Steps to producing an in-game model

[FictiousWill]

Ok, this is going to be an on-going dev diary of sorts, where I post my progress on one small part of the development process. This time I'll be taking a sketch by our illustrious concept artist o_nobody_o, and (hopefully) converting it into a 3d in-game skinned/textured/animated/interactive model. I'll post in this topic as I go, explaining little quirks and eccentricities of the process and hopefully what all is involved.

If you don't want to see all the in-progress posts, you can click here (http://www.sourcefortsmod.com/boards/showpost.php?p=87908&postcount=92) to see the post of the final product.

A note about this particular item. SF is going to be having new game modes in upcoming versions. Some of these modes will be created by the developers, but the new entity set should allow for map makers to pretty well create whatever they want. At least one of the new official modes is going to need a 'goal' of sorts (as in the soccer kind) that needs to fulfill the following requirements:

1. It's got to have some kind of arpeture. So you can throw a ball in it or whatever. Some way to use it to score.
2. It's got to be big enough for that to happen. This is going to be a decent-sized prop. Going to need some significant texel detail. (A texel is a pixel displayed in 3d. I'll get into that more later)
3. It's got to look cool. Goal posts are boring. It's got to look all techy and badass. Think 'power reactor'. Cables and shit coming off. You know, the whole deal.

Here's the concept that o_nobody_o produced. I think it's perfect. It's got the big hole to throw stuff into to score, it's big, should scale well, and it's got detailing around the outside to make it look cool. It even has big power cables. Ideal.
http://img442.imageshack.us/img442/2199/concept1nc9.jpg

A few things I like the look of off the bat: The cables. They'll be polygon intensive, but they should look really nice, and what's more I can use a repeating texture on them and save texture space while they still look good. Plus KGtheway2B is awesome at making normal maps for stuff like cabling. Very excited to see the finished product there.

The little fans. Brilliant idea of o_nobody_o's - I can animate those and make them spin, maybe have them emit steam or that heatwave effect. Visually interesting. Also can't wait to see those in action.

The sides of the reactor - I can save texture space by repeating geometry. Perfect. Since a lot of the model repeats, I can re-use a lot of uvw space and up the model material resolution without sacrificing memory space. I'll get into this and explain more later when I uvw the model.

Anyway, I've started meshing it out in 3d studio. Updates to come.

[Nocashvalue]

Sexy.
Woo for content :D

[FictiousWill]

Currently doodling in 3ds max. This is going to be one of the eight segments to ring around the central bucket part of the model. The way I see it at the moment, the model will be constructed of four component parts, one of these segments (duplicated eight times), the power cables (probably a mirror of a single cable), the bucket itself, and any detailing to go in-between.

http://img183.imageshack.us/img183/9715/mesh1qf4.jpg

[RedXIII^]

Looks like something good has come from the leak at least.

Nice concept.

[Mudkip]

Woot, new game modes and entities!

That looks really nice. Model the big cables individually though, it would look strange if they all had the same shape, I think.

[FictiousWill]

The cables will start as splines that I will sweep out into cylinders along their paths. I'll probably make individual cables for each side, then mirror them to the other side.

In the meantime, here's the central bucket, and a chamferbox to add some detail.
http://img457.imageshack.us/img457/9718/mesh2cr9.jpg

[FictiousWill]

Different styles of modeling come in handy for different modeling tasks. Often I can start with a simple shape and extrude or bevel faces to arrive at the final shape I'm shooting for. This is typically called 'box modeling', and it's useful for static objects like tables, chair, guns, and usually anything that's got discernable edges.

The power cables are going to need a slightly different approach, just because of how they drop down from the side of the model. Even though these cables won't move, they need to appear 'draped' and as if they could possibly become dynamic, even if they're not.

It could be possible to box model the cables, starting with a long segmented cylinder and pulling the vertices about, but it would be pretty tricky to get them all to line up correctly. For this task I need to spline model them.

In spline modeling, you lay out a line of vertices in a spline, or an array of points. What's useful about the spline is that it can be moved or modified at each segment with just one vertex, instead of 18 or however many sides the final cylinder will have.

