1. When building a DM map, what is your general methodology... Do you construct the outline of the map and get the interconnections established first, or do you completely finish each area (including lighting, entity placement, etc..) before moving on to the next?

Levelord
...usually I get brain farts that are immediately fleshed in the editor I rarely have a pre-drawn outline. Often, though, I'll have sketches of sub-sections of a level.

Sverre Kvernmo
I'll frequently decide on a general layout before I begin, either on paper or in my head. This is definetely easier with DM levels since they tend to be smaller and less complex than single player levels. I've never made a level that ended up the way I originally planned it though. The level will, and should, take on a life of its own as you find out which ideas work and which don't.

Giggler
A quality deathmatch map is all about layout and game balance. Before starting the actual map creation, its a good idea to decide upon a general layout, and sketch it out on paper. Although I don't finalize lighting and entity placement as I create the layout, I keep in mind where the major weapons and powerups will be. This helps gives the balance and flow needed for a good deathmatch map.

Tim Willits
I first start out thinking about what I want to accomplish in the level, where it goes in the game, what the player needs to do before finishing it and what puzzles they will face. I then build as much of the level without worrying about lighting and some special detail. I build the entire level before I fine tune any lighting.
 
Gyro Gearloose
I'll probably always be experimenting with construction techniques, even if I ever manage to stumble on one that actually makes this stuff any easier for me! I try to start with an overall concept, like "a single floor fragfest with a central Quad and lots of lava" or "a level that plays like DM3" or something like that before I start editing. I often make some drawings on paper and sometimes little flowcharts to visualize what rooms will hook up to what. I do use the editor as a sketchpad for things once I get into it, so it's not like I know how things will work before I touch the mouse. Each level seems to require its own experiments and compromises before it's done. I have no problem deleting heaps o' brushes if I decide something's not working the way I want it to.

For the most part, I try to detail rooms and passages as I go. Plumber was an exception to that - I roughed out the entire thing first without any lights or detailing, and then went back over it to add those things - and the detailing phase just seemed to drag on and on and on until I couldn't stand to look at it anymore. It might work for some, but I don't think I'll use that approach again. Sadly, Plumber also turned out to barely work as a DM map, but I think it was mostly due to its cramped passages rather than its topology or construction method. I hadn't figured out r_speeds yet either.

After a level's basically done, I use lots of walkthroughs and playtesting to try to get the amounts of health, ammo, and armor placements sorted out. I've got some buddies at work to help test with on a LAN, so I get a chance to see how the map works in a 3-4 person environment. It would be really cool to have some sort of beta service on the net though, where a group of folks would sign up to be willing to be ginuea pigs on a server that could host maps in the fine-tuning phase. For that matter, it would be great to have more servers on the net running user-made DM maps!
 
Paradox
Usually most DM maps require a non-linear layout. So, creating the entire layout of the map first, leaving details for later, is usually the best way to progress.

DaBug
I would first have a theme or general idea of what the level is all about. I'm a big proponent of having a level look and feel like its a somewhat identifiable place. I also try to have a good idea of what I want to accomplish in the level, which includes things like number of players its designed for as well as what weapons and powerups are included. I love to have a finished level have places that players make up names for as well as being well balanced(i.e. weapons,armor,& powerups).

2. I have noticed in several maps that a good way to improve the looks of a level is to place brushes with different textures on the same plane as other brushes. Seems to add depth to the area with no real cost. For example, on the wall beside a "func_plat" you might place a brush with the "working gears" texture. The brush with the gear texture does not stick out/in from the wall, but it is on the same plane, as if painted on. Is this done for visual effect, or is it a concious decision not have it stick in or out to reduce face counts, and therefore improve framerate? (whew, gotta take a breath!)
 
Levelord
...well, if a breush like you describe does stick out, it will add to the surfaces being displayed and therefore slow things down a bit. Even the added texture will cause the framerate to drop as BSP has to now divide the surface into smaller sub-surfaces. I'd say, bottomline, that I strive to make things look as "real" as possible and I'm constantly battling with the dreaded Dragon of Framerate while doing so.

