I've been following the Unreal 3 developement closely (like a real gamer ) and i thought about the possibilities for Nexuiz.

First I thought: I hope they release Unreal engine 2 in open source, so we can advance Nexuiz into a better weapon system. then I realized we have a strong foundation already, and it would be really hard.

Now I'm thinking: where is nexuiz going? If we remain on the Darkplaces engine, playing the quake 1 engine to its limits, we need a to move in to a new building, we could keep upgrading Nex on the quake 1 format, or we could move on to a more efficient rendering and gaming engine.

What would we have to do to migrate to another engine (commercial OR open source)?

If we really can keep upgrading Nex, is it worth it? What I'm asking is if we could be getting current Nexuiz capability with less work by migrating to another engine.

I don't mean to say Nex isn't a great game, I just think we might improve with the help of a new engine. I do realize that there is probably a lot of work involved in migrating, I'm only asking for more educated opinions.

iSuck wrote:Now I'm thinking: where is nexuiz going? If we remain on the Darkplaces engine, playing the quake 1 engine to its limits, we need a to move in to a new building, we could keep upgrading Nex on the quake 1 format, or we could move on to a more efficient rendering and gaming engine.

That is kinda like telling your sysadmin at work that he shuld change from windows XP platform, cause its running DOS format.

iSuck wrote:What would we have to do to migrate to another engine (commercial OR open source)?

wtf? commercial? wont and cant happen. The game and content is allready GPL. I also think noone would bother to put work into it if that happened.

Well... WHICH engine would you change to?

I can tell you - this was considered about a million times, and we found out every single time that there is just no better engine to change TO.

And why should they open source the unreal engine 2? They haven't open sourced unreal engine 1...

You'll have to remember that Darkplaces isn't just the most advanced open-source engine, but it is constantly evolving, compare stereo screenshots of the original Quake, and Nexuiz if necessary.

It has been said and said before, that Nexuiz blows away even UT2k4 in terms of visuals.

In some respects.

and it doesn't license seperate middleware to do it's hyped stuff either (COUGH*SOURCE*COUGH)

you wouldn't need open source unreal engine 2 since the modding possibilities of unreal engines are extreme...
Look at deathball for examle, or that tennis mod, or w/e
and I've been modding ut2k3 for some time

But it would be a rewrite, not a migration...
and I'm against it... I like the darkplaces engine
I wouldn't mind if it wouldn't support quake1 anymore though...

Code: Select all

ed@sn45g ~ $ no
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n
n


Just for every time this topic comes up again .

The darkplaces engine can do some fantastic stuff and with the right artwork will look as good as any other engine. All we need is features added incrementally, this is why I voted for improvement, the engine isn't standing still as it is anyway.

Blµb wrote:you wouldn't need open source unreal engine 2 since the modding possibilities of unreal engines are extreme...
Look at deathball for examle, or that tennis mod, or w/e
and I've been modding ut2k3 for some time

But it would be a rewrite, not a migration...
and I'm against it... I like the darkplaces engine
I wouldn't mind if it wouldn't support quake1 anymore though...

It would also mean redoing all content from scratch, because we do not have a license of it other than the GPL, which certainly does not allow using it in a project using a non-free engine.

It simply won't be Nexuiz any more. It will be something else. Why not make something else... right, why not do it. However, it would never replace Nexuiz. It would just be a new project...

I completely take back anything I said about darkplaces possibly being too limited. That was just my computer. I played nex on a desktop with a 3.5g processor (compared to my 2.0g laptop) and it was so much better than I thought it really was.

The only thing that was improvable was the quality of the models. We REALLY need to make new ones. That is the next step, looking at the capability of nexuiz in full quality is really inspirational to open source gaming, right up until you use chase_active, or play a close up fight. (I would be willing to help, I just learned normal mapping, and I only need a zym exporter for blender...)

Sorry for bringing this up, my computer is the problem for me, I need to mooch one of those fancy computers from school off of the tech guy. (they have full fledged 4ghz cpus with 1.5gig ram, it's crazy, and that's not even including the pile of shiny new nvidia gpus sitting in the back room, uninstalled)

The biggest problem is the player models, the weapons look fine.

Other than that the r_shadow_bumpscale_bumpmap should be higher (~15-20 IMO) by default, 32 is a little high, but 4 is way low.

about values like 16 or 32 being unrealistic....

there's always this question... what do you want?
should it be realistic? or should it LOOK nice?

I'm for the latter, and 32 definitely looks better, i'd go with 24 as default too...
If you think it's unrealistic, then go activate newtonian physics before making things LOOK bad hrhr

Speaking as someone who does not have the skills to contribute to the Nexuiz project, let me just say this:

Why are people worried about the engine when we can't even get someone to redo the player models (I don't care what they look like really, I'm talking about polycount and it's affect on overall performance.) Or even a menu system that doesn't suck for that matter...

