How good is darkplaces really?

Anybody who shows nexuiz to one of his hardcore gaming friends immediately gets asked the same question: "Why does it look so shitty?"
I want a straight answer from the community because i'm tired of hearing the EXACT SAME CRITICISM over and over again.

I have looked at lots of gameplay footage on http://www.gametrailers.com looking for things darkplaces cannot render and my list is pretty short:
1. Full motion video textures.
2. Warping of the environment. This seems to be a post effect commonly used around explosions.

In my mind if you somehow import the level geometry from, for example Kane and Lynch 2, it should look just as good.
If i'm unaware of some limitation of darkplaces please indicate.

Keep in mind that this is just the community making the game, not a whole game company.

the maps of nexuiz were not so good-looking, thats the whole problem.

and darkplaces cant do some more things like physics (there are some videos about that at youtube, but the physics are buggy), including playermodel physics. its a good engine but you cant say "i dont see anything which the cryengine can do and darkplaces cant do, so darkplaces must look as good as the cryengine"

the maps of nexuiz were not so good-looking, thats the whole problem.

That's my theory also.

and darkplaces cant do some more things like physics (there are some videos about that at youtube, but the physics are buggy), including playermodel physics.

True. Let's add that to the list.
3. Player model physics

its a good engine but you cant say "i dont see anything which the cryengine can do and darkplaces cant do, so darkplaces must look as good as the cryengine"

Why not?

Keep in mind that this is just the community making the game, not a whole game company.

Really no excuse in my mind. It's understandable if we set a low bar but that's not a reason to.

The art is simply not (yet?) at a level similar to AAA commercial titles.
It takes several full days if not weeks of an full-time employed hightly skilled artist to create a SINGLE playermodel of the quality found in Crysis, so how can you expect a game to look similar when done in the free time of some (probably) not as skilled artists?

Thats really the whole point!

Of course since Xonotic's art is also open-source, we can build on it freely, and maybe some day it will look as good as commercial games.

Of course since Xonotic's art is also open-source, we can build on it freely, and maybe some day it will look as good as commercial games.

no, it wont. there is a reason why games which look good have a few hundred developers.

and Ihsan: you want to expand the list? check this out: http://en.wikipedia.org/wiki/CryEngine_3

Of course since Xonotic's art is also open-source, we can build on it freely, and maybe some day it will look as good as commercial games.

no, it wont. there is a reason why games which look good have a few hundred developers.

There are more than 600 people on this forum. What a company has is focus and quality control because of it's limited resources and expectations of what makes it competitive in the market. Opensource has infinite man-hours because it has no deadlines and doesn't pay it's devs. I think our artwork standards are just too low but again, just an opinion. I don't think enough hands touch the models and maps before the are released so they never get polish they need. Hopefully git will make the iterative development of artwork a reality, or maybe there need to be organized teams to get features and artwork done in certain branches. I have no answer to this really.

and Ihsan: you want to expand the list? check this out: http://en.wikipedia.org/wiki/CryEngine_3

There's a bunch of filler in that list because they must at least make it sound better than other engines ("Uber Shader" thecnology?) BUT there a few notables:
4. Multi-core support(?)
5. Soft shadows
6. Subsurface scattering
7. Deformables and soft body physics
8. Rope physics
9. Depth of field

its a good engine but you cant say "i dont see anything which the cryengine can do and darkplaces cant do, so darkplaces must look as good as the cryengine"

Why not?

Throw the same scenery at DP and watch it crawl to a halt. That's my guess anyway. Cryengine has been optimized for the latest hardware, where DP is afaik still based on older technology. You don't see stuff such as automatic terrain lod, automatic culling, entirely realtime lit worlds or whatever Cryengine manages to pull off with pretty impressive framerates. Performance takes a hit at an unfortunate rate as you add stuff into a DP map, not to mention how unrealistic and dull the lighting of q3map2 looks (at least at the default settings), so you just can't easily get up to that great quality without sacrificing lots of performance it seems.