Sverre Kvernmo
There is still a cost even if the brush with the decorative texture isn't protruding from the wall. The brush thats level with the wall it's imbedded in costs roughly half in number of faces. As for the reason for using one or the other decoration, thats mostly a taste issue. Some textures look good level with their surroundings since they are part of a themed set of textures. Others might not align up perfectly and have to be carved into the wall to look good.
 
Giggler
Having that texture not stick in or out of the wall does save 6+ polys. Its all a matter of how much framerate you have left to "spend" in any particular room. If that particular room was only running with an r_speeds of 200-300, then the extra detail would be worth it.
 
Tim Willits
It really depends on the level designer and what they want to accomplish in the area they are building. Its always better to not have textures flush with each other but to give them a little depth. BUT, if you are having trouble getting an area to run smoother then flushing textures together is acceptable. Avoid it on the floor.
 
Gyro Gearloose
I'm a clueless guy on that one! :) That's a technique that I haven't used a whole lot. You inspired me to do some testing, however, and about all I was able to figure out is that it looks like there is a penalty associated with the use of coplanar brushes that I don't understand.

As a test, I made a 256 unit cube, dropped a light 300 in the center and a DM start to one side looking toward the center of the room. I then sliced all the walls into 32 unit high horizontal strips, textured them with alternating textures, and copied the result to two different maps. In the second map, I nudged the interior faces toward the inside of the room 16 units. I ran qbsp, vis -level 4, and light -extra on the maps. Curiously, all three polygon counts on the r_speed test were actually worse for the first map than for the second! The first number of the n/n/n triplet, which I believe is the total number of polys being considered for drawing, was actually double that of the map that had the hidden surfaces in it. I rechecked things and became a gibbering idiot.

Fortunately, when I sliced the walls up further into checkerboards (64 32x32 squares per wall) and did the same test, I got more sensible results. The first map (the coplanar faces version) then definitely ran faster and with considerably fewer drawn polygons than the raised-red-squares checkerboard wall version.

It makes sense that raising or indenting a surface if you don't have to is bad, because it creates as many new polygons as there are edges on the surface being extruded (plus additional bsp nodes and possibly cuts and so on) but I'm scritchin' my haid about what I saw with coplanar faces versus going ahead and indenting/extruding in an area that's going to have low polygon counts in either case. For areas with higher polygon counts, the coplanar faces technique makes good sense.

Paradox
Usually, details like that are done for many of the above reasons. Changing a face versus adding a extruded brush can help keep world splits, and visible polygon's down a great deal. Keeping a level playable, and pleasing to the eye are a delicate balance. More often than not, things that look really cool are nuked due to the framerate-monster. Low framerate is espically important on strictly DM maps, giving consideration to internet players, already suffering from lag, and the possibility of many players firing rockets in the same room. (When a rocket is fired, the surface cache is rebuilt for the dynamic light. That is very slow on most machines, therefore, many players firing rockets in a very complex room slows the game down considerably.)

DaBug
I would say this is done for visuals, and for the most part your always fighting face counts. I try to get things working like I want and then work on making it look better and the whole time watching my face count.

3. When you run into the situation (and we all have) where you have designed an area that looks totally bitchin, but runs kinda slow, what steps do you generally take to improve it again? Do you compromise lighting effects first, or do you restructure walls? Or is the best idea to just throw the damn thing away and start over?
 
Levelord
...this is where MOST of my time is spent. The first thing I do is try to simplify the area. Often the same "effect" can be acheived with less polygons and surfaces. I rarely limit lighting except perhaps dynamic lighting.
 
Sverre Kvernmo
It's hard to give a good general answer to this one since all those cool looking areas are totally different, with different problems bogging them down. The first thing to do is to use "r_draworder" to make sure you can't see into any areas that aren't important to the coolness of the room. Next, consider what makes the room so groovy and try to rescue that aspect of the area, at the cost of all the others if neccessary. i.e. if you've come up with some fantastic new light effect then the sacrifices should be made to the architecture and vice versa. The best cure for avoiding this dilemma is to test you areas before they become too complex, keeping constant count of the face-count.
 
Giggler
I've never had to throw a room away, although sometimes you just have to bite the bullet and put an awkward wall in somewhere, or tighten a room up to save some framerate. The first thing to go is always dynamic lights, they are a framerate killer. The next step is to alter the layout so that less rooms, or parts of rooms are being drawn from the problem areas. Lastly, if any one section is still too high in r_speeds, some detail has to go.
 
