Texture Space, Theory and Practice.

[Archangel of Death]

I was specifically talking about not having to shrink stuff, because clearly Argh does not think the small (or not so small) loss in quality is worth the small (or not so small) gain in performance. If your going to accept my at least 60% statement... which you haven't denied yet, and if anything I think it consumes more of it than that... then 512w*512h*60% = 157286.4/256h = 614.4w minimum to not have to shrink stuff... and I hear video cards don't like off sized textures, and sometimes doen't like non-square ones much either.

However, I am not saying he shouldn't bother filling that space with something. Contrarily I think he should. Fitting a corpse in there is probably a better idea than the one I had too. Has to be done anyway!

[SpikedHelmet]

Just by estimation, by removing redundancies and smart organization I'd say it could easily fit in a 256x256

[Neddie]

Hey, guys, I've cleaned up your thread, I hope you don't mind.

[Zpock]

If you want to do light baking you can't have mirrors. I guess you could lightbake and then rearrange the UVmap to mirror some spaces tough. Mirroring can also be very ugly and break immersion if it's obvious. Things like dirt being the same on both sides look really stupid.

So choosing to not mirror at all is sometimes a good decision. There's a reason the pros did it.

Also everything is about how much work things take to do versus gain. If you want to be lazy in some regard to get other things done, why not. We only have so much time and energy before death claims our souls.

personally I think using texture space efficiently is a good idea. At least so you don't have huge gaping holes, altough I wouldn't obsessively make sure there's 4 pixels between every edge or something.

Or maybe I'm pretty obsessive after all:


These are laid out in a grouped manner too, I just made four packed squares of stuff belonging to different areas and combined them and then finetuned.

[SwiftSpear]

Something that I think deserves to be said...

In Evolution I use almost exclusively 512 textures. They aren't always very well mapped, bbut one thing is constant no matter how your UVMAP is done...

The resolution on the uv DOES make a difference. Even if you have wasted space, If you cram all that crap into a 256x256, your texture will be much less detailed.

It's the nature of the beast. Generally do this. Try not to waste space, but if it happens, don't jump off of a mountain over it.

Ok... really dude... There is nothing in evolution that needs too much more than 128x128... you mostly just use what is effectively repeating textures on units that are primarily geometric shapes. Huge portions of the units could just be mirrored and remirrored using a single tiling texture. This is just about entirely bad advice.

It's not about "cramming all that crap" it's about making space for the stuff that really matters. That's the entire theme of this thread here, although argh still makes the violent misconception that making more space for your important model realestate somehow equals more unused space in the UV, which is hard to see as anything other that straight up bad advice. Using a 512x512 texture for a unit that could get away with a 256x256 texture is telling the video card to dump texture memory into RAM 2 times sooner, and therefore meaning half the video performance with an equal amount of on screen units, get past 40 individual units in play or so.

Forb: Now, I really can't tell you how poor your models are compared to my models, since I have no models. That being said, I know enough about modeling to know that the majority of what you've done so far is roughly the modeling equivalent of stickman drawings. Don't presume you've figured out any good modeling practices yet. At the end of the day ya, don't spend hours and hours setting up elaborate UVs that then take you days to skin for moderate results, learning how to do efficient skins comes with practice primarily. Looks are more important than space, and using a higher res texture space is a much more forgiving learning environment. That doesn't therefore make it good modeling practice though. Exactly the opposite acctually. Ultimately UV mapping is puzzling, if you don't approach it from the perspective that you must constantly personally improve at it, you probably won't. Don't get used to bad practices. They make things easier, sure, they also make your modeling work a joke though.

[Forboding Angel]

SS you git.

I choose to use that style in evolution. A style that is liked by a lot of people.

I can model anything I need to, whether it be ultra low poly, or very high poly and complicated. I personally prefer the approach and style I had planed from the start. I'm not going to all of a sudden redesign everything because someone doesn't understand what a cartoon is.

I am very capable of setting up and optimizing UV's. You presume far too much. A style is a style, and I am not going to all of a sudden start introducing complicated and intricate models and textures into evolution. It is a style. A very cohesive style and one that I like a lot. The entire point of it is geometric shapes, it doesn't take a genius to figure that out. Moreover, it is my game to do with what I please. I have never asked you to like it and in as many ways as possible you have demonstrated that you don't. That's fine. I have no problem with that. You can state and restate opinions all you want but I could really care less.

You apprearently haven't seen more than 1 battle in evolution... There are generally half a zillion units duking it out. THe only slowdown comes from ceg stuff that I haven't optimized yet. Perfect example would be the bombs, they spawn waaay too many particles, but it doesn't break anything and is therefore a minor issue considering that I already have the new effects done and optimized, all that's left is to change 2 lines.

I am not worried about my modeling being a joke. I'm not here to wave my modelling epenor around. Syles are styles kids, deal with it.

