http://i.imgur.com/UdTTMt1l.png 

nice lightning block in the middle. 

A few days ago Lunaran posted a pic of a really nice looking Quake4 map. Upon looking through the rest of the screenshots, this caught my eye.

It was sexy and I knew I'd seen it before, like here in Doom3 or here in Quake2.

These areas use tall, pillar type brushes that look structural. In between them you can place walls, eye candy, or whatever. It makes mapping flexible and it looks cool. Ordinary walls now have a bit of interest and the utilitarian look if the beams/pillars adds believability.

So, since I'm over a decade behind on mapping theory, I gave this idea a shot. This also gave me an excuse to give making textures a second try.

BAM.

It's generic as hell, but still cool. This is part of the big map I'm working on. The new textures will seem jarring, as there aren't any others elsewhere, but, meh.

Download the map and take a look for yourself, or Download the textures if you'd like to use 'em. 

nice textures! 

I think we found our next speedmap session. 

are those your textures?

If so, some advice:

1. edges of features should line up with pixel boundaries in a consistent way. For example, your floor texture has 4-pixel wide openings, but the space between openings seems to be 4.25 pixels, so the third slot is obviously .5 pixels off, and then by the 5th slot you are back on grid. This creates a lot of unnecessary and inconsistent antialiasing.

The wall vent in the top right screen shot, far right edge is an example of repeated features completely not lining up with the grid, so they are all blurry and weird looking

2. repeated patterns like vents and ribs also should be powers of 2 in size. This way even when mipmapped down they retain their crisp readability. Your floor vents are good because they are 4 pixels wide with 4 pixel gaps (once you fix that gap width) so even on the third mipmap they will be crisp. So try to make things 1px, 2px, or 4px in size, and gaps between features should be 1,2, or 4px wide.

3. All of the textures in your screenshot have almost the same color across the entire texture. It's a very flat, brown look. Adding some sort of contrasting material color can really help make your textures pop. It doesn't have to literally be a "Color", it can be more brown but of a different value (for example you have one panel that is slightly darker brown, try making it even darker so it stands out against the lighter brown you are using everywhere else. It can also be the same color but a different saturation, for example taking that brown and fading it almost to a dirty grey color. And of course you can add more of a real color too. The red pipes in some quake2 maps is a good example of color that pops well.

4. finally, a good way to maintain color harmony with different textures -- once you have more than one color :) -- is to make sure that in any given texture, some pixels share the primary color of other textures. The quake palette helps with this of course. if you look at the quake metal textures, the grey textures have blue pixels and the blue textures have grey pixels, so the eye sees those repeated colors and it pulls the whole scene together better. Textures that don't look good together often have no common colored pixels. 

@necros - Thanks!

@FifthElephant - IF you do, you might wanna download my first attempt at textures and mix them in for some much needed contrast.

@metlslime - Thanks for good advice. I fixed a few other textures that suffered from the misalignment issues you mentioned, but then just got lazy... I should fix these. What they would look liked mipped never even crossed my mind, nice pointer.

And yes, they're all exactly the same color. I should do something about that. hehe... looks like Alien3.

And your final point is very nice, I'll put all your advice to use my next texture making go around (third times the charm, hopefully).

Tell me what you think of this idea, and if it's not a very good one on a technical level:

I planned on greyboxing an entire map, focusing purely on gameplay. Then making a single, giant texture for each wall, all uniquely drawn. It's basically like megatexturing quake style. I think this could yield some really cool results, however, my technical ignorance causes me to wonder how bsp/engines would deal with that.

Just imagine: a map without any tiling textures, just a big, uniquely drawn canvas full of detail... 

It's worth trying but you should do an early stress test to see if it's even viable -- there are a number of steps in the texture loading pipeline that probably never had to handle 100s of megs of texture data. You'll probably need a giant -heapsize, for example.

Also, map load times will go up linearly with the amount of texture data to load. I like how quake takes 2-3 seconds to load a map. Changing that to 30-60 seconds doesn't sound like fun to me as a player. 