Here are the power cables in spline form. Ready to be swept into cylinders. When I do that, 3ds will extrapolate cylinders around each of the vertices relative to the rotation and orientation of their neighbors, and produce a smooth cylinder along the whole length.
http://img408.imageshack.us/img408/8648/mesh3if6.jpg

[Nocashvalue]

This is rather interesting ;O

[FictiousWill]

...like this:
http://img117.imageshack.us/img117/1447/mesh4ah2.jpg

I also extruded some faces towards the end of the now-cylindrical cables to form the segmented ends of the cable connectors. At this point I'm almost ready to uvw unwrap the model, the first step in the skinning process.

[FictiousWill]

When I uvw map (the 'uvw' comes from the coordinates used to specify how the polygons map onto the flat texture material - xyz is in 3d space, uvw are the translated coordinates in 2d space) I usually collapse the mesh into a single part so that I can see all the faces on the uvw map at the same time. This is helpful because I can make sure I use all the space on the texture, and thereby maximize texel resolution and only map a single polygon to a uvw position once. This is why I only have one segment of the side of the model done, instead of eight. I'll only duplicate it when it's fully uvmapped - that way I only have to skin it once instead of eight times.

Nevertheless, it'd be fun to see what the model will eventually look like, even if it's not skinned yet. So I duplicated the side parts eight times anyway, mirrored the cables, and made sure all the parts lined up ok.

http://img297.imageshack.us/img297/8984/mesh5uk6.jpg

[King Simmy]

...like this:


I also extruded some faces towards the end of the now-cylindrical cables to form the segmented ends of the cable connectors. At this point I'm almost ready to uvw unwrap the model, the first step in the skinning process.

looking good :)

[Nocashvalue]

Looks awesome - can't wait until it's textured :)

[SGorilla]

Wow, you are a great modeller, this is very interesting! :D

[FictiousWill]

Thanks :D

Of all the steps involved in producing a model, uvw mapping is probably the least immediately rewarding step. The task is to take every face of the 3d model and flatten it on a 2d plane. It's rather tedious, and when it's done, the model looks exactly the same as it did when you started. But it needs to be done, the model will never make it into the game without uvw mapping, so I need to bite the bullet and get it done.

Polygons all have uvw mapping coordinates applied to them already, but they're generated by 3ds max, and only based on a single axis, so they're useless as far as my uvw mapping needs are concerned. Here's a look at what the uvwmap looks like to start with:

http://img117.imageshack.us/img117/8575/uv1zw0.jpg

That jumble of green lines on the left is the model flattened on the uvw map without any work done on it yet. You can make out one of the power cables I think, and some other parts of the mesh there. Most of it overlaps completely and is unusable, so I'll start unwrapping the faces manually.

[FictiousWill]

First I separate all the parts so I can see what I'm doing. See that blue square in the center of the big white checkered square? That's the space I have for the final texture - All the faces have to fit in there. It's not a question of space, there's tons of room because the faces can be scaled however much is necessary - but it is a bit of a puzzle to get them all to fit optimally, that is, so they use as much space as they can. The more space a polygon gets on the uvw map, the higher its in-game resolution is and the better it will look.

That thing that looks like a radiation symbol is the face of one of the fans, and the rest of the parts are pretty much recognizable now.

http://img66.imageshack.us/img66/7135/uv2if4.jpg

[kas]

This is sweet...hurry and post more!

[SwiTcH]

:O Nice Job FicWil! Keep up the good work

[SGorilla]

This shows how hard and time consuming it is to model and texture, good work! :D

[TommieV]

All looking really sexy, just wondering though how big this goal thing will be in-game. Probably, bigger then a player?

And, I also want to thank you for posting this. Making models is time consuming, but I bet reporting back everytime you add something isnt going to speed up the proces at all. However, Id like to see this section used more in this way :)

[Mudkip]

Yes, but what do the cables connect to???

[FictiousWill]

That's a good question Mudkip.

.... :P

Um. Damn. Well I'll stick boxes or something on the ends I suppose. I hadn't thought of that. I had assumed they'd disappear into map walls or level geometry, but we don't want to restrict the mapper to have to do that. I'll add ending bits to the cables, or at least include them as a submodel.

[Mudkip]

Maybe you can make a companion model which is an extension to the cables. Then they can still disappear into walls, but the walls don't need to be so close to the goal. Then put the ending bits on that.

