cant do anything apart from darkly lit corridors.). Maybe have a map setting for inside or outside.
I think a good solution to the above problems (Speed and lighting) would be to have diffrent build targets in the source that turn on and off different features so people can recompile a version without ____ fancy features - or with diffrent lighting tweaks, no networking etc.
Just some thoughts.
( ) I can bring valid information into a discussion.
(x) I just repeat what I heard from id software haters four years ago without any double-check of sources.
mafoo Wrote:(Doom 3 engine for example cant do anything apart from darkly lit corridors.).
Bradamante Wrote:(x) I can bring valid information into a discussion.
( ) I just repeat what I heard from id software haters four years ago without any double-check of sources.
mafoo Wrote:(Doom 3 engine for example cant do anything apart from darkly lit corridors.).
Proof from Carmack his godly self:
http://uk.pc.gamespy.com/pc/doom-3/539049p1.html
Stop changing the subject. Back to dim3 2.5b.
Not to delve into a Doom 3 war, but you are looking at one of the biggest problems with an engine like dim3; it has to be generic. When making Doom 3, you have the ability to tailor the engine for just the specific things you'll need it to do; remember how certain engines would be "good indoor" or "good outdoor" kind of thing.
This is the nature of the business; obviously, dim3 has a ways to go before it's good at "everything" but there is going to always be a bit of the "jack of all trades, expert at none" going on.
The best path is just to try things out, post them up as betas, and see what happens.
[>] Brian
Very true Brian, though there are engines out there, that work quite well for many environments. Doom 3 is a nice engine, but it's the graphics of the game that are so impressive, not the engine. My experience with Doom 3 and Doom 3 engine based games, is that they load really slowly, and suffer many frame-rate issues.
Even the new Quake Wars engine isn't that impressive, and also suffers from long loading times. Better look elsewhere for a good engine example in my opinion.
Current Progress for next Beta:
1) Fixed a big bug where physics could be completely broken (you'd float around the map and fall through everything)
2) Added walkToPosition and walkToPositionSlop
3) Minor editor updates
... the big thing I'm working on is rendering, I want to get this solid before too long. I have a better system (before it was a mixture of old segment stuff vs mesh stuff) but I've cleaned that all up. It's a *bit* slower but more exact, and speculars are part of the lighting process and should work as they are supposed to.
Right now they only work on opaque segments, though, so it might be a while (or I might need something more complex) to get them to work on transparencies. I'll work on that after I verify I'm going in the right direction.
I've also fixed all the triangle/quad lighting problems.
[>] Brian
ggadwa Wrote:Current Progress for next Beta:
1) Fixed a big bug where physics could be completely broken (you'd float around the map and fall through everything)
2) Added walkToPosition and walkToPositionSlop
3) Minor editor updates
... the big thing I'm working on is rendering, I want to get this solid before too long. I have a better system (before it was a mixture of old segment stuff vs mesh stuff) but I've cleaned that all up. It's a *bit* slower but more exact, and speculars are part of the lighting process and should work as they are supposed to.
Right now they only work on opaque segments, though, so it might be a while (or I might need something more complex) to get them to work on transparencies. I'll work on that after I verify I'm going in the right direction.
I've also fixed all the triangle/quad lighting problems.
[>] Brian
Hot diggity! Good fixes here Brian, especially the quad/triangle one.