But where we lack the most is still art. I'm not certain a good looking map can be made only by having a mapper slap together his layout, someone else's textures and prop models to one. Instead the mappers, texturers and prop modelers need to work together to make textures and models that fit exactly that map, and perhaps also let the mapper adjust his map around the texturers/modelers ideas. But that's something you'll hardly see in an unpaid volunteer project, it seems hard enough to get the separate parts

---

There are more than 600 people on this forum. What a company has is focus and quality control because of it's limited resources and expectations of what makes it competitive in the market. Opensource has infinite man-hours because it has no deadlines and doesn't pay it's devs. I think our artwork standards are just too low but again, just an opinion. I don't think enough hands touch the models and maps before the are released so they never get polish they need. Hopefully git will make the iterative development of artwork a reality, or maybe there need to be organized teams to get features and artwork done in certain branches. I have no answer to this really.

+1

+10^24, rather, to all of that :-P

I can understand that the beta would have maps that aren't necessarily top quality, because something should get released soon to keep the community active. But for the next version, I'm thinking polishing media should be the top priority. All media needs to have a consistent feel, maps should have consistent details and overall quality, size, flow and such factors. I'm not too much in favor of eg. having one HUGE map with ons/vehicles, and the others being smaller deathmatch arenas. Instead ons/vehicles need to be tweaked to favor maps of the sizes that all the other gamemodes use, so as to not even give a mapper a reason to try and create a huge terrain map, DP JUST ISN'T BUILT FOR THESE! If you ask me, Xonotic would mostly consist of indoor arenas with a few exceptions here and there, but also these exceptions maps would be mostly indoors.

There are more than 600 people on this forum. What a company has is focus and quality control because of it's limited resources and expectations of what makes it competitive in the market. Opensource has infinite man-hours because it has no deadlines and doesn't pay it's devs. I think our artwork standards are just too low but again, just an opinion. I don't think enough hands touch the models and maps before the are released so they never get polish they need. Hopefully git will make the iterative development of artwork a reality, or maybe there need to be organized teams to get features and artwork done in certain branches. I have no answer to this really.

You have to realize that not every one of those 600 can actually model or do anything useful. Less than 100 people have git branches or have contributed tangibly to the project. Zed Shaw explains a more extreme version of this phenomenon.

Instead ons/vehicles need to be tweaked to favor maps of the sizes that all the other gamemodes use, so as to not even give a mapper a reason to try and create a huge terrain map, DP JUST ISN'T BUILT FOR THESE! If you ask me, Xonotic would mostly consist of indoor arenas with a few exceptions here and there, but also these exceptions maps would be mostly indoors.

I disagree with some of your points, but that is a topic for another thread.

About outdoor: I'm not sure if switching to a sandbox system would be a step forward or backward for darkplaces and I don't know how a hybrid sandbox/bsp would work either. Someone more knowledgeable needs to say if "sandbox system" should on the list or not.

P.S. I just noticed an interesting quote by one of the makers of Resistance 2: fall of man on CGSociety
“We had six character artists, working with up to six concept artists.” “The environment department was up to 30 people at the crunch time, and of course, the environment artists work all the way up to the completion of the game.”
- Samuel Sharit

Just 6 character artists, 6 concept artists and as many as 30 environment artists! I think we need much bigger teams for making maps!

Just 6 character artists, 6 concept artists and as many as 30 environment artists! I think we need much bigger teams for making maps!

Luckily I'd say the majority of the contributing community are proficient using netraident. Although I agree entirely the methods used to contribute maps are completely unorganised.

Great map making for Xonotic requires 3 stages of development, by people assuming 3 different positions. This is waht I have found to be a great method for producing high quality maps, and sharing the work load between members. It is also a self quality assuring system, with only the best levels being worked on.

1) Mapper: makes a level, with varing levels of quality of layout, theme, optimisation or visual apearance. Levels becomes popular due to high quailty layout. Which is often the case as levels that are just visually pleasing are played for a few weeks and don't catch on. This role can be done by any contributer in the community, although this is a great position for those new to mapping.

Possible examples:vede, MintOX, Grasshopper, Cortez666

2) Dev: Skilled mapper identifies good layout and increase visual appearance, optimises resources, and alters the theme to suit Xonotic. This is done primarily by dev's. Dev's are very busy with commitments to other areas of the game so there time and resources are scarce. This is good because there limitations of time becomes a type of quality control, with only the best levels being selected.

Possible examples:FruitieX, MireaKitsune

3) Detailer: Highly skilled mapper which contributes to Xonotic by primarily mapping. Paying an attention to detail, mainly increasing the visual appearance of the level. Once complete, the detailer passes the level back to the dev for final optimisations and submission. I like to take this role in level construction because it allows dev's to spend more time on other parts of the game.

