The role of the level-designer

Well the Splinter cell team handled it differently and IMO they did a good job. It is also the way it is usually done in movies with lots of specialized people so it's not like its a totally new thing.

Chaos Theory looks really nice, they did an excellent job with their normal and specular mapping, which you'll see quite a bit of in the game with the scary dim lighting etc.

I just hope it's a bit less Goddamn linear!

Quote from FrieChamp

Due to the steady increase of detail in levels, also the role of the level-designer is changing and handled differently in each company. The level-designers at Crytek for example focus on general game-design and the design of the missions, but are not involved in the creation of objects, textures etc. It's a cooperation between the design- and art department. ......

So you say there is hope for mappers what have sucky texture and model skills?

Quote from Bluestrike

So you say there is hope for mappers what have sucky texture and model skills?

Yup. Different jobs at game companies are getting more and more specialized. It's getting to the point where game environments are getting too visually detailed for most traditional level designers to handle. If done correctly, the workflow combination of a separate artist and level designer can actually be faster and produce a better, more polished, product in the end. This also leads to the artistic theme of the entire game being more consistant.

For instance, look at a game like UT2k4. A very large portion of the maps were made by a large amount of different people. As a result, you get a game whose maps are very inconsistant. It works for UT2k4 because it's just a crazy game with no clear visual theme anyway. One minute you're fighting on an asteroid with two towers and the next minute you're in and Asian setting. But the game tends to feel less like one whole game and more like an assembly of mods.

HL2 was so polished and looked good because they did separate the level design from the artist tasks. Same with Doom3. Farcry. These current gen games, IMO, show how the roles of artist and level designer are breaking up and how well it works out if everyone is talented and specialized. Just think of next-gen engines and how high-poly they'll get. It's almost too much work even for the level designer that can do relatively high-poly modeling and 2048 texturing.

I think in the next year or two, level design and artist will fully separate across the whole industry because of the boost in has-to-be-modeled-detail-because-brush-can't-do-it. Brush, as a visually appealling tool for the LD, is dying. Brush is going to get to the point where it's only really useful as a method for optimization. But you say, "What about HL2 and Doom3? There's still a lot of brush there!" and I'd say that you are correct and it works well, but the next couple of gen leaps in engine tech and hardware tech are going to make HL2 and Doom3 look like Quake1.

And don't worry, it's not like the level designer doesn't have anything to do anymore. He's now plenty occupied with gameplay/mapflow creation, blocking out the entire game, scripting sequences, and most important of all, making the game fun. Fact is, most LD's are mediocre at modelling and texturing. When a company is trying to create the best possible game, who do you think will get the task of modelling the statue that goes in the courtyard in the map? Artist, because he'll do a much better job at it.

Gearbox does it this way and as a result, the game is better, visually and consistantly. And I still get to do a big portion of tradional LD things. I didn't make the tree, but I get to put it where I want to, and it looks better than anything I could have created myself.

My 2 cents.

I tend to agree with tom

Wouldn't for BiA it be: "OK, got a lovely tree here from the artists. Now, I just need to consult thousands of maps, accounts and photographs to calculate the EXACT position of the thing during D-Day, and if I get it even a millimetre wrong, then I'm sacked"?

Actually, id is a bad example - their team is VERY cross-talented and a lot of them did all kinds of work across multiple areas of the game.

I'm very pleased with the workflow that the Doom3 tech allows. There's no comparison to patch meshes in any other editor, and with those you might as well be modeling. I would love to see a company either a) take the fine control of details from a modeling program and toss it into a level editor (such as with Radiant and patch meshes, particularly texturing), or b) create more effective plugins or add-ons for a 3D package to allow quicker and easier texturing for

I mean, imagine building a set of pipes or a railing. right now, I can plop down a brush, turn it into a cylinder, and rotate and bend it every which way to my hearts content, all with perfect texturing the whole way (or easily tweaked and fixed within seconds if I throw it out of whack). Now imagine an editor where you could not only do that but, by grouping, weld points together and define them as their own model and instance, then plop it around like a model itself.

The thing I love the most about my job is I really do get to do it all - I've documented levels, blocked them out, and am soon getting into a lot of enemy placement and gameplay stuff. Only thing I don't or won't do is the advanced scripting, which is beyond my tiny little programming knowledge anyway. But I also build mostly from scratch (well, when I'm not on someone else's map, hehe). It's really cool as a designer to be able to keep the ideas to a smaller group of people - more hands is good for productivity, but it's sometimes disheartening to lose the control over some of the things you really aimed for.

I definitely agree that designers are getting more and more involved with the really important stuff, like what tom mentioned of the gameplay/flow, blocked out stuff, scripting, and fun in general. But I don't think that means that you have to take all the fun construction work away, at least not just yet.

EDIT: Tequila, you rule.

Quote from Tequila

Wouldn't for BiA it be: "OK, got a lovely tree here from the artists. Now, I just need to consult thousands of maps, accounts and photographs to calculate the EXACT position of the thing during D-Day, and if I get it even a millimetre wrong, then I'm sacked"?