"Tiling textures are really just a crappy form of compression." -John Carmack

"No, they're a great form of compression." -Lunaran 

Carmack 1, Lunaran 0.

yes killpixel studying the level details of lunaran and others like him that are good at architecture is the best way to improve your mapping skills. 

Speedmapping could be a great idea with this set :) Totally Doom-base style (which I love).

<3 

Crappy form of compression, great form of workflow. 

Glad ya dig it :D I rather like the doom base look myself. 

Carmack will decide on a viewpoint that he finds "pure" and then use that as the justification for a huge amount of decisions that must be correct because they're derived from a pure assumption.

We all heard the story he kept trotting out about "being able to see what brick will pop out of the wall" in a cartoon because the brick is cel shaded and the wall is a painted background and that's why lighting should all be unified and real-time. Great, sounds terrific. Lighting wasn't unified up to that point, of course, because it allowed you to make pretty deep case-specific optimizations that made your game actually run fast, a lot faster than Doom3, but now you've talked yourself out of that completely for pat-yourself-on-the-back points. Anyone else would be called naive.


Tiling textures are, from a certain very grognardy point of view, a form of compression. Sure. And if your goal is to perfectly simulate the real world, yes I guess it's not 100% accurate. In the context of the reality of having to make a video game, however, it's really pretty damn good.

1) Very simple to implement. Already supported by 100% of hardware!
2) Artists intrinsically understand it and can work with it easily.
3) Players barely notice, and environment artists barely notice once they've started playing and stopped looking.

I no longer take the view that the goal of a video game is to perfectly simulate a real world, because doing so just so you can fill it with absurd gunfights and guys who wear tires for hats is dumb. Building a game with nothing but unique surfaces and textures that the player will then drive past at 40 mph in a dune buggy is dumb. 

but that concrete wall looks literally the same as THIS concrete wall!

MY IMMERSION! 

I planned on greyboxing an entire map, focusing purely on gameplay. Then making a single, giant texture for each wall, all uniquely drawn. It's basically like megatexturing quake style. I think this could yield some really cool results, however, my technical ignorance causes me to wonder how bsp/engines would deal with that.

Its a trap! Unfortunately the only person who is really going to notice all this detail is you because you created it. Most players will notice the first room, but after that it, it will just become a blur of detail overload.

I never really understood the fascination with the metal Doom3 texture set, its mostly fixed sized panels painted over generic corridors. I really preferred the wide atrium, outside areas and underground temples to the constant alpha/beta/gamma lab corridors sections. 

Taken in context, extolling the virtues of megatexture, his comment makes perfect sense.

But, sure, if you want to talk realities of shipping games you certainly wouldn't want to listen to John Carmack. What's he ever done? ;) 

Let's also bitch about how Quake maps are slower/harder to build/longer to compile than Doom maps because of all the bullshit innovations. 

All realistic games suck. 

I see your point, and you may be right.

However, I quite liked the (macro) detail and uniqueness of RAGE. It didn't feel like detail overload and the environment always had my attention... of course, that could be because the game had absolutely nothing else to offer.

The idea of greyboxing is to give the level designer total freedom in terms of gameplay. Later, the artist can "make it work", making giant, area specific textures which gives the artist flexibility needed to do that.

So far, my little test is coming out much better than I expected. The mapping is easier, the texturing is easy and it's looking really cool.

I'll post my results in a week or so. 

Number of games (non-id & excluding expansion packs) shipped on Quake3 engine licenses, according to Wikipedia: 15

Doom3 engine: 6

Rage engine: 2


Do the Unreal or Crytek engines only support one lighting path? They must, if it's such a brilliant streamline. How about fully virtualized textures and geometry? 

mean this?
https://www.youtube.com/watch?v=00gAbgBu8R4
This is so 2011. 

wasn't that video panned as just a play for investment capital? 

Well, you know, billions of voxels are really just a poor form of compression. What we need are actual digital atoms. 

Seems like a pretty shitty form of compression, although then again we seem to be getting some hints that it's more efficient than we think