Possible examples:Cuinnton ?????

If people assumed these roles, we could get mapping high quailty much more organised! I would also liked to see more Mappers assume the postion of Detailers!

Cuinnton, I don't apply to any of these rolls? D:

By the way, things are organized. It's explained here: http://www.xonotic.org/team/

For level design, it basically works this way: The mapper makes a map. The mapper then presents his map to FruitieX (the current level design coordinator) on whether or not it can be accepted into the game. FruitieX will give a yes or no, and if he gives a no, he'll say why (hopefully). If FruitieX denies the map entry into Xonotic, the mapper can go fix the problems that the map had and present the map again and again, until FruitieX hopefully gives a yes.

Developers are not always mappers. They can contribute to the game in a variety of ways, which can be seen at the link I gave above. And the role of "detailer" is nonexistant. If anything, it's merged with the role of mapper, because the mapper is responsible for putting detail into his own map.

Just straightening things out here, go back to your pesky engine complaints!

I'll check off some of the advances DP sports:
Realtime World Lighting/Soft Shadowing - most commercial Xbox360 and PS3 games don't feature this.
Traced reflections...i.e. Realtime Reflections - No console game has this.
Relief Mapping - Amazing. Period. Few console games have offset, none have relief.
HDR - Common, but beautiful with Bloom.

What Cryengine adds:
Radiosity, or light bleeding. One surface is lit by the light reflected off another. See Lightmark for example.
Good LOD - LOD that is efficient and works...mostly. Other games have even better LOD.
Some detailed water rendering effects - color warping, secondary reflection map.

What DP could use:
Postprocessing distortion - Think warped glass, heat, explosions, even water could be tricked with a 2 sine functions.
Radiosity - The maps would gain some realism.
Improved LOD and efficiency - Hard. Very VERY Hard.

Usefullness of this list:
None. Postprocessing Distortion(PD) and better efficiency would be nice, but those take time. PD would help improve performance and enable, for example, red planet to keep the warped glassy doors without the massive framerate loss. Explosions do look nice when they warp the area, and a shockwave would be awesome.

There's only so much weight you can put on a donkey before it refuses to walk at all

I would check 5 and 9 off. As shadowmapping (technique used for soft shadows) has been implemented for a while now. And DOF is just a shader which can be pulled from blender or various other sources.

"Postprocessing distortion - Think warped glass, heat, explosions, even water could be tricked with a 2 sine functions."

Is also just a GLSL shader, and can be done easily in Darkplaces AFAIK.

Edit see here: http://www.youtube.com/watch?v=29_0qL1gTOc

Are you sure that is a distortion?
I'm not sure how much of an impact the current ripple refraction effect has, but is it an extra render on top of the reflection render?

Uh, the water relfection on that video is not the GLSL shader, the blur is.

See, blur is a distortion of the color, different from a mathematical distortion of the image itself, especially a masked distortion.

Hmm... but both (edit to clarify: blur and masked distortion) are GLSL driven post-processing effects. What you write into the shader really doesn't matter so much as long as you got the GLSL support. Thus you can do easily both AFAIK.

I think it depends a bit on what your graphics card is capable though (DX9 vs DX10)... for the engine GLSL is GLSL though, no difference.

But I am not sure how far GLSL effects are exposed through QuakeC (CSQC that is) in Darkplaces... that could be IMO the main problem. I know in Cube2 you can embed GLSL into Cubescript and thus pass it over to the renderer.

Edit: Disclaimer.. I don't really know what I am talking about... since I am not a graphics programmer But I think what I wrote is correct.

Reflection as a post-processing effect? That's now how reflections work... (somewhat related, this reflection was done in postprod)

Reflection as a post-processing effect? That's now how reflections work... (somewhat related, this reflection was done in postprod)

No I was replying to the other comment. The reflection is likely a GLSL shader applied to a surface/texture

keep in mind the base of darkplaces is 14 years old

And the base of Linux is 19 years old...

(i.e. this doesn't mean much, DarkPlaces is purposedly able to run Quake, but it has received many many improvements on top of the original id engine)

And Yo mamma is 50 years old, yet.... wait lets not go there.