[4tune]

So will we use the gravity guns to shoot the object into the goal or something?

This is pretty interesting I never knew what the modeling process was and now I'm wonder why people actually go through with it...

[NocturnO]

This is very nice will :O

[TommieV]

4tune, the same reason as why people paint or make music.

[FictiousWill]

uvw unwrapping is now done. And the resulting model looks like this:
http://img117.imageshack.us/img117/1447/mesh4ah2.jpg

Yeah, it looks the same as before. See what I mean about not very gratifying? The uvw map however, is nice and clean:

http://img441.imageshack.us/img441/6342/uv3sr5.jpg

If you've ever opened up in-game media assets before you're probably used to seeing model textures in their unwrapped form. This template forms the guide that literally wraps that texture information onto the 3d model in realtime. With the uvw map mostly done, I can duplicate the eight segments again and then burn lightmaps. I'll get into what's involved in that for the next post.

[teamstevo]

gj will!!!!

[FictiousWill]

The next step is to burn lightmaps for the texture. While it's possible to import the model at this point, it wouldn't look very nice in-game because it would be lit by vertex, meaning that the only way the game engine can apply shadows or highlights would be to shade corners of triangles light or dark. The engine does do this, but it is vastly supplemented by a slight lightmap burn, which I do now.

In this step I essentially instruct 3ds max to sandblast the model from every possible angle with simulated bouncing raytraces, much like how a mapper runs vrad.exe on their map to generate lightmaps. This means that nooks and crannies in the model will be darkened and highlights untouched, giving the model an overall more solid look.

Burning lightmaps takes a little time, probably about 10 to 15 mins, depending on the size and complexity of the model, and the size of the resulting skin that I'm applying. This model will use a 1024^2 bitmap, so I'm rendering out the lightmap at that scale. Here's the render at 3 mins in:

http://img171.imageshack.us/img171/3798/lightmap1wr0.jpg

[Billeh]

Fic, you should have it so that cables bend into the floor, and if need be add connectors around the cables to make like sockets on the floor

[FictiousWill]

And in the time it took me to type that post, it finished the rest of the lightmap. Normally when I'm burning lightmaps I go on irc and sing lightmap burning songs, but typing explanatory posts is far more interesting.

This image is scaled down to 512 from the original 1024 map.
http://img401.imageshack.us/img401/6021/lightmap2ti6.jpg

It will form the underlay of the model's eventual texture material.

I'll see about turning the cables to attach to the floor. That I think should work out well without breaking the uvw map.

[FictiousWill]

One thing to notice is how repetitive the uvw map is on the bottom. That's where the unwrap of the main bucket is, and a lot of it is wasted space, either obscured behind geometry or simply repeats eight times. So I altered the uvw map to only map that area once for those eight segments, and hopefully gained a little uvw space.

I'll probably use the remaining gaps in those cylinders for the mapping to go on the feet of the cabling, which I'll probably do now.

In the meantime, here's an image of the model with the lightmap applied. This isn't a clay render, this is realtime - rotating around in 3d. The lightmap effect gives a model a slight rendered look even in realtime - this is the effect we were after.

http://img441.imageshack.us/img441/8844/lightmap3ew8.jpg

here's a better look, an animated gif showing lightmapping before and after application:
http://img405.imageshack.us/img405/2850/lightmapfv5.gif

[4tune]

Fic thats looking good. I don't know why but without the lightmap it reminds me of concept models used in movies. But if this is the net will be seeing the opening in front of us like a soccer net or will the opening be above us like a basketball net?
4tune, the same reason as why people paint or make music.Nah. Making stuff with music is pretty simple in comparison in my opinion. If you write a song with a random key and chords every day you get about on average 10 good ones per year... Except lyrics I have no lazy ass way around those...

[Khuskan]

Nah. Making stuff with music is pretty simple in comparison in my opinion. If you write a song with a random key and chords every day you get about on average 10 good ones per year... Except lyrics I have no lazy ass way around those...

...Take that back :-(

Sure you can stick a bunch of standard chords together and some cheesy words, but would it be -music-?

[FictiousWill]

Mappers will be able to place the model (if they choose to use the model) at any angle or orientation they desire. I'm sure there will be some interesting maps with goals placed vertically or upside-down and all variations in-between.

