Noob modeling/texturing thread

[demagogue]

A little tangent. I've been playing around with Endorphin recently (the procedural AI animation app). Apparently, the animations are exportable. Have you guys considered or actually used it for AI animations? I'm just getting started playing with it, though, so I don't know all the ins and outs, and may not know exactly what I'm talking about, but so far it's a lot of fun and seems pretty powerful in generating whatever stock animations you might want ... and I'm just curious if what I'm playing around with and learning could be useful to the cause.

Edited January 3, 2008 by demagogue

[greebo]

Ok, then I misunderstood "repository". I thought it was just a place to store the highres textures for future alterations. But you still need the specular map as a TGA on Ultra High. I just tested it... Don't want to picky, but I wrote a little replacement for the list...

The Dark Mod is running on "High Quality" by convention. We should have mentioned that earlier, but we're not supporting the Ultra-High setting, because it would make our repository a few hundred megabytes larger for a barely noticeable quality win.

 

Rest assured, that we will ship with a readily fabricated DoomConfig.cfg file, so that end users won't have to bother with playing around with the settings. It will work out of the box.

 

@repository: We're using SVN for our internal working copy, so repository is the term for the place we store our files in. We've got multiple repositories.

 

Another thought now. Since the only difference between ultra high and high detail apears to be the use of uncompressed textures, why not create a real ultra high mode, which doesn't limit us to dimensions of 1024, but 2048 instead for example. Nothing would have to be changed. The TGAs would be 2048x2048 or something like that, but the compressed only 1024x1024. Ah wait, we'd have to make a smaller normal maps as well for the lower settings then. So it would be a bit of extra work, but so damn detailed... =)

2048x2048 textures are above our convention, because they take an awful lot of memory. 1024x1024 is really enough for 99% of all the purposes. The only exception would be some ultra-large terrain model which spreads over half the level, but it's always possible to create models using multiple shaders, so there is no real need for textures this large.

[STiFU]

So for my testing purposes I should just switch to High detail and leave those TGAs out, I guess. Ok then, thanks for you answers. But you should really clarify this informations a little closer in the wiki, so that this question might never arrose again...

 

And I got a new question already. Is there something like a shader manual for Doom 3, similar to the one for quake 3? I googled, but couldn't find anything. I need to know the parameters for the material files, especially the ones to create a full-bright material, like a lit window or a skybox. I know there is a special technique for the skybox, but it should be possible to create a full bright material anyway, shouldn't it?

Edited January 3, 2008 by STiFU

[Baddcog]

We don't have a full list of parameters for materials as far as I know. I don't want to get into that at this stage really, I just copy/paste from the wide variety of materials we have already.

 

This is a material I use on my steambot lantern, the Blend Add stage makes it bright:

 

lanternbot_lit
{
 glass
 noSelfShadow
 noshadows
 forceoverlays
 qer_editorimage models/md5/chars/steambots/bc_lanternbot

 {
 map models/md5/chars/steambots/bc_lanternbot
 }

 {
 blend add
 map models/md5/chars/steambots/bc_lanternbot

 red 0.7 // with the rgb values you can regulate the intensity - 0 to 1
 green 0.64
 blue 0
 }
{
	blend   specularmap
	map	 _white
	rgb	 0.5
}
}

 

a link to ID's material page

 

I think the only thing not listed is our specific surface types. I'll try and remember to find that list for ya.

[STiFU]

I got the list already from your wiki, but thanks anyway. That link you posted is very helpfull, so thanks again...

 

For quake 3 there was a neat tool for creating the materials, called Quake 3 Arena Shader Editor (Q3ASE), with a proper preview and all. I wonder whether the author could be bothered to lay open the source to us, so someone could add the doom specific features to it? Q3ASE enhanced my working with q3 shaders very much, because this way you can just learn by doing... The best thing would of course be to integrate such functions into DarkRadiant, since you got the preview already implemented and would only have to write an interface for creating the materials and a function to reload all or just the changed materials in DR. (Keep in mind that I am not demanding anything here, it's just a gentle suggestion, which in my oppinion would enhance DR to a great amount)

 

About that full bright thing, I actually had a very similar solution and now I realized that it was just the preview that didn't work. I should test things ingame more often...