You gonna start ripping into rcdraco or simbase now because their models aren't amazingly textured?

I simply said, that if you have a model that you end up wasting space on, there is no need to jump off of a cliff because of it. It happens.

Style: The combination of distinctive features of literary or artistic expression, execution, or performance characterizing a particular person, group, school, or era.

You know what, you are 100% right swift. End discussion. Now wasn't that easy? Course there is the slight problem that I don't care...

[SwiftSpear]

I don't mean to be personal forb... but really, I understand that there is a specific style you're going for with evolution, that DOESN'T connotate that I can agree that any evolution unit needs a 512x512 UV map. I apologize if the conclusions I drew were off the mark, but I really still can't wrap my head around exactly how that happened...

I haven't played too much evolution, no, my computer is on the verge of being unable to run the mod because it's too graphically busy. Not graphically good, just busy.

Also, I have no qualms with ripping on draco or peet's mods. Sure, simplistic is a style, that's why there's about a million stick figure animations on newgrounds, some of which are acctually quite impressive. Once again, that being said, the fact that it's a style, or that it's popular, doesn't mean it comes anywhere near to matching quality with something more completely done. Now... I've always been one to promote gameplay over graphical obsession... that being said, all other things being equal, I don't waver in my mind much which is better after that. There are a tonne of stronger arguments I could make outright against the 'style' if I really believed it was harmful, ultimately, I don't really see it as that big a deal, but becoming addicted to it is probably one of the worst things a game developer can do to themselfs. So forgive me if I preach just a little bit of caution.

[smoth]

Zpock, exactly, and argh doesn't do any advanced things like baking.

My main point is that the wasted space could be used for something at least. Argh always champions his efficiency, well texture space in memory is one of those things. He could make PURE more efficient and use less texture memory.

Forb, the thing is that you have very rudimentary textures which present little in the way of information. By and large the models are basic geometric shapes, not unlike what genblood used to use for features. We all know it is your style decision but you have to be understanding that we are not going to immediately warm up to in when all the other projects outside of nano, kp and fibre are going for somewhat believable models and textures.

As far as textures, I do not see any detailed lines, any pipes, greebles, dirt etc. you have no details that need that texture size. Most of the particles maintain higher detail levels then your textures. UNLIKE the other mods that means yours could have lower system requirements if you used small textures.

[rattle]

If you want to do light baking you can't have mirrors. I guess you could lightbake and then rearrange the UVmap to mirror some spaces tough.

I do bake mirrored stuff but it takes some time to prepare the model for it. Basically you have to cut a lot and offset UVs by a full unit so they don't get baked. And apply certain subsurf modifiers to all hard and smooth parts on your model which can be a royal pain, but this turns an average bake into an awesome one.

[Treeform]

why not just use blenders mirror modifier?

[Wolf-In-Exile]

If any of you want to make a career for yourself in modelling/texturing, or simply want to be good enough to actually have the right to call yourself a "good modeller and/or texturer", learn from the right sources, learn the correct standard practices. *hint* The Spring forums isn't that kind of place, judging from 'popular' sentiment about modelling/texturing practices here.

That being said, 2 basic points to keep in mind (which any competent game artist in the industry will tell you):

Poly efficiency matters. The reason has been given many times already so I don't see the need to reiterate it here.

UV map efficiency matters. The less wasted space on a UV map, the better the quality of the texture, because you can put even more detail in it.

[Zpock]

Yes the proffesionals in the industry are so awesome in their everytime groundbraking amazing work and we should all stop our feeble attempts at modeling and spend our time worshipping them instead.

You have a point tough, spring forums arn't the best place for modeling. Low polygon mechanical modeling is pretty hard to find good stuff on in general tough.

This is a good place:
http://www.geocities.com/paulthepuzzles/aardvarks.html

[Peet]

Laziness is a better excuse for boring and simplistic models than "style", imo

[Treeform]

Peet i agree - will use that form now on.

[Wolf-In-Exile]

Yes the proffesionals in the industry are so awesome in their everytime groundbraking amazing work and we should all stop our feeble attempts at modeling and spend our time worshipping them instead.

Your words, not mine. I never said anything about worship or stopping, nor do I remember addressing you in my previous post, but I guess that's moot now.
I'm simply giving sensible advice, and to me learning from professionals makes more sense than preferring to learn from amateurs.

I find your sarcasm rather self-contradictory, since you mentioned that "the pros" use mirroring sparingly in your earlier post, and then agreeing that ''I have a point''. If your modder's pride got bruised, well, the truth hurts? I don't think I need to remind you that constructive criticism is worth its weight in gold for any aspiring artist to improve.

[Pxtl]

Am I the only one who finds this flamewar hilarious in comparison to the cavalier attitude mappers have about huge files packed full of texture information?