Tim Willits
Its not always best to throw things away, I rarely throw an area away. Its best to rework the area until you can create the look you want and still have it run fast. This helps you learn and the more experience you gain doing this the faster you can build areas that look good and run fast.
 
Gyro Gearloose
The problems of slowdown in Deathmatch levels have seemed to me to be both overrated and underestimated at the same time. Lessee if I can get out of that one!

The slowdown problems that a lot of people seem to think of with Deathmatch levels are the same problems associated with levels in general, or even with 3D graphics in general. Limitations on polygon counts and scene complexity forced by the graphics engine and hardwareis going to be a stumbling block for all of us probably for many years to come. My first attempt (Crucible, which was primarily intended to be a DM level) with the Quake engine was a classic example of the Wrong Thing - great big rooms with lots of occluding objects. It was a perfect candidate for just chalking it up to experience and starting over on another level.

Later, I did levels where I went into them with the intent to keep tabs on polygon counts and framerates as I added passages and rooms. That approach allowed me to start trying to figure out what worked and what didn't, and why some really complex-looking levels run faster than some that look simpler. GYDM (not on any FTP sites, but you can snag it off my page) is my best effort at a decent-performing DM level using the technique of adding some stuff, checking it out, changing things as needed, adding more, testing, etc. This worked well because it even told me where the low polygon-count areas were, which meant that I could afford to add detail and shadow-casting stuff to those areas to not only make them look better, but actually bring their counts up a little so there was less variation in framerates from one area of the level to the next. It wound up looking pretty dense to me, and yet there aren't many places a player can be where the polys take a big jump, so I think the approach worked well.

As to dynamic lighting effects, I must confess ignorance. I understand that they can have an adverse effect on performance, and I find it interesting to note that some of the id levels use spawnflags 2048 on lavaballs to remove them for deathmatch, but I haven't actually done much r_speeds testing there, so I'm a clueless guy when it comes to deciding whether to cut down on polys or lighting effects to get framerates up. I would suspect that they're related though. In other words, I'll bet that lighting effects probably won't hurt much if restricted to low-polygon areas. In other words, if a flickering light or a lavaball is only illuminating 20 nearby polys, it probably won't affect framerates nearly as much as if it's casting light across 200 polys.

What I meant by slowdown problems being overrated is that the primary purpose of a DM level is to be fun to play. I don't know how many hours I've spent happily gibbing and being gibbed in just the START map, but it always amazes me to bring up START, type "r_speeds 1" at the console and see such terrible numbers! :) My point here is that there will be cases where the numbers may not be the best, but if it seems to work alright and people have fun playing it, maybe that's okay!

What I meant by slowdown problems being underestimated is that Deathmatch performance issues usually go beyond the polygon counts and 3D data structure problems. Deathmatch performance tends to be communication bound rather than graphics bound. Lag, latency, and bandwidth problems seem to turn out to be much smellier ogres than polygon counts for network deathmatching. All I can suggest here is to avoid the use of big rooms for people to meet in, even if they have low polygon counts and high single-player framerates. E1M7's lower floors stay under 200 polys, yet modemers disconnect like crazy rather than stand around frozen hoping for someone to exit when that level comes up.
 
Paradox
Lighting effects are always the first to go. Next would be a restructure, of the area itself, or surrounding areas that might be visible. Its very rare that cool stuff is just totally thrown away.

DaBug
I don't generally use any special lighting effects on DM maps. Lighting surfaces is just too costly IMO. Big rooms can kill ala E1M7, can't have a whole bunch of projectiles flying around the level. but if its designed correctly ala HipDM1 then the big room feeling can be accomplished and work really well.

4. What is your general feeling on powerups/rocket launchers in DM maps? Should they always be behind secrets or in tough to get to places ... or do you think they add to the experience, and should be as easily attainable as other weapons (but in smaller numbers obviously!).
 