Edited January 3, 2008 by STiFU

[greebo]

A material file editor is indeed filed in our issue tracker, but that's not going to make it in very soon. It's a lot of work getting this right, as you can imagine.

[Maximius]

http://artofjoe.blogspot.com/2007/10/digit...g-tutorial.html

 

I found this interesting tutorial for Blender, a little advanced for me but some good tips in there. This may be o ld news but worth repeating.

 

Especially check out the "Decimate" modifier, it allows you to reduce polys rapidly. I've reduced the alembic down but I have to do some "surgery" with Decimate because some parts look fine and others look too low res. I have a nice version at around 1010 faces but it could be lowered with some detailed reductions elsewhere. 500 is a good base then the neck and such needs to be brought up a bit. Maybe a 700? BTW for noobs when you use the decimate modifier it tells you the face count, its the only way I know off the bat to find out.

Edited January 4, 2008 by Maximius

[OrbWeaver]

Especially check out the "Decimate" modifier, it allows you to reduce polys rapidly.

 

It will also screw up your nice, symmetrical, well-designed mesh structure by turning it into a set of randomly-divided triangles which meet the required mesh count with as little change from the original shape as possible. It's useful for terrain or other "chaotic" objects, but you won't want to use it for anything which is supposed to be regular or symmetrical (like a human).

[Maximius]

It will also screw up your nice, symmetrical, well-designed mesh structure by turning it into a set of randomly-divided triangles which meet the required mesh count with as little change from the original shape as possible. It's useful for terrain or other "chaotic" objects, but you won't want to use it for anything which is supposed to be regular or symmetrical (like a human).

 

Sigh.

 

 

Ok I got it. This whole kit only costs 500 faces.

 

http://aycu21.webshots.com/image/39660/200...44729271_rs.jpg

 

Reading on importing models to D3.

Edited January 5, 2008 by Maximius

[Crispy]

No application that does something complex (such as 3D modelling, music production etc) will seem intuitive if you are used to a different application. There is a good article about this on JoelOnSoftware which I can't be bothered to look up right now, but it points out that even the supposedly well-designed and intuitive Mac interface seems clunky and strange to Windows users, because it is unfamiliar.

Yes, I've read the article. Point taken, though my transition from 3ds max from Maya was a lot less painful than this one. Maya still presents everything to you up-front; everything you can do is in a menu, and the menus are easy to navigate. Blender tries but doesn't quite get there.

 

I realise that Blender is designed to be mostly used via the keyboard, and there's nothing wrong with that; I used keyboard shortcuts a lot in Maya too. But I was able to learn them at my own pace (rather than being forced to learn them up front) because everything important is either in a menu, the left toolbar, or one of the well-laid-out settings panels on the right. Blender does have settings panels, but their layout is very confusing (even just finding the right settings panel is fraught with difficulty) and chock full of opaque abbreviations. Worse, the documentation often tells you to go to controls that don't exist any more / have been moved. The screenshots don't match reality.

 

It would probably help if the documentation said "okay, here's the menu item for applying a texture, here's where to go to find the menu item, or if you're using it a lot you can use the keyboard shortcuts". It doesn't. It says "press Ctrl-Shift-4, W, X, V, and connect the texture to the boondoggle" (assuming that you already know all about connecting textures to boondoggles). Don't lecture the keyboard shortcuts at me; educate me about the layout of the GUI so that I can find them out myself when I'm ready. Don't give me a fish; teach me how to fish.

