We need to talk.

[RyokoTK]

Ever since the advent of functionality to transcend beyond single texture set definitions, more and more maps have been cropping up that are using textures from many sets at once. This is obviously the point of the functionality, but the only problem is that the end result is usually hideously ugly. I'm not exactly a trained art professional, but I can tell you why they're ugly with relative ease. It boils down to two simple points:

1) Textures are designed to work well within their own sets. That's why they're called sets.
2) Texture sets are inherently unique from each other. Otherwise, what's the point of choosing one set over another?

Having separate sets was spurned from technological limitations, but Bungie came up with different sets to represent different types of environments. If you want a Pfhor spaceship, you use Pfhor, and only Pfhor -- because that's what it's there for. There are no other sets that are intended to represent Pfhor ships, and the Pfhor set isn't intended to represent anything else. The other sets are a little more diverse, but the point is this: there is rarely any designed maps where two separate sets would work equally well. The individual textures work well with each other because they're all designed to represent the same type of area, but they're unique from textures outside of their sets because they're designed to represent different areas.

The end result is that it's difficult to just jam two sets together and expect it to be visually appealing or artistically coherent, and yet this is precisely what people are doing. These maps end up being ugly and garish, and the same map would work much better if it had just stuck to a single set. This is one of the cases where the visuals impact on the overall experience -- besides the fact that it's difficult to convey a certain environment when you use all five sets, the distracting colors and style changes make it difficult to immerse yourself in the experience.

Having multiple texture sets is a great thing, in my opinion, but only if one can use them properly. It is much more effective to use two sets in two separate rooms than it is to jam them together. This keeps a map from looking samey, and also lets a single map be much more diverse on the grand scale -- certain textures require certain architectural styles and habits to be most effectively used, and if you can properly segue from one set to another, you can have two distinct regions in one map with a natural transition.

There are three visual elements to a texture that distinguish it from other textures:
1) Color
2) Pattern
3) Material

For two texture choices to mesh properly, these elements need to be the same or similar, the difference needs to be sensible, or there needs to be a good reason for them not to be. It's hard to use these two together, because they share no common elements:


On the other hand, these two textures work reasonably well because the color choices are similar and they appear to be of a similar material. The themes can be shared.


When you want to mix your texture sets together, make sure that in any given area your texture choices blend as well as possible.

For example: say I want to make a map, and I want to use three separate texture sets to do it: Water, Sewage, and Jjaro. Water and Sewage have a fairly unique synergy between them, because Water's color themes and Sewage's color themes are close enough that you can mix-and-match better. It's easy to actually blend these texture sets to great effect, as long as you know what you're doing. (Sewage's green textures are better than the blue ones as far as this goes.) That doesn't mean that you can use whichever texture willy-nilly, just that you can do so sensibly. The Jjaro textures do not enjoy this synergy with any set, and therefore, if I want to use Jjaro, I should limit it to a single distinct region of the map. Say you want some sort of ancient Jjaro facility buried underneath a S'pht building, I dunno. The point is, they don't blend together, but you can still have a transition, and with a little work you could still get it to seem natural and sensible.

There are a few exceptions to this policy:
1) The Pfhor set doesn't really work well with anything. It hardly works well with itself. It's tough to transition into and away from this set, even, because it's so... well, alien. It's good in that regard, but that doesn't make it any more usable. I have never seen the Pfhor set integrated well with any others.
2) Some textures are unique, but they're great textures on their own and can be used as decoration in many situations. You can't paint the walls of an entire room with it, maybe, but you can use it extensively to decorate pillars, stairs, sides of panels, and so forth. Below are some examples:


In summary: learn to use multiple texture sets better. Rooms that use a different texture for every single side are not visually pleasing.

[treellama]

Perhaps the map contest should require the aesthetic use of two texture sets?

Thanks for this guide. Scarbo was originally going to have rock textures from all three planetary sets, but they really didn't mesh as well as I'd hoped. I'm sure it's possible, I just wasn't good enough to pull it off.

[RyokoTK]

I've had success combining the water rock textures with the lava rock textures, more so in darker rooms than lighter ones.

That blue rock from sewage is disgusting, though.

[Sharkie Lino]

Thats most of the reasons that I like single-set maps, because of the environmental feel but also because they easier to work with when you just have one texture set to use, at least in my opinion.

Now of course when used properly, they can make a map look great and give it a very unique feel.

I remember some of those maps we were playing on last night. If I remember correctly one room had a combination of textures from Jjaro, Pfhor and, Water. That didn't look too good...

[RyokoTK]

I remember some of those maps we were playing on last night. If I remember correctly one room had a combination of textures from Jjaro, Pfhor and, Water. That didn't look too good...

That is the map that inspired me to make this topic. I don't remember the name. The architectural quality was pretty high, which is why it's so sad that the texturing was so abysmal.

Edit: "Before These Crowded Streets" is the name of the map. Who's responsible for this?