Argh is wrong and right at the same time. He's wrong because he's arguing that proper sizing and space-usage isn't important in UVMapping (it's the most important). He's right because he says not to sweat it too much - if you can make 2 units with decent texture usage or one with perfect texture usage, your players will probably appreciate having 2 units more.

[Forboding Angel]

Am I the only one who finds this flamewar hilarious in comparison to the cavalier attitude mappers have about huge files packed full of texture information?

Argh is wrong and right at the same time. He's wrong because he's arguing that proper sizing and space-usage isn't important in UVMapping (it's the most important). He's right because he says not to sweat it too much - if you can make 2 units with decent texture usage or one with perfect texture usage, your players will probably appreciate having 2 units more.

Well I can see how you would get that conclusion, but mapping is an entirely different animal. A much larger animal for that matter as well

I 100% agree with your second paragraph.

@ peet, thanks, I will start using that as well. :)

[Zpock]

Yes the proffesionals in the industry are so awesome in their everytime groundbraking amazing work and we should all stop our feeble attempts at modeling and spend our time worshipping them instead.

Your words, not mine. I never said anything about worship or stopping, nor do I remember addressing you in my previous post, but I guess that's moot now.
I'm simply giving sensible advice, and to me learning from professionals makes more sense than preferring to learn from amateurs.

I find your sarcasm rather self-contradictory, since you mentioned that "the pros" use mirroring sparingly in your earlier post, and then agreeing that ''I have a point''. If your modding pride got bruised, well, the truth hurts?

The sarcasm was meant to be funny, sorry. I think you'r position is a bit extreme. There's a lot of crap (supcom anyone?) and good stuff produced by pro's and modders alike in my opionion. The only difference is they do it for money, so their sometimes uninspired altough I agree they generally keep a higher standard. Also professionals work in huge teams and throw a lot more hours at their work. As a hobby modder you might be better off cutting corners and saving time. And while you might be better off ignoring some of the guys "advise" here there can be some good stuff too.





The s44 is less detailed of course (since in COH you have fewer units), so it might not be the best example. I just think looking at the s44 model, or say MR:Ds or smoths or arghs isn't all that of a bad an idea. They do a good job with the time and resources they have.

[AF]

The end user doesnt care how you uv mapped your units, as long as they work and they look pretty theyre fine.

[Argh]

A lot of what I do is very deliberate, to save time. The less time you waste on a skin, the faster you can get 'er done, and move on. If it looks great, and is at spec (in this case, 512 for anything that isn't ginormous)... move on!

But there are other issues, as well. For example, if you end up with too high of a detail area on one small area, because it had the only shape you could fit into your puzzle, then it may make painting it a pain. Why have to have a special bevel, just for that piece, when you can keep everything at about the same LOD and make it look nice, fast? The pros are getting paid by the hour- I'm getting paid zero, yet I also have a workload.

Lastly, on mirroring:

It's bad to mirror if it's going to be overly obvious. That's pretty much my whole philosophy. I think saying, "don't mirror", or "always mirror" is extreme. Take it case by case. It gets easier and easier with experience, to figure out what should get mirrored, what shouldn't.

However, this has a mirrored body:



So does this, on the upper deck below the turret:



So does this, on the entire upper surface of the rear, most surfaces of the gun, all the wheels, and other areas:



I used a lot of subtlety to hide it, but packed as much detail into it as I could, by using a mirror. In fact, most of the work in PURE has mirrors of major portions of the bodies. This is one of the ways that I get everything to look good.

Oh, wait, last example: how many mirrors are on this guy? He's the uv space everybody's poking at, btw- so, does he lack detail, or what?



And, in fact, GMN, the guy who's doing most of the major modeling work, has gotten incredibly good at providing me with models that are ready to mirror all over the place, because he knows I'll use it.

How to make it not overly obvious, but keep the realistic look is difficult to explain, but I'll try.

Basically, what you want to do is to keep from making easily-identifiable marks. Those cue the human eye that they're looking at a mirror, and ruin the illusion.

What the hell does that mean? Well, it's not exactly a science, but I've found that marks that are vertical aren't easily identifiable. Marks that are purely horizontal, ditto. Marks that are highly diagonal, however, scream, "MIRROR". Dunno why, I think it's just the way our pattern-recognition firmware works

Also, major changes of hue or value do so, and major areas of high contrast, but that's something that should be obvious. Especially hue. If you mirror with, say, a black surface with yellow raised-armor surfaces, you'd better be very careful when doing your weathering. Whenever I'm not sure if I can get away with it, I fake it by using some muddy airbrushing instead, to provide some browning but confuse the issue visually.



Lastly, no, I don't do "advanced" stuff, like baking. Personally, I regard it as a huge waste of time- you can achieve equivalent results, faster, by just using bevels and airbrush. If Spring ever supports bump-mapping or normal maps, then hey, I'll happily use them- that's a completely different story. But baking? No. Huge time investment, no real benefit at the polycount we're talking about.

Pretty much my entire opinion about that topic.