I've started on the underlay skin, which means a coating of noise and dirt to weather the model, and I've added an alpha channel to knock out the texture where the vent to the fans are. This way I can have a nice-looking vent with fans rotating beneath and not have to make polygon-eating vent meshes. In all liklihood, the fan itself could be an alphamapped single polygon, but then I'd get alpha-fighting and that could cause problems. It doesn't hurt to leave the fans as polys.

http://img461.imageshack.us/img461/1858/skin1jp6.jpg

[4tune]

Fic I need to eat your brain for your talent.
What you said about the net's directions could be pretty interesting. As long as it looks good in all angles we could probably see a scenario where we have to bounce the object off a wall to get it into a net that faces in the opposite direction.
...Take that back :-(

Sure you can stick a bunch of standard chords together and some cheesy words, but would it be -music-?Sadly yes. Don't forget alot of music is composed of three basic chords (Not all music. But alot of it especially things like pop...not Britney Spears pop) Hell Greenday uses only started using a fourth chord with American Idiot...Then again they suck.

[FictiousWill]

Here's the mesh and skin in their current state. At this point I'm going to take a break and get some food, do some work and run some real-life errands. The model is nowhere near done, I'd say this is the half-way point. When I get back I'll finish up the skin, animate the little fans and do a test export to the game. Then I'll need to make some aggressive LOD models so it will run smoothly and share screen real estate with other in-game assets. And at some point I need to plug those damn cables into the ground.

http://img101.imageshack.us/img101/2734/skin2sw3.jpg

[dacrusader]

dude this is crazy and awesome sweet content!

[4tune]

Fic I think you killed imageshack. The image doesn't show up. I'll just assume it has something cool like deathrays on it.

[Mudkip]

Works fine for me, and looking good too!

So... 2.0 is halfway done now, right? You'll release it tomorrow, right? :p

[Flare]

Here's the mesh and skin in their current state. At this point I'm going to take a break and get some food, do some work and run some real-life errands. The model is nowhere near done, I'd say this is the half-way point. When I get back I'll finish up the skin, animate the little fans and do a test export to the game. Then I'll need to make some aggressive LOD models so it will run smoothly and share screen real estate with other in-game assets. And at some point I need to plug those damn cables into the ground.

http://img101.imageshack.us/img101/2734/skin2sw3.jpg

Thats the sexiest thing you've made yet. You've outdone yourself! Good job!

[4tune]

Now I can see it. I'm a bit disappointed that it didn't have deathrays though.

Fixed? (http://i3.photobucket.com/albums/y82/HungryHungryHippo/Fixed.png)

[Billeh]

Their invisible death rays. Brilliant work ficwill ^^

[KGtheway2B]

Just wait till I get my grimey little hands on this :D

I guess I can add onto this page too once fic has finished- showing midsteps and whatnot.

[Mudkip]

Do it! Do it! Do it!

/chanting

[FictiousWill]

I spent a little more time on the texture. What's nice about the texture map is that it can be edited and altered at just about any step from here on out. So I'll probably leave the skin in its current state for the importing process and KG and I can edit it further once it's in-game.

http://img366.imageshack.us/img366/4080/skin3hg6.jpg

Next step is to animate and produce a collision model. I'll probably have to get to that sometime tomorrow afternoon.

[Tony]

Has anyone else noticed that despite all the bullshit people give FicWill, he's actually..

Really fucking talented?

[kas]

Yes. And I love getting media releases this way!

[MasterDana]

I agree, step-by-step for the win.

[4tune]

It looks like it is shaping up quite well even though there still aren't any deathrays.
So assuming since it's sourceforts I can't wait to see how building can affect this. (Unless this game mode wouldn't have building)
Has anyone else noticed that despite all the bullshit people give FicWill, he's actually..

Really fucking talented?Don't look at me. I liked those player models he made. (Then again I like Tribes so maybe I just like them for the nostalgia those kind of models give me)

[eeeentRo]

Has anyone else noticed that despite all the bullshit people give FicWill, he's actually..

Really fucking talented?
This model I actually like, though I admittedly don't really know what the hell it is. Whatever it is, it looks good. Still not really a fan of some of the other stuff, but this is pretty detailed and cool.

[Klip1337]

fuck I wish I had your talents. I can make the model just fine, but doing textures and stuff... i suck at.

