[GUIDE] Balance, layout and your A/D map.

  • Site Migration: See bugs? Report them here. Want something changed or have an idea? Suggest it here.


Old Man Mutant Ninja Turtle
Mar 4, 2008
So you want to make an A/D map, preferably featuring 3 stages. But what do you need to consider to make sure Blu don't steamroll or Red from forming an impenetrable bulwark? How about this...

People tend to overcomplicate matters for their A/D maps; indecisive area sizes and inappropriately lengthy corridors, ill thought out secondary paths and a lack of understanding as to how players will respond to their environment. We consider the engine, and we consider the player.

Chapter log:
Considering the basic layout.
Considering Blu spawn.
Defining an alternative route.
Balancing alternative routes.
Progressive difficulty.
Making your map unique whilst following this template.
Final thoughts, the engineer and the sniper.
Additional information.

The long winding road.

The best trend to exercise for a map that will inevitably occupy a length of space, such as A/D, is to wrap it around itself. This not only means you have less of a chance to reach your map boundaries but that you have more opportunities to increase optimisation and vary your maps gameplay aspects with bends and corners. This is also true for 5 CP push maps which regularly take on an S shape.




One small step for Blu.

A reasonable spawn setup from which Blu assaults Red will usually consist of either 3 minor exits, or 1 major and 1 minor exit (or similar mixture there of). This relates directly to player count and is relevant to area transitions; CP1-CP2. Firstly, a team will (at maximum) consist of 16 players. This becomes a small amount when you start to split them up into any more than 3 routes that will inevitably reach a single objective defended by up to 16 enemies at once. So to have any more than 3 seperate paths is a bad idea and breaks player focus. These spawn exits should allow players direct or at least indirect (special jumping) access to significant play area's (defined below).








What are my options as a player?

High ground, low ground, and middle ground. These are fairly self explanatory but can exist in their own varying ways depending on the approach of the level designer. Particular routes will be favoured by particular classes and will be catered for by the level designer to compliment the obvious gameplay tactical choices any player will want to employ.

You can't defeat me now, i have the high ground!

The high ground will be favoured by the sniper where he's hard to attack and far from sight. Whilst the soldier can fire rockets into his enemy before jumping down to finish off the job and inevitably complete the map objective. They are also favoured by players whom prefer to penetrate further into enemy territory with lowered risk of being intercepted and achieving the element of surprise.

A low blow indeed!

The low ground will provide the best flanking position due to greater territory penetration. Less flak will be received from the low ground as visibility is limited which means it is more likely to be overlooked by defenders in comparison to larger and more likely threats; the high and middle ground. This route will be favoured by spy's and pyro's and other surprise player formations (a surprise ubered heavy etc).

In contrast to the standard assault mode, payload mode can contradict this trend due to the single moving objective. The objective which travels forwards along a predefined path defines the main route itself, and may occupy and subsequently redefine the higher or lower grounds as the main route. However, in the name of balance it normally occupies the middle or lower ground. So the low ground frequently becomes the main route. As illustrated below.




The middle ground is usually the main assault force direction and will likely consist of most of the available classes. These classes will pack a heavy HP punch consisting of buffed heavies, soldier's, pyro's and demo's etc. The main route/middle ground will be flanked either side by the higher ground which penetrates a moderate amount into the next territory (or merely a bay to shoot from in relative safety and even jump from), and the lower ground which will also penetrate moderately or even significantly beyond the main routes entrance to the objective.



The main route should be wider to accommodate the greater congregation of players. Too small and explosives become over powered, which unbalances classes as well as teams. The main route should also include moderate cover as players will attack and counter attack frequently. Cover can come in the form of corners or detail props/brushes. The secondary routes can be smaller, fewer players will traverse these paths and less space is required; these routes are also frequently one way, so less cover is required. Also, these routes will generally branch from the previous area and