That's why we have Uber-GPS's that have a rewind time feature so we can see into the past.

I don't think even the doom 3 style of level design will make it past the next-gen or next next-gen. BSP will always be around, you need giant flat planes for walls/halls almost everywhere and it's great for optimization/culling. It's great for core architecture, gameplay flow control.

However all the fine details, columns, computers, wires, lights, power plugs, shelves with books, pens, phones, foliage, etc. Those need to be modeled, and regardless of how cool patch meshes are, they simply can't go there. If you could even get that much detail in, getting the UVW map right would be insane, and let us not forget ANIMATIONS. A pipe or railing or thick wiring is ok, and even terrain, but I think going anything beyond that is asking a way too much. Programs like max/maya/etc. have focused on this for years, and artists train on those programs for years. To build an editor with that power would take lots of dev time and training time when it’s really not needed.

An all-control-level-designer-program might work in doom3 but comparatively to HL2/farcry/etc., doom 3 is very low detail and simply would not work in those games. Take an average room from doom 3, take away the normals, fog and steam effects, and make it full bright. Now compare it to the same thing in hl2 and that's what I mean. I guess comparing the look of the levels in the two editors would be perfect to explain. Doom 3 relies so much on these things, and although it’s a nice look, I don't think it will make it past these next-gen's. I just don't see an all-in-one editor even getting close to the level of detail next-gen games are hitting.

Try to imagine the next doom game on Earth (if there) with the current engine. A mixture of outdoor and indoor areas, office buildings, homes, malls, military bases... Lots of artist time is going to be demanded or it really just won't match the level of detail that generation of games will be at. Unless the majority of the game takes place inside a giant metal military complex and the like, some people like that...

BSP will definalty disappear, I dont think its going to survive the next few years. Which is great on one hand, and on the other hand not so great

Great to quickly fix or add a small thing with a cube or so and to zone things off and easily texture them but on the other hand bsp isnt really the fastest way of working and it really is oldskool, next unreal engine wont have BSP anymore anyway afaik and engines that DO have bsp today the day are more of a rarity already than the other way round

You should start to view level editors more as the composers and finishers and not the actual tool to do everything with. We never made everything in an editor in the first place. Things like textures were photoshop. It would be crazy to try and equivalent Max with your game editor. Hard to get such complex program out in just a year or so and not easy for your artist and anyone new you recruit because they have to learn yet another program before they can start to work which slows down.

Editors should be highly specialized in composing levels, easily placing instanced meshes around. Easily making shaders and materials, lighting, particles, special gameplay entities, adding stuff like occluders, simple exporting and saving, testplaying. And all previewable ofcourse

That kinda stuff you need, its the hole in the market because Max cant do it well

And that will prolly be the 3rd job in the future (I hope), Gameplay Dude, Model Dude, The Finish Off Dude

Tom fact that UT2004 is a mess is not our fault, look at ONP, everyone made their own stuff yet (except the themes) its still coherent..

UT2's fault was that were too many people with different skilllevels, too many different mesh artists and specially lack of art direction from the leads are the cause

Quote from DD

doom 3 is very low detail and simply would not work in those games. Take an average room from doom 3, take away the normals, fog and steam effects, and make it full bright. Now compare it to the same thing in hl2 and that's what I mean.

I find it interesting you're saying that from one game on the engine, and it's not like HL2 doesn't have plenty of extremely simple design in it, either. Sometimes less is more, you know. Not every room needs 8 million polys of in-game or high poly source material to look right. And just wait until the crop of licensees starts pushing things for higher minimum spec hardware.

Quote

However all the fine details, columns, computers, wires, lights, power plugs, shelves with books, pens, phones, foliage, etc. Those need to be modeled, and regardless of how cool patch meshes are, they simply can't go there. If you could even get that much detail in, getting the UVW map right would be insane, and let us not forget ANIMATIONS. A pipe or railing or thick wiring is ok, and even terrain, but I think going anything beyond that is asking a way too much. Programs like max/maya/etc. have focused on this for years, and artists train on those programs for years. To build an editor with that power would take lots of dev time and training time when it’s really not needed.

Well, most of those objects I want models for anyway. I'm not denying the strength of modeled props by any means, but some of those examples I would rather do myself. Wires and pipes can be so variable from room to room that I'd much prefer the control of manual construction.

Really, Doom3 is already very close to what I described. I can take geometry that I've built in the editor, export it straight out to LW, tweak it, and save it right back in as a model in maybe a minute at most. If it needs animations, I give it to an animator. I'm not talking about building a new editor, I'm just talking about merging functionality.

I'm not trying to say that things won't change, nor argue that they haven't yet, because they have, and very rapidly at that. Teams are growing to larger than ever before - Raven used to have 2 designers and maybe 10 people per project, now there are at least 10 designers alone, it seems. And to me, that's almost more reason to hold on to designers who both construct and design. Why should there be more and more work on the already greatly extended role of the artist when I can do some of it just as effectively or efficiently myself, and often quicker? Need a torn up wall panel? I can build that in the editor. Railings? Build it in the editor. Pipes? Build it in the editor. Wires? Build it in the editor. Even a twisted mangled beam, I can build in the editor. For me, that's the bat of an eye, and for an artist, it's more time to spend on the assets that really need it.