[FictiousWill]

Before I crash for the night I produced two LOD models.
If you forgot what LOD models are, I explained how they work in detail here:
http://www.sourcefortsmod.com/boards/showpost.php?p=75481&postcount=139

I also made little feet for the ends of the cables and stuck those on. The final model set will be the main bucket and a cable segment. The mapper will then be able to decide whether or not to use anywhere from 0 to 8 of the cable segments to hook into the bucket, depending on what their map geometry allows.

http://img379.imageshack.us/img379/1116/lod1ir9.jpg

[inzehhat]

coo beans.

[NocturnO]

<3 !!! (ten).

[Yurg]

FicWill is indeed a God among men!

No, really. And think how many more times he has to crash.

[Sayyan]

Really nice looking model there Fic (as has been said a lot) I really like the style of it, but just a little thought: Will there be seperate textured ones for different style of maps? If you think about the variations between sf_skyway, sf_castles and sf_tread, one texture style isn't going to fit nicely in every map combination. So is there any thought about different texture styles? And how easy is it for map makers to re-textures models? (i.e will the bug where after playing a map with different textures from the standard, all other maps will be re-textured, be fixed?)

Sorry if that made no sense whatsoever, but again I love the model, and I love this style of content release, it puts the whole development process into some perspective for people.

[Trubblez]

I'll have to praiz u on that peice of artwork there.

[GCool]

'Tis indeed brilliant. What were you saying about animations though?
I meen, it looks awesome, but it might be worth putting the cables in a different model (I think you said that) and then have different animations on them of them kind of... breaking off the ground and flying round a bit. Could be good for a gamemode.

[FictiousWill]

Thanks guys -

The next step for the model is to produce a collision model. The source engine uses a nearly universal (as far as engines go) technique for model collision, separate from bsp brush collision. Each model is defined by two coordinates that specify an overall bounding box, and within that box the lower resolution collision box is defined. This is to allow fast collision and raytrace detection. When a traceline is fired in the engine (and these happen just about constantly, either to determine the path of a bullet, or just to determine the direction the player is facing) the engine first pares down the number of visible portals the player is looking into.

You all who like to enclose your maps in boxes and call it a day just because it looks the same in the end, this is just another reason why map optimization matters.

With a list of the current and visible portals and bsp areas in mind, the engine can draw a traceline out and determine if it intersects with any model bounding boxes - not collision boxes yet, just the overall bounding space. This way if it's obvious that the traceline doesn't come close to a model, it's not necessary to calculate if a collision occurred.

Once again, if your map isn't optimized correctly, it's going to have to check every portal and every bsp area, and that's often many many times more processor load than is necessary. More processor load = slower calculation and a slower game.

If the traceline does indeed intersect with a model bounding box, then it's time for the engine to check each facing side of a model's collision mesh. A collision mesh element must be convex (collision meshes are made up of multiple convex parts to simulate concavity) to perform accurate detection. In the case of the bucket part here, I need to arrange 16 separate convex shapes to roughly cover the model. This is the model that will determine if bullets hit or where players can walk or bump into the model.

The collision model is never rendered, so potential z-buffer sorting isn't an issue, and the convex hulls may intersect as is necessary to form the concave overall shape. I've covered most of the model with collision hulls at this point, I still need to cover the cables and then perform some optimization steps - I'll get to those in the next post.

http://img401.imageshack.us/img401/3970/collision1bg9.jpg

[Flare]

TEH SEX! Thats beautiful, man. Lovely work.

[mc_nebula]

VERRRY NICE!!!, HOW MUUUCH??

yea, this rocks...

This is the way to go for the development. You are more likely to keep us hooked whgile you develop if you do this sort of thing. But a word of warning, dont do it too much, we want some surprises...right? I can tell that what has been said throughout this thread has affected development. That is a good thing IMO.

Thanks and nice work Fic!

[FictiousWill]

Here's the finished collision mesh in wireframe. The cylinder segments that surround the cables should provide passable collision detection, while the hulls surrounding the bucket are generally precise to the shape of the underlying mesh.

http://img408.imageshack.us/img408/5383/collision2ki4.jpg

[nip]

its so hawt...


how bout some red and blue lights inside?