Levelord
...personally, I hate power-ups in DM. On the other hand, I love rocket launchers. Placement of any of these goodies depends on the specific area that they're in. Usually you want to use goodies to draw players to certain areas, but the goodies themselves can be out in the open, or hidden away. You also want to make absolutely sure that the placement of a goody does not allow any one player to dominate simply because they've captured the goody's area and are holding the area with the goody's benefit.
 
Sverre Kvernmo
I'm not a great supporter of keeping the important items secret. Once one player has found where the goods are, he is at a tremendous advantage, further unbalancing the severely unbalanced game Quake is in deathmatch. While the other shotgun-toting players are searching franticly for the hidden stuff, "Mr Secret" can pound everyone into the ground. Great fun for the dominator, frustrating for everyone else. Of course, this won't happen once all the players know the map, but then what is the point of keeping them secret? I personally always put at least one rocket launcher in a fairly easy-to-get area, and there are always two of them in the map. This helps avoid the S& M style of play described above. Remember that for the good players Quake really has only three items Quad, rocket-launcher and red armour (and _maybe_ the lightning gun). Place them carefully.

Giggler
No secrets in DM maps!! Powerups can be fun, but there has to be a risk in trying to get them. DM4 has great placement of the Quad Power, because you have to jump over the lava to get it. On the other hand the Quad in DM3 is just there to get, and whoever is around when it respawns gets it. This type of placement favors the people who camp near it. Regular weapons need to be placed differently though. Rocket launchers cannot be put in hard to get places, such as the one in DM5. Whoever has one, can easily stop others from getting at it, since that requires hitting the switch and running over the catwalk. I love a map where everyone gets a weapon on respawn, and the game isn't revolving around some guy with the quad.
 
Tim Willits
I like rocket launchers and power ups to be in difficult spots to get to, it helps create valuable areas in a level and getting the rocketluancher becomes another goal as well as killing other people.
 
Gyro Gearloose
Both! :) One of my favorite levels is DM2. I enjoyed it since the release of QTEST, the Deathmatch Test version, but I really started to appreciate it after I watched the must-see Romero/KillCreek Deathmatch recordings (www.ionstorm.com). It's a level that's perfect for two-player DM, yet it's got two rocket launchers, two red armors, two megahealths, and a Quad, and everything's right out in the open (given the railing jump to the armor/health pair and the rocket jump to get the Quad). I'd a thunk it would've been overkill, yet it's aged to perfection - straightforward yet tense, with copious amounts of guts flying everywhere.

In contrast, I've also had great fun in E1M1 trying to dig a skilled camper out of the grenade room while they bounced grenades upstairs getting frags on people in the exit room and waiting for the pentagram to respawn rough, yet a cerebral environment in which to spatter brains thither and yon.

I think both approaches can work well in different levels, but of course they aren't guaranteed to work well in any random configuration. It's hard to beat grabbing some deathmatch buddies who can give you some good feedback and getting them to help out with some "research". It was interesting to see the small change (but major improvement) that was made to DM2 between QTEST and the 1.01 release. I'll bet the change was a result of many hours of deathmatching in that level.

Paradox
In DM maps, balance is the most important thing. Putting the rocket launcher in a good position, not too close to other insane powerups or superhealths. Putting the supernailgun around some megahealth to give a player stuck with the lesser gun a balanced chance with a rocket launcher armed player. Putting the red, yellow armor in unique places. etc..etc... Balance is the most important thing in a DM map, peroid. Keep powerups away from the higher power weapons. Put armor, health help close to respawn spots where a lesser weapon is. Consider where and how each respawn spot will be used, and taken advantage of. Try not to create camping spots, or places on a map where somebody can easily dominate for the entire game. Keep it as fair as possbile for all those playing.

DaBug
Depends on what you're trying to accomplish. Sometimes having a rocket launcher out in the open works as in DM2 but it always helps to reward skilled players with somewhat harder to get weapons and health. If you're gonna have just one then it better be in a wide open place. Usually a little play testing decides these things.

5. What do you feel are the significant differences in designing for internet play (via QW) as opposed to LAN... I've heard keeping the brush count lower is a must, but I've also heard rumors that the rocket launcher can cause lag due to all the smoke trails and stuff it creates...

Levelord
...hmmmm, I haven't really accomodated this difference in my DM levels. I always keep the framerate as high as possible in either situation.