 

"But the philosophy of Blender is to have one hand on the mouse and one hand on the keyboard!" say the purists. Tough. If you want to make it easy to use, you need mouse-only control for the newbies. Encourage them to graduate to keyboard control, yes, but don't shove it down their throats before they're ready. That approach is a blueprint for failure.

 

Find this control on the bottom bar of the 3D view, and set to "Closest", or hit SHIFT-TAB (which activates "Median" not "Closest", but I have no idea what the difference is).

 

 

Then select the vertex you want to move, hit G, drag over the vertex you want to snap to and hold down CTRL. Instant snap.

Oh, so that's where it was hiding.

 

Took me a moment to figure out that you have to be in Edit mode to do that (obvious only in retrospect).

 

I'm not sure what you want to do here, do you just want to render the scene from a different position? Try using SHIFT-P to create a sizable, draggable preview overlay in any 3D view, which responds to the underlying rotations or translations of the 3D window.

Aha. Now that's exactly what I needed, thanks. Knew there had to be a way.

[STiFU]

I prefer to work on a low poly level from the beginning. On the one hand it's cleaner and on the other, it's much easier to optimize it later on. Also I think over all you still save some time, 'cause you woun't run into nasty problems that could possibly be caused by such a triangle reducing feature.

[Baddcog]

I like to try and model as close as possible to the final outcome also. I do know how Max's optimize works so I often use it to my benefit but I start out modelling knowing how I'm gonna use it, it's not just a randomm after thought.

 

Also, if you start out with a certian budget in mind you can usually get pretty close once you understand polycount. I've gotten to the point where I can look at objects that someone else made and get pretty close to polycount off the top of my head, and also looking at an object and thinking 'I can probably do that with so many polys'. Of course that comes with years of practice too.

 

But I agree with Blender not being too user friendly, I haven't used too many 3d programs but Max has got a really clean user friendly layout. I still think I use my mouse 99% of the time, I do know a few keyboard shortcuts but the menus are handy and easy to use. I also notice that Max, Maya and other programs tend to use similar terminology for alot of stuff. Seems like Blender uses names I've never heard of and that don't really make much sense.

Like decimate for example, sounds like throwing a nuclear bomb at your object and hoping for the best Optimize sounds alot closer to what the actual function should be. I often want to optimize stuff, but rarely would I want to decimate something I've worked hard on.

[demagogue]

Decimate literally means to make something 1/10 its original size ... like saying one-tenth-ify.

That was the factoid on an episode of Monk to show how anally literal and quirky he was, to correct someone on that. That's the same feeling I get from Blender using it.

 

By the way, nobody responded to my post on Endorphin. Has anybody played with it? Or motion builder? It seems like it should come with the object-making territory.

Edited January 5, 2008 by demagogue

[Baddcog]

Never heard of endorphin, kindof set in my ways right now, learning DR anyway.

 

I believe some of the animators are using motion builder, I haven't tried it though.

[demagogue]

Is "DR" something obvious I should know ... oh dear.

 

For the record, and because I'm having fun with it I feel like describing it a little, Endorphin is a system where you give the AI a script and a set of constraints and they basically animate themselves with the cognitive/biomechanical tendencies of a real person (and hopefully someday animals) ... e.g., they shift their weight or side-step to keep their balance while they're acting, depending on what forces are hitting them ... so it looks more natural and is faster to produce than doing it by hand. It's something to use along side motion builder.

So far I'm really liking it.

 

A demo:

 

There are a ton of YouTube videos of people playing around with it, but mostly just of goofing around.

 

Edit: The reason why it sort of excited me ... I was thinking about conversations in dromed, you line up a queue of actions an AI performs. I thought it would be esp good for that kind of thing.* Also some of the more "acted" events like an AI dying, getting hit or knocked out, slashing their sword ... but even simple things like the walk-cycle can be done fast and look natural.

 

* (Speaking of which, I'm not sure how TDM handles the equivalent of conversations in dromed ... do you have a system like that, or is something just handled by general scripting?)