[$lave]

That's Windbreakers latest map. The texturing is similar to his last map, which also used All-Textures and was released in sovereignty.

Also, I've only recently (sadly) been becoming more and more aware of this, especially since I released Riki. Being a large map, I had room to use tons of textures, but I continually feel that I should re-texture most of the map. I guess part of it, is that I (and maybe other people) feel that when they use All-Shapes I should be using it as much as I can, even though I know it wasn't intended that way.

One other thing, the blue rock from the sewage set is hideous, but I actually REALLY like the hi-res (green) version. Should people keep in mind what textures will look like hi-res, or stick to what looks good in software? I ask this because, in my experience, there are some combinations of textures which look really nice in software, but not with the enhancements, but also vice versa.

[Crater Creator]

I think visual_mode.lua will change not just how people texture, but how they map as well. Before that, it was fairly impractical to really design a map with multiple texture sets in mind. In EMR, we would do a handful of texture swaps in a level, changing a texture we didn't need via mml to another one we did need for that level from another set. But it was tedious, and while in Forge you just had to hope it would look good, and align properly, when you ran it in the game.

Now that we can use any texture anywhere, mapmakers have more options. For those doing total conversions, I would look into rethinking texture sets entirely. Instead of coming up with twenty-odd textures that match a particular liquid, you can use different criteria. For example, have texture sets like Terrain & Liquids, Interiors, Detail & Trim, and Panels. Ryoko's suggested techniques still apply: don't lay down a grassy floor inside a building where it doesn't make sense. But I think for those doing all-new textures, redefining the texture sets (and scenery sets to match) could be more convenient, and ultimately lead to better maps.

[treellama]

For those doing total conversions, I would look into rethinking texture sets entirely. Instead of coming up with twenty-odd textures that match a particular liquid, you can use different criteria. For example, have texture sets like Terrain & Liquids, Interiors, Detail & Trim, and Panels.

Loading a texture set requires using RAM for every replacement texture in the set. You don't want to do it for no good reason. So, organizing textures based on level environment (the way Bungie did it, go figure!) makes more sense.

[Crater Creator]

Granted. Every dev team will have to weigh the increased convenience against the decreased performance. I predict over time, the use of multiple texture sets in a level will become more commonplace. So if it becomes somewhat of a norm (and I haven't noticed a performance hit, using the .dds TTEP), it's worth re-thinking how textures are grouped by set.

[RyokoTK]

it's worth re-thinking how textures are grouped by set.

How is it actually worth it? It makes little difference how the textures are organized except for the RAM issue, and of course the thematic issue that is the point of this topic.

[Windbreaker]

I've been planning to scrap that map.


With my few latest maps I've been trying to base the map around a single texture set, only with 3 or 4 other textures throughout the whole map. But the placement of these textures is fairly sparse.

I guess my original goal with those two maps I created was that I was trying to explore some different avenues and create some new interesting ideas. With my latest map (unreleased) I think I've had a better view and become more experienced on how and where to place textures that should match up pretty well. My mistake with those previous maps was that I just went hog-wild with all the textures at my disposal. You need to use all these textures in moderation.

[Crater Creator]

There's the performance difference in RAM usage, I acknowledge that. There's also the pragmatic difference in how people texture. We've always textured in Forge with one texture set, and that texture set by necessity was themed around a liquid. That setup has been adequate, but not without sacrifices. For instance, if you make a water level with Infinity's shapes, your terminals all have to be half a world unit tall, and they have to have a red symbol that's best associated with Tycho or the Pfhor. Sometimes, the mapmaker would've preferred a terminal with a green symbol, or one that would fit in a 1/4 WU high panel. Once mml allowed you to swap textures, the mapmaker could use any style of terminal in any level. But there was still cause to consider if that was worth it, because you couldn't see what it would look like in Forge, and because of the performance hit.

Visual mode.lua removed the former disadvantage, though not the latter, so it's now more feasible than ever to organize textures sets according to how you texture. For those that find it most intuitive to work with texture sets as they are now, great. But another mapmaker might, for instance, want to do texturing in several passes. Suppose he prefers to first lay down outdoor terrain textures whose alignment isn't important, then the interiors of the buildings, and lastly all the switches, rechargers, and terminals. He would then benefit from a 'terrain' texture set, an 'interiors' texture set, and a 'panels' texture set. Or maybe he's doing a team-oriented netpack, and he'd like to be able to access all the textures featuring red or blue colors at once.

These are just examples; developers will no doubt conceive of other ways to group textures which they find most convenient. My point is simply that dev teams in the early stages of their projects can now consider how splitting their textures into sets in different ways may improve their efficiency in texturing.

[DeepThought]

I think this topic should be stickied.