I'm not saying this is a fault of any of the current developers who contribute - kudos on all the great work to make such an awesome game. It seems just to be a lack of time and physical number of people actively working on it. Hopefully the number of active devs will grow and things will improve.

I'm not even pushing the game very hard right now due to GPU limitation so I have a lot of room to grow into with the existing engine.

The Darkplaces engine is not the quake 1 engine, and is not limited by quake 1 engine limits.
The Doom 3 was ultimately based off of the quake 1 engine, even though it doesn't share any similar code. The quake 3 engine is based off of the quake 2 engine, which is based off of the quake 1 engine.

The problem is not so much the engine as it is the content. The textures could use some work and models do need some work. Take a look at Call of Duty 2, that game is using the quake 3 engine.

What I'm trying to say is: Darkplaces is an awesome engine that can compete with many commercial engines.

Ok.

I think there are two things that I would really like to see:

Primarily new player models.
I mean no offense to the original developers, but the models are really last generation. There have been myriad threads asking about player models. I would happily contribute, but I still have conversion problems from .blend to .zym. Could someone help me with that? maybe a link? I learned about normal mapping (without actually adding a separate normal image) in blender, so I really want to help with this. I also think that different races need different gibs, my robot should leak oil, not bleed. And I'm pretty sure the Quark doesn't have a stomach.-agian, I'm willing to make the models for this too.

The other thing is equality between Operating Systems.
I said in my earlier post that I played on a windows computer, and it pwnt. On the same settings on my mac, I played it, it lagged a lot, and it didn't look nearly as good with bloom, HDR, antiostroic filter, etc. In fact, on a mac, HDR and Antiostropic filters actually decreasd the quality, and make a grid of translucent spots all over the screen. I think I should just get a second computer.

iSuck wrote:The other thing is equality between Operating Systems.
I said in my earlier post that I played on a windows computer, and it pwnt. On the same settings on my mac, I played it, it lagged a lot, and it didn't look nearly as good with bloom, HDR, antiostroic filter, etc. In fact, on a mac, HDR and Antiostropic filters actually decreasd the quality, and make a grid of translucent spots all over the screen. I think I should just get a second computer.

There's nothing we can do about that, the OpenGL drivers on osx are horrible... They advertise features that don't work and GLSL seems to be broken beyond usability, so most of the fancy effects will either not work at all (like deluxe mapping) or are slower because they have to be done in software (like bloom or hdr).

MAYBE Apple will fix their drivers in the next release if just enough people complain. I see no other way to get somehow working GL drivers.

Of course, it may be possible to work around Apple's bugs by doing some things differently - but no single Nexuiz/DarkPlaces developer has a Mac that can do GLSL (I am the only one who has a Mac at all, and it can't do GLSL because it's just a Radeon 9200). Currently there are no single OS X specific parts of the OpenGL code, it's using the very same OpenGL routines on all operating systems, which is fine on Windows, Linux, and FreeBSD, but fails on OS X.

We really need a developer that knows much about OpenGL and GLSL on OS X. If there is anyone here, try looking at the source (I especially wonder why Bloom/HDR are so broken, as they do not even USE GLSL) and finding the problems. If you find any, or if you have questions, or if you just want to talk to the engine developers, join #darkplaces on irc.anynet.org and talk to LordHavoc or the others, but don't expect too much of them - LordHavoc is the one who knows the engine best, but it may still be worth asking in the channel if he is not around. If you have a patch that fixes some or all of these issues, just upload it somewhere and post its URL in #darkplaces, and someone with commit rights will find it and have a look at it (but I think I'll then make a Nexuiz build with this patch and post it on this forum first).

Thanks in advance, if any Mac/OpenGL developer reads this.

[url=http://www.realtech-vr.com/glview/]OpenGL Extensions Viewer[\url]

I just found this tool for Windows and OSX.

What I find interesting is that under OSX I have two options available for the renderer, the Intel GMA 950 OpenGL Engine - which the utility reports as supporting only a fraction of the OGL Extensions, 48/55 total.

But using the Apple Software Renderer 100% of the extensions are supported, with the exception of the 2.1 extensions.



Also, I understand the desire and simplicity to keep the same (standard) OpenGL codebase across 3 OS's but Apple does have their own specific extensions (13 in all) that are available to use. Here is Apple's Developer page for OpenGL http://developer.apple.com/graphicsimaging/opengl/

I understand you are the only Mac dev for Nexuiz, and don't even have a Intel Mac / GLSL capable machine to work with which means there is practically nothing happening for the Mac side right now.