[FictiousWill]

Now that the collision hulls are in place they need to have all their faces set to the same smoothing group and material id, or they won't be exported correctly into .smd format and will be unreconizable by studiomdl, the hl2 .mdl compiler.

Each hull has every polygon set to smoothing group 1 here, and then I apply the reactor_sheet material id to all the polygons on all the collision hulls. The planar uvw mapping is irrelevant, the polygons have default uvw coordinates, but they will be discarded. The important thing is that they now share the material id with the underlying visible mesh.

Lights inside as well as steam effects and light glows are all added by the mapper. We don't want to imbed them in the model if they can be added using dynamic map entities that we may or may not want visible later on.

http://img511.imageshack.us/img511/167/collision3kp0.jpg

[FictiousWill]

Now I need to figure out how best to scale the model so it appears correctly in hammer (and therefore at the correct scale ingame).

I open hammer and drag out a few default solids to approximate the size and shape of the model in relation to the player. I can then export the brush solids into .dxf format and import them into 3d studio to use as guidelines.

http://img530.imageshack.us/img530/5161/scale1ld6.jpg

[FictiousWill]

The .dxf file is imported into 3ds, and the model scaled up to meet the size of the brush guideline. Now we can get an idea of the model's eventual in-game size.

http://img471.imageshack.us/img471/3329/scale2yh5.jpg

[DJimpy]

that was e-mazing fic. you want my first born child?

[FictiousWill]

Oh we're not done yet. Now I have to export to .smd format and compile down to an .mdl file. I'm going to start with the bucket and then do the cabling. So from here on out the project splits into compilation of two distinct models.

The visible mesh is exported as a reference mesh, the mesh that all animations will assume to have present. The collision mesh is exported as its own reference smd file, and the idle animation (the fans spinning) is exported as a five-frame animation .smd.

Then I create a batch file to call studiomdl with a .qc script that I have written that defines all the parts of the model - what .smd files define which parts and how the compiler should treat those parts to form the eventual final model.

If all the paths are correct, it will compile successfully to .mdl and be visible in the model viewer. I haven't placed the materials in the mod folder yet, so it will appear with the path error purple checker material, but that's ok as we were only testing to see if the geometry exported correctly at this point.

http://img122.imageshack.us/img122/2894/export1kq2.jpg

It appears the collision model failed to export correctly. This is either because I didn't collapse all translations on the collision hulls, or because valve decided not to make an official 3ds max exporter, unlike every other game engine toolset on the planet. Now we start the trial-and-error process of collapsing collision hull rotation transformations to get the collision model to appear correctly in the model viewer. Because it can't just be a matter of exporting it from 3ds and then importing. Oh no.

I doubt there will be an official valve plugin for 3ds max any time in the future, so we've got to work with what we've got. It's just an unnecessary shame that it's so easy for other people. (http://udn.epicgames.com/Two/ActorX)

[AlphaSE7EN]

Who knew you could convince so many adolescents to suck your dick with a few screenies?


good job, you rule <3

[Mudkip]

You all who like to enclose your maps in boxes and call it a day just because it looks the same in the end

I threw up a little in my mouth when I read that. God, I hope nobody is actually that sloppy!

Model's still lookin' fat and sassy, Fic; in fact, twice as fat as I expected. I'm thinking I'll probably build a raised catwalk near this thing so that it's easy to just drop the bomb/flag/platypus into there. Is that what you have in mind?

[FictiousWill]

Yes that's how I assume the model will be placed, either with surrounding ramps or placed in a pit of some kind - with the overall shape of the bucket being that of a pretty even sized cylinder there should be no end of placement possibilities for mappers.

Collision model and geometry export #2, better but not there yet. If I get this in three attempts it will be a record.
http://img161.imageshack.us/img161/7694/export2gq2.jpg

[nip]

when will this be able to work ingame? o.O

[o_nobody_o]

Fic, I must say, you've got some extreme talent. And to think, you're making my life easier! That concept 'SKETCH' was just to give you an 'IDEA' of what I was thinking. I didn't even get a chance to finalize anything, and here you are running with it.... AWESOME!

Fantastic job!