Oi, this is probably horribly thought out. It's too early to be discussing something online... I wanna go back to bed...

A very interesting topic for sure. I've been actively interested in level design for the best part of six years now. I've built up a solid technical ability over my HL mapping days but have always wondered how working in the industry compares to an (my) amateur workflow. I knew there were many facets to level design at a professional level (gameplay, scripting, texturing, brush-based geometry and model-based geometry etc.) but didn't realise each company had its own approach to the process or quite how much they differed. Thanks for the read - it was very informative. Hi btw (first time poster).

Welcome to HELLL! er.. or... Mapcore as we call it around here.

I like KF squirrels integrated idea, I've no problem with MAX, my issue is when I have to make a model to slot somewhere in the level, it's a bastard to try ang get the position and scaling of it correct and Hammers DXF export is so shitty that it's rarely on any use. Just a button that overlayed things in MAX so I could get the scale and positioning bang on the first time would be good.

Is that when MAX is using cms or inches?

I've actually got a file with cubes as various sizes to use a reference, but it can still be quite tricky to get things exactly as i want them. Guess it's just a matter of practice though.

Quote from TomWithTheWeather

That's why we have Uber-GPS's that have a rewind time feature so we can see into the past.

Ah, naturally!

Quote from FrieChamp

There aren't "any" BSP brushes in FarCry. Floors, walls, ceilings are even modeled objects (in different sizes) you place in the map and align them to get a room. It really is annoying when you spent most of your time in hammer before you start mapping with Sandbox. Maybe I''ll manage to convince one of the senior programmers here to add BSP for the next game (step backwards? )

Yeah I noticed that when I tried to make a level some months ago... good thing the majority of farcry is outdoors. It also explains why my performance indoors in farcry was noticable worst. At least on my machine and I have a pretty kick ass rig.

As far as I know Farcry uses OSP (oct-space-tree, very goo for outdoors/open indoors) or a form of + awesome lod + fog/distance cull + and I think more options for indoors (manual zoning). That's why it can easily go from outdoor to indoor and still cram lots of detail, no?

Sandbox is an awesome editor, I wish I could bread lots of features from it, hammer, unrealed, and doom3 editor to form an uber editor.

Quote from KungFuSquirrel

Really, Doom3 is already very close to what I described. I can take geometry that I've built in the editor, export it straight out to LW, tweak it, and save it right back in as a model in maybe a minute at most. If it needs animations, I give it to an animator. I'm not talking about building a new editor, I'm just talking about merging functionality.

You can do this in hammer, unreal and many other games. I was doing this with hl1 mapping in the past, but after repeated experience I learned I was just producing lower quality/less optimized models by doing so. I still did/would export bsp from the editor just to get scale correct, but I learned to never use the bsp as a base for a model.

In hammer, say I wanted to created a water tower then export to max to model it. I would need several brushes to create the different segments or else I would create invalid geometry. The same is true in doom 3 ed (lets say we aren't using any patch meshes). After importing into max you'll find all these needless faces on the backsides of brushes inside parts of the water tower, on boths sides of intersecting planes! Hope that makes sense, anyways that's a real pain to clean up.

I think the max payne 2 editor has a feature semi-like what you want. You can select a group of brushes, do some conversion and automatically remove those faces, do some beveling/extruding and other simple stuff, and it will convert those brushes into a model. Then you can instance it all over the place. Even at the point most of the stuff I saw created looked like it needed artist modeling time.

Quote from Hourences

BSP will definalty disappear, I dont think its going to survive the next few years.

BSP or some form of it will always be useful. Especially in heavily indoor games. Outdoor games may use plain distance culls + or manual zoning (VERY TEDIOUS) and get away with it, but that only works in outdoor games. In heavily indoor games you want to cram as much detail as possible in a confined space. Some rooms may be long or just plain big and others small, lots of rooms need to zig zag close to each other, etc. Trying to occomplish a well performing game in that type of enviorment, and do it fast is just way harder and less efficient using modeling programs and no BSP like culling.

The workflow, speed, efficency of dev teams using automatically generated nodes/zones for culling using BSP will totally blow away any competitor who thinks they can do the same using max and maybe + manual zoning or distance cull only. They just will not be able to match the detail and performance.

Here's a great example: Compare an indoor area of Halo PC, performance and detail, to a game that was released on PC when HALO 1 XBOX was which uses BSP/OSP/the like.

The term "BSP" many mappers use refers to in-editor built geometry. To me it means the method used for culling. BSP will not go away, nor will the use of it applied to editor built geometry.

Also last I heard on unreal engine 3 was that it was switching back to additive "BSP". This to me implies that they realized it's importance and spent time improving it. And IMO the game epic plans to release using UE3 is NEXT-NEXT Gen.