Oh well. I still have a lot of fun playing Nexuiz and know that there is a lot of optimization that can happen for the Mac in the future.

[/url]

melk wrote:What I find interesting is that under OSX I have two options available for the renderer, the Intel GMA 950 OpenGL Engine - which the utility reports as supporting only a fraction of the OGL Extensions, 48/55 total.

The GMA950 is a very low end GPU, it's designed to be cheap and energy efficient, not to support as many OpenGL extensions as possible.

But using the Apple Software Renderer 100% of the extensions are supported, with the exception of the 2.1 extensions.

Well, since it's a software renderer they can implement any extension they want to...

I don't think these Apple specific extensions will help us much. Also, the main problem is not that OS X doesn't support some extensions - the main problem is that it claims to support them, but Nexuiz then tries to use them and it ends up in some broken way (see the all-black menu when we used GL_ARB_texture_non_power_of_two, which the OS X drivers claim to support).

In the future, we may have some use for GL_APPLE_float_pixels to make higher quality bloom/HDR, and GL_ARB_imaging for the same reason. But first, the bugs need to get fixed.

So, what is the proper channel to nag Apple about fixing their OpenGL drivers?

im really not?! sure! I think it was for redendring but now i got info that it is responsible for why there are no turrets and vehicles in Nex and why there are no facial expressions (ofcourse its kin of models fault they would have to be modified as well). Can someone explain to me like to a dump ass what is an engine!

DeVsh wrote:im really not?! sure! I think it was for redendring but now i got info that it is responsible for why there are no turrets and vehicles in Nex and why there are no facial expressions (ofcourse its kin of models fault they would have to be modified as well). Can someone explain to me like to a dump ass what is an engine!

I understood approximately 5% of that. Something about what an engine is? Well, in simples terms, it's the thing that cars, boats, planes have. You know, the thing that makes a lot of noise. Land rovers has them too, hate those things. It's basically what makes those things go, tick, breath, and live.

the game (graphic) engine if you would like to explain!

Well, how can one explain something like that that's easily understood. In it's most basic terms, it is the "world". You have all the rules there, physics and how things mix with each other. It's very complicated making an engine if you don't know programming and the like. I think wiki knows more than me: http://en.wikipedia.org/wiki/Game_engine

in thought it was only for redendering. well why do we wanna change the engine we could stick to this one and modify it so some things are possible. And for more demanding heave modifications (turrets etc.)

I vote stick with darkplaces, focusing on improving it.

Darkplaces is a very well written, very advanced engine in it's current state. You're not going to find many feasible replacements out there. It's very good at doing alot of things. Is it perfect? Not quite, but it's still one of the best opensourced engines out there.

A couple of areas that could use attention -

Better scripted shader support(I think currently it's limited to 1 stage, which really limits what you can do).

Better particle system(realtime scaling needs to be implemented here, or if it's there, Nex needs to use it).

Keep working on the netcode. It's much better than it was, but still doesn't feel quite right.

Reflective water If not, at least UT2k4 style faked reflective water(easily done with an envmap shader).

Turrets do not need special engine support

Well, I played the Bioshock demo, Gears of War, and a few other Unreal engine demos on my friends 360. I was really impressed at the quality, I thought it was a funny thing that video games, even in interactive and open-ended possibility, surpass the graphical ability of most animation studios non-realtime animation.

Unfortunately, no matter how hard anyone works on the Darkplaces engine, it would only support that kind of quality on a serious gaming rig. The normal maps on the players are incredible, I hope Nex will support normal-mapped player models in the future, and that mappers use more spec-mapping in their textures.

I was also wondering about the status of the "making that lazy second core do something" project. The PS3 has EIGHT, 3ghz processors, all doing a different peice of the render. If we got just two cores working to render together, instead of stranding that one half of the dual-core alone, we might make a dent in the completion of realism in Nexuiz.

iSuck wrote:I was also wondering about the status of the "making that lazy second core do something" project. The PS3 has EIGHT, 3ghz processors, all doing a different peice of the render.

It's acutally only one full blown cpu with eight co-processors (of which two are disabled in the ps3 afaik)
http://en.wikipedia.org/wiki/Cell_%28microprocessor%29

If we got just two cores working to render together, instead of stranding that one half of the dual-core alone, we might make a dent in the completion of realism in Nexuiz.

The cpu doesn't render, that's done entirely by the graphics card. So multi-core support wouldn't help directly with rendering speed, because on any modern multi-core cpu, Nexuiz doesn't even use 100% of one core. The bottle-neck is usally the gpu.