[4tune]

I'm going to steal the whole Borat thing that was mentioned earlier in a life scaring way.
http://ffmedia.ign.com/filmforce/image/article/711/711738/borat-20060607053153666-000.jpg

Awesome job on that though.
Fic, I must say, you've got some extreme talent. And to think, you're making my life easier! That concept 'SKETCH' was just to give you an 'IDEA' of what I was thinking. I didn't even get a chance to finalize anything, and here you are running with it.... AWESOME!

Fantastic job!I wish everyone I've ever worked with did that.

[nhammen]

I love your work on this... except for the textures on the pipes... it looks like im seeing artifacting on them. Just letting you know, in case that is what I was seeing.

[Stieffers]

I love your work on this... except for the textures on the pipes... it looks like im seeing artifacting on them. Just letting you know, in case that is what I was seeing.

That's just the 3dsMax viewport renderer auto-scaling the texture for performance reasons.

[Sauce]

LOVERLY work FicWill. I always said you were a good modeller, and you still are! :D Thanks to o_nobody_o as well; there would be no model without a concept!

I must say Will that this style of releasing works very well. It worked well for sf_airships, and it's working hella well for this.

[FictiousWill]

Since the 3d studio exporter refused to correctly reset rotation translations for each of the collision model hulls, I was forced to rotate them manually in the .smd reference file. So the original exported rotation placed the object at 90 degrees out of alignment, with a rotation of exactly pi, so I rotated the collision model back by exactly pi/2 on every node line, and re-exported and compiled. Now the collision model matches and lines up.

I then exported the material sheet from photoshop to vtf .dxt5 format (to preserve alpha map transparency) and wrote a corresponding .vmt material file for it. With these files placed in the mod directory, the materials should appear in the model viewer.

http://img250.imageshack.us/img250/3238/collision4fr1.jpg

[FictiousWill]

I can now build a testmap in hammer and place the model as a static prop. The alpha maps on the material are disabled because they cause z-fighting with the other polygons, so I'll need to recompile with the $mostlyopaque .qc command, or probably convert the fan grates to polygons after all. The source engine doesn't perform alpha sorting well, and we've got to work around it.

Here's the model ingame with no alpha or normal maps:
http://img293.imageshack.us/img293/5340/ingame1si7.jpg

...and the collision mesh is functioning correctly:
http://img113.imageshack.us/img113/7079/ingame2mu5.jpg

We're not done yet though. Now I need to export each of the LOD models into smd reference files and add them in the .qc script. Then we can start on the cables.

[Upseeri]

I'll sacrifice 100 cactus fantasticos to you

[KGtheway2B]

So the original exported rotation placed the object at 90 degrees out of alignment, with a rotation of exactly pi, so I rotated the collision model back by exactly pi/2 on every node line, and re-exported and compiled. Now the collision model matches and lines up.

Haha, funny- exactly pi :rolleyes:

[SwiTcH]

I'll sacrifice 100 cactus fantasticos to you

You'd probobly die too then. Anyways:

This model looks like its going along VERY great, and its good that everyone like its. But yea its wierd how alot of people questioned his models before and questioned his skills but now they worship you :O.

[RedXIII^]

Well the leaked weapon models/player models are so so.

But this is nice.

[Tony]

Hey, Red?

Stfu.

:)

[Flare]

This is indeed a lovely model. This is a great thread for someone wanting to put a model ingame, it's very informative!

[Fail]

Well the leaked weapon models/player models are so so.

But this is nice.

I agree.

[Pincus]

Hey, Red?

Stfu.

:)
Wow... To say the least.

[RedXIII^]

Hey, Red?

Stfu.

:)

Nah thats ok, I have an opinion, so i'll voice it.

[Tony]

So when did people start taking me seriously, for starters?

But realistically, we all understand that people think the models that have been seen, released or not legitimately, are from good to bad and everywhere in between. What's the point in asserting it again? Yes, of course you're free to say it, but if it's unnecessary, then it's unnecessary, duh.

[RedXIII^]

I felt the need to make my appreciation known for this model, sorry.

[dacrusader]

So when did people start taking me seriously, for starters?


When you started sounding serious...

[FictiousWill]

And at this point we're finally just about done. The model is ingame as a set of three separate submodels, all swapping lods at the right thresholds and lit with env_lightglow and light entities. The cables rotate around the origin of the bucket so they can be easily placed to attach to any of the eight radial parts, and the fan grates have been exported as separate models that discard their alpha map at 64 units for lod. The animation is functional and the little fans spin in their sockets.