 

Edit2: They also have a middle-ware version called Euphoria which integrates the reactions directly into the AI, so e.g. they naturally shift their weight to keep their balance in-game on the fly. Endorphin just cans it in advance and plays that when called. While that would be pretty cool to see, esp for things like melee combat, I'm thinking that's pretty far beyond what you guys want to be doing for TDM (to say nothing of licensing issues, even if it is non-commercial). But even Endorphin by itself is pretty powerful.

Edited January 5, 2008 by demagogue

[Baddcog]

That sounds pretty cool, I had a hell of a time making a decent walk anim for the rat. That's why I'm leaving the human anims to the more experienced animators. It can be fun but it can be very frustrating some times too.

 

DR is Dark Radiant I assumed that was becoming common knowledge. I've messed around a little here and there but most of my time has been modelling, now I'm spending more time trying to learn the editor. I'm already liking it ALOT more than Dromed, but like Dromed there are limits as with every engine. The learning curve isn't that bad but it's good to mess around with it and learn the ins and out before jumping straight into a building a mission.

The cool thing is being able to use brushes to create objects. I created some barred windows and a large rusty barred gate with brushes and patches, also did some cellar doors with metal hinges, all in the editor, then you can add door props. So if you want a new crate, door, lamp, ect... you don't even need a modelling program. You can deffinately do more in a modelling program, but for simple stuff you can just whip it out in the editor.

 

I'm sure you can script any behavior you want with the AI, and I'm sure that's how it's gotta be done as we don't have a Conv system like Dromed had. I'm sure basic scripts will be developed in time by the team or by authors that everyone will be able to reuse. I'm counting on it in fact scripting isn't a strong point for me.

[demagogue]

DR is Dark Radiant I assumed that was becoming common knowledge.

 

Haha ... of course. I was trying to think of some modeling plug-in or animation app or something, from the context, but knew it was going to be right in front of my nose. I'm sure it will become second nature when I start using it, too.

Edited January 5, 2008 by demagogue

[Crispy]

I was thinking about conversations in dromed, you line up a queue of actions an AI performs. I thought it would be esp good for that kind of thing.* ...

* (Speaking of which, I'm not sure how TDM handles the equivalent of conversations in dromed ... do you have a system like that, or is something just handled by general scripting?)

I'm not familiar with how conversations work in Dromed, but Doom 3 does have a system where you can control actors to a limited extent by placing nodes in the game and connecting them up. It's part of the patrol path system for NPCs. We haven't gotten around to integrated it into our AI yet; it should be straightforward, just tedious. It would be a useful introductory task if a new coder were to sign up. Hint, hint everyone.

 

If that fails, there's always scripting, which is well suited to that kind of thing.

[Baddcog]

It would be cool to have some base set-up for those of us who can't script.

 

Dromed had a drop down menu type system, each window had 10 options. Option 1 would be play motion...., option 2 would be play sound file...

If you used those 10 options you could start on option set 2 (with 10 options...)

I think there was a total of like 7 sets, then you'd have to trigger a new Conv.

You could run thru them fairly fast if you had 2 Ai talking back and forth because you might only use 1 or 2 options per set per AI.

 

But it was fairly easy and staright foward, just a pain to muddle thru all the windows and find that one line you messed up. You could end up searching 20 windows for something stupid like a mis-spell.

[Crispy]

I haven't checked so I can't promise that this is true, but I think D3's NPC functionality has similar capabilities, except you place "path nodes" (entities, on the map) for each action and link them together. I agree it would be nice to have a solution for this which doesn't require scripting, so if D3's system doesn't give you enough control then it would be nice to add such functionality ourselves. But no promises.

[demagogue]

That node system sounds decent.

 

For the record, the cool thing about dromed convo's is that it isn't limited to just one AI. The inputs could be like: #1 AI-1 speaks, #2 - AI1 gestures, #3 AI2 speaks, #4 all the torches in the room burst into flame (maybe this was a blueroom AI frobbing buttons...), #5 soundeffect for the torches-event, #6 AI3 walks in, #7 AI3 speaks, #8 Garrett blacks out, #9 Garrett teleports and "wakes up" in a new area ... etc.