EDIT: And I kind of agree with Crater Creator that multiple texture sets is more convenient for a scenario. For example, my former scenario Dookethon only needed four texture sets. However, the texture set for duke nukem reached its limit and I could not fit any more textures. I thought that it might be a good idea to make another texture set for level-specific material, similar to what EMR did. For example, if the level was a store, then you can have a sign that says "Bob's Supermarket" or something. However, I thought that that would be pointless, because it would be way too hard to implement without actually seeing the texture when I place it in forge. VM.lua solves this problem.

Also, I was considering an idea of multiple texture sets being used as Crater Creator. For example, one set would have Ground (grass, mountain, sand...), Panels, Building, Cave, Underwater/snow. That way one can use all the textures they need for a scenario without feeling limited to 130 texture per set. I am aware that this would cause slowdown, but I have played multiple texture set levels smoothly on a computer made in 2001.

I have actually seen some maps that take advantage or multiple textures and look cool. I remember this one level that was capture the flag and the side for one team implemented blue textures from the water set and the other team had red textures from the lava set, and it looked good.

[irons]

I have played multiple texture set levels smoothly on a computer made in 2001.

Was this in a scenario that used high-resolution replacement textures? That is what Treellama means when he worries about computer performance. I don't know if EMR changed since the last time I played it... but at that time, it didn't even use the faster DDS textures (instead opting for JPG and/or PNG). That means really long level load times. Even with DDS, the uncompressed textures can still take up huge chunks of RAM and VRAM. Assuming five texture sets of 100 textures each, with 512x512 textures, the uncompressed size is over 100 MB of (V)RAM, which is ridiculous on older computers. Bump the size up to 1024x1024 and you have 512 MB. My poor laptop couldn't even load the landscape palette in VM.lua because it was running out of space, and that was just with the regular old textures.

I remember this one level that was capture the flag and the side for one team implemented blue textures from the water set and the other team had red textures from the lava set, and it looked good.

All of Magenta Filter (the "official" CTF pack) had this scheme.

[Crater Creator]

I don't really expect a computer from 2001 to handle 1024x1024 textures in the first place. I can't think of other games which use such a high resolution and still have system requirements that a typical 2001 computer would meet. Most Quake 3 based games, for instance, only use 512x512 for the largest textures, like large terrains and skyboxes. I think we've bumped up texture resolution so much because it's easier to improve graphics through high-res textures than it is through 3D models or other things. If you're curious, EMR's textures are all 256x256 JPGs (much to my complaint; I pushed for lossless PNGs), except for the landscapes which are DDS. Version 3.1 is supposed to be all .DDS textures.

Anyway, what I mean to say is that a scenario developer should already be weighing the costs and benefits of using hundreds of high resolution textures. I think we all acknowledge the performance impact of multiple texture sets. What Ryoko was getting at, and what interests me more personally, are the artistic considerations.

[The Man]

On the whole I agree that using Pfhor textures with, say, water textures doesn't work that well, but at the same time I think it's possible to use the contrast between texture sets to create an effective map. I actually find Jjaro and Sewage work rather well with each other because a lot of the textures use the same basic format - in fact, if you compare a number of similar-numbered textures between them the layouts are so similar that they practically come off as color swaps of each other (though there's a lot more difference than just hues). Just for the fun of it I've replaced Sewage maps from the original Trilogy with the corresponding Jjaro textures, or vice versa, and only had to do a small amount of retexturing for aesthetic purposes afterwards (nearly all of it confined to two or three textures which are discernibly different between the two sets). Actually I'm using a highly expanded version of a sewage-textured "Ne Cede Malis" in the scenario I'm developing and it's come off quite well.

In short, I agree in principle with what Ryoko says in the OP, but I think it should be taken more or less as a general rule rather than something to adhere to strictly.

Now that visual_mode.lua has been created, is it possible to use textures from more than five collections? I haven't tried or even researched whether the AO engine is even capable of that yet.

[Volcanon]

highly expanded version of ne cede malis? Isnt that map already giant?

Anyway, Jjaro and Pfhor might work well together, as they share similar patterning. But then again, there is something to be said for creativity created from restriction - Most maps that were restricted to one texture set look perfectly fine.

[The Man]

It's not that big compared to other levels in the solo scenario, much less some of the levels from Eternal and Rubicon. The new version adds about 400 polygons, currently.

[chinkeeyong]

It's not that big compared to other levels in the solo scenario, much less some of the levels from Eternal and Rubicon. The new version adds about 400 polygons, currently.

That looks just like Ne Cede Malis...

[localwarming]

That looks just like Ne Cede Malis...

But, um, you know, with things added on.

[irons]

I thought so, too, until I saw that he added more to the north:

[Volcanon]

Ha I confused ne cede malis with that last stage with the gigantic space station whose name I forgot.

[The Man]

Aye Mak Sicur. I bet you're not the first person to make that mistake, so don't feel too bad about it.

But yeah, the first half of the map is fairly similar, except that it's a vacuum level and thus there are now oxygen rechargers next to all the pattern buffers. The second half is (obviously) completely new, though there is a bit of architecture I copied directly from the first half.