All-in-all, I'm left with about 30 3ds source files, a number of .smd references and animations, a few .qc scripts and the compiled .mdl and material files that have been placed in their correct locations in the mod directory.

Development team members can place the model in their maps now, any further updates that I apply won't change the size, shape or appearance enough to change the way it is used or placed in hammer.

So to wrap things up, here's the original concept:
http://img442.imageshack.us/img442/2199/concept1nc9.jpg

And the in-engine result:
http://img257.imageshack.us/img257/7639/ingame3du6.jpg
http://img75.imageshack.us/img75/3017/ingame5rb9.jpg
http://img332.imageshack.us/img332/2231/ingame4gh6.jpg

Elements I might return and improve are the cable connectors - they're more detailed in the concept art, and look nicer with large cylindrical attachments. I may also remove collision altogether for the cables, as they're a pain to walk over at the moment. Any other additions will probably be raised during testing.

That's the process, more or less. Hope you found it interesting and/or gave you a glimpse of the development process, at least as far as art assets are concerned.

[Nocashvalue]

Looks gorgeous.
:D

[Billeh]

You know the lights around the edge, are they coloured to the model or did you put them into the map?
If there in the model, are there possiblities of red or blue, also you should make some stay on and some flicker, might looked good ^^

Looks awesome never the less

[Bovineconspiracy]

That's amaza-zing! Just read through the whole thread (well, Will's posts anyway), and it looks fantastic. 3d modeling and scripts and coding and all that stuff amazes me and it's something I could never do myself. Great job!

[nip]

Hawt

[RedXIII^]

If the devs talked more openly like this then SF would be a better place.

Good work there Will :)

[SwiTcH]

VERY nice job :O

[hobbes]

hmmm...those edges look pretty sharp, someone could hurt themselves.

Safety > Gameplay

[IactoSophos]

Nice models! Just going to read back through and find out what they are for... :)

EDIT: Ah, one of those 'new gametype' thingies eh? Bravo!

[MasterDana]

I'd hit that.

[dacrusader]

hmmm...those edges look pretty sharp, someone could hurt themselves.

Safety > Gameplay


Or you could trip over the pipes.....

I'm seriously concerned!

[Sauce]

very nice. now hurry up and start working on the next one! :P

[kas]

That's the process, more or less. Hope you found it interesting and/or gave you a glimpse of the development process, at least as far as art assets are concerned.

I loved it! Make moar media releases like this, it's kick-ass!

[cellarmation]

It looks great. I am also pleased the devs are listening to the fans and releasing a bit more stuff :-P.

Will you be able to walk over the pipes or will you have to jump? What should happen to the ball when it hits the center? will it just disapear into a func_flagzone?

[zerocool]

So, who's signing up for my new Ficwill Unites Cables Klan? -={fuck}=-



[Reserved space to repost his ancient "I love my Willy" picture]

[Vanderdecken]

Wow. Well done FicWill. Nocollision on the pipes = the way to go, or provide ramps over them (in the model I mean).

[GCool]

Or the mapper could make ramps, or the /players in game/ could make ramps. Why do it for them and limit what they're capable of in the model?

[Vanderdecken]

Or the mapper could make ramps, or the /players in game/ could make ramps. Why do it for them and limit what they're capable of in the model?

Because if it was left to the players it would never get done, and someone would invariably think 'What's this ramp doing in the middle of nowhere?' and remove it, not realizing what it was for.

[InfernoX]

That's the process, more or less. Hope you found it interesting and/or gave you a glimpse of the development process, at least as far as art assets are concerned.

Dr. Will, dat there creation looks like a disintergrater...bucket

[stume618]

The ramps should be built by players, not mappers or modelers, in my opinion.

And to topic: EXCELLENT looks so nice ficWil

[Vanderdecken]

The ramps should be built by players, not mappers or modelers, in my opinion.

And to topic: EXCELLENT looks so nice ficWil

Two months late, and worth it.

[Sauce]

its always worth it when you mispell "Will" :P

[Vanderdecken]

its always worth it when you mispell "Will" :P

He's thinking of Wil Wheaton, too much Star Trek on the brain.