 

It's a pain like Badcog says in the respect of keeping track of any individual event in the queue, but it's pretty cool in that you can knit together a wide variety of events from AI's talking and moving to sfx to changing AI or object properties, to interface changes (e.g., freezing Garrett), etc, etc, all in sequence, (I think you can input time delays between them, and trigger events that have their own timers to go off, so there's good control of the pacing), so pretty easily you could cook up a complex dramatic scene full of disparate events.

 

Anyway, if you do find time to add a system like this, think about this kind of approach, where a wide variety of events are allowed into the queue (also, not just AI events, but e.g. sfx and property-changes directly, so no need for middle-man blueroom AI). Some ideas to innovate might be allowing simultaneous events, maybe the possibility for branching paths so that one AI1 can do his thing in his own sequence while another AI2 does his own thing in his own sequence (without having to ping-pong in one master list), maybe even if-then/else branching paths could be very cool (Deus Ex like, so like if some game event happened before, the convo goes one way, another way if it hasn't) ... definitely a more elgant/straightforward interface that makes it easy to scroll down the queue in one window (not window hopping) and make the input intuitive, what you type in makes sense with what happens, etc. Anyway, I'm just yammering off a wish-list of ideas. But having this sort of thing a little controllable and automated would be very cool and would encourage dramatic, more narrative-based FMs ... a very good thing.

Edited January 5, 2008 by demagogue

[OrbWeaver]

"But the philosophy of Blender is to have one hand on the mouse and one hand on the keyboard!" say the purists. Tough. If you want to make it easy to use, you need mouse-only control for the newbies. Encourage them to graduate to keyboard control, yes, but don't shove it down their throats before they're ready. That approach is a blueprint for failure.

Oh, so that's where it was hiding.

 

I agree; I think they are trying to add more menu-based options (originally there were many functions which were only available by keyboard shortcuts). There is also supposedly a project going on to rationalise and clean up many of the more cluttered buttons panels, but it never seems to make it into any of the point releases as yet.

 

Took me a moment to figure out that you have to be in Edit mode to do that (obvious only in retrospect).

 

Yeah, I assumed you wanted to snap vertices together. You can snap whole objects by selecting them both and hitting CTRL-C, then "Location" to copy the origin, but this is probably less useful.

 

To be fair, the snap options were added fairly recently so for a long time your complaint about not being able to snap vertices would have stood.

[STiFU]

Is it normal, that the normal D3 surface sounds don't work without a further setup either? I have a material like this and the sound is never altered when I step onto my surfaces:

 

textures/darkmod/wood/panels/tiling_1d/trim_small_precious_001

{

wood

qer_editorimage textures/darkmod/wood/panels/trim_small_precious_001_tiling1d_ed.tga

diffusemap textures/darkmod/wood/panels/trim_small_precious_001_tiling1d.tga

bumpmap textures/darkmod/wood/panels/trim_small_precious_001_tiling1d_local.t

ga

}

[Baddcog]

What do you mean exactly STiFU? If your material is wood it should sound like wood when walked on or hit with an arrow, an arrow should always stick too.

 

One exception is on a model that you use a generic collision model shader on, that will override the wood. iInstead you'd need to use tdm_collision_wood as a shader on the collision model.

 

@demagogue,

Another option with Doom is cutscenes, if you played the game you'll know what I mean. They were used extensively. While I prefer normal conversations for gameplay reasons (I can move around while they happen...don't feel like I'm being transported away) they can still be used to great effect and you can control camera movement and everything.

Camera movement might need to be exported from 3d program.

[STiFU]

In the surface parameter list on the wiki it's said that those are the default doom 3 surface parameters, but they don't work for me either. The sound doesn't change when I walk on it...