Sverre Kvernmo
When designing for internet play, keep in mind that 99% of the players will be lagged, no matter how low your brush-count is. They'll have a lot less coordination than you have when testing the level. Don't make a QW-map into a dexterity test with narrow ledges, pixel-perfect jumps, narrow doorways and tough rocket-jumps. If you do, you're designing for the LPBs (myself included ), and your map won't get much play. As far as I know, the nailguns are the worst speed-hoggers for internet games.
 
Giggler
I am assuming that rocket launcher smoke, etc. is all client side, meaning that playing over the Inet would not affect it. It is true that in bigger areas, more player positions could possibly be calculated at once. So with that in mind, the r_speeds for Inet maps should be lower than those made for the LAN. Aside from that, the biggest difference here is that most every Inet map should be made for 16-32 players, and LAN maps are often for a smaller group of players.
 
Tim Willits
The best thing you can do to speed up internet play is not use laser shooters in a level. Besides that I wouldn't worry to much about making it fun over a LAN and over the Internet, just make the level fun.

Gyro Gearloose
LAN setups are usually not near as prone to either the bandwidth or latency problems that players encounter over the internet. In a LAN, you typically have 10 Mbit ethernet adapters and hopefully some decent routers and gateways, and ping times are in the sub-40 ms range. Over the net, not only is latency a big factor, but bandwidth can be limited. Some clients will be on 28.8 modems, where there are not enough free bits per second in the datastream to handle all the packets needed to cover 16 people in E1M7. Not gonna happen. While there are too many variables in the whole internet performance equation to be able to figure out a set of rules that will always give optimal performance, I think it's fairly safe to say that things that cause network packets to be sent can have a greater detrimental effect on performance than things that don't. In other words, the brush counts are, as always, important because letting them get out of hand will lower the framerate on the clients, but even rock-bottom brush counts will not help the situation when there are 12 people in one room all blasting away at each other.

The rocket smoke trail question is an interesting one that I don't know the answer to. My guess would be that use of the rocket launcher does indeed generate more network traffic than, say, a shotgun blast because the server has to tell the clients where a rocket is and when and where it hits something, while a shotgun blast takes place "instantly". I would also guess that nailguns are expensive from a bandwidth point of view. Even so, I'd advocate trying to design the level so that not everyone winds up in the same room very often rather than cut down on rockets, grenade launchers, and nailguns! :)

Now you've got me wondering about whether the server tells a client when a smoke puff or a bubble has vanished or whether the client takes care of it on its own...

Paradox
Really the difference isn't that great. DM maps created for LAN or QW dont have to differ that greatly. Just keep in mind how much stuff is going on in the level at once. For maps that are geared torwards Internet play, keep the amount of moving objects (func_trains, spikeshooters) low because the game has to update the information for each of those objects every frame. Making maps with lots of stuff going on slower over the internet due to the increased packet length. LAN dosen't suffer as much from this problem due to the near-local speed of a good ethernet.

DaBug
DM2's moving lava floors suck when you're lagged. Big rooms like E1M7 suck when you're lagged. I pretty much design with the same things in mind for both LAN and internet play.

6. How important is the visual aspect of a DM level to you? I've played levels that looked great, but sucked as a map ... and others that looked pretty bland, but were just excellent for DM'ing! I personally, like a level to look good and play good, but that's just me. Any comments?

Levelord
...if I have to pick, I always opt for playability. That's the reason for being there. Making the level look cool comes later and it's often limited for performance. In fact, most of my aesthetic work is done for the sake of the singleplayer.

Sverre Kvernmo
I guess I weigh them both equally. While the playability issue is what keeps people playing your map, first impressions are all-important. You may have designed the greatest level ever but noone's going to know if it looks like shit. This is even true down to the level that you should make sure the player start-points look at some cool view.

Giggler
Heh, all of my DM maps I made before joining the industry were showcase maps, with high detail and very little layout and planning. So I guess it depends on what your reasons for making it are. < g> In general, layout and entity placement come first.

Tim Willits
It takes talent to make a level look good and play well. Your first goal of level design should be to make every area look great. Any idiot can slap some walls up and accidentally make a level play well but to make it look well takes talent.

Gyro Gearloose
I definitely agree with you. I tend to get bored with a level that doesn't have some visual appeal no matter how well it works for deathmatching.
 
Paradox
Gotta do both. Keep the level looking good, and playable for all. Crapply looking levels leave a horrid first impression on players, despite if the level has the best layout on Earth. Although, in time people (if they give it a chance) realize that the map is fun due to layout. Then, if a level runs slow, or the layout is mediocre people usually play it for a while, then become fed up with the unplayable speed of the man, despite good looks or otherwise.

DaBug
Has to look good and play good.

7. Which style of level do you enjoy making the most and why? I like medieval/metal castles, but I've seen some kick-ass military DM levels as well. Which styles are the toughest to pull off and why?

Levelord
...I like to capture as much realism as possible, so anything contemporary or from the past is best with me. I'm uncomfrotable with futuristic settings because I have a hard time "seeing" them. My absolute favorite is medieval, although I had a blast doing the Duke Nukem Los Angeles levels.

Sverre Kvernmo
For me personally its easier by far to create ancient/medieval levels. Both the architecture style and the texture set lend themeselves to cool abstract shapes. With futuristic maps I feel more constrained and the challenge is greater to create cool, believable scenes. I enjoy both styles though. The closer you go in time to the present, levels become harder to create and allow for less expression by the designer. Working on Redneck Rampage was a chore at times. )

Giggler
It all depends on what textures you have to work with, and what the creator has an imagination for.

Tim Willits
I like castles and medieval stuff the best. That takes passion to make. The ancient architects where very romantic people.

Gyro Gearloose
To tell you the truth, I'm still working on that. When I was a kid, I spent hours and hours drawing mazes for some reason. I remember the twisted glee I'd feel when I finally tracked down an adult and got them to wind their way through my scribbles while I watched, usually laughing and trying to distract them. Weird kid. I really do appreciate my parents for not institutionalizing me. Anyway, I get the same kick out of the whole process of making levels, even when they don't turn into anything worth uploading. I like the look of metal and lava and shadows being cast a lot, so I've been trying to play with that lately.

The military/base levels are the hardest for me. I've tried to do some but I just don't have the knack for it. Actually, I suck so bad at them that I PGP encrypted the evidence and hurled the disks off a bridge for fear of blackmail.

Paradox
I like doing a variety of maps. Originally, Tech levels were my favorite but I now really love doing everything.

DaBug
I love them all. None are really tougher then others. I could see some problems with trying to make realistic outdoor areas but other then that its pretty much getting the face count to work and balancing the items.

8. Any comments or suggestions you would make to the current crop of level designers out there based on what you've seen recently? No need to name names, but is there anything that's caught your eye in either a good or bad way lately?

Levelord
...actually, I haven't seen much. Most of the good designers have been snatched up by game companies and their work is now clandestined. I haven't really seen anything on the net that excited me lately, but if I did, the designer may get snatched up, too (wink, wink, knudge, knudge).

Sverre Kvernmo
Sadly I haven't been able to download and play other people's maps as much as I'd like to lately. I think the last really cool freeware stuff I saw was a couple of maps by Martin Hryinewicki (last name is probably spelt wrong - sorry Martin:). Hipnotic's add-on pack was great, with some much needed fresh approaches to Quake level design.

Giggler
Chicken Feet.

Tim Willits
My only advice to people is not rush to get your level on the internet, take some time with what you do and make it great. Put some love into and search your soul for something better than what's out there.

Gyro Gearloose
Whenever I get the chance to go download some levels and check them out, I invariably find some really impressive work and things that are a lot of fun to play. About all I can say here is many thanks to the people that create cool stuff for everyone to enjoy, including levels, patches, Quake pages, development and technical info, news, and everything else associated with this most excellent hobby. I haven't used a weekend to change the oil in my truck for almost a year now.

Paradox
A lot of maps have caught my eye. The overall quality of levels on the net is increasing, as more editing groups form, and more TC's come closer to completion.

DaBug
I think the hypnotic(ritual) add-on rocked in level design. Designers should look at those maps and carry the architecture ideas over into DM map making. It's ok to have things look cool in DM levels :) I love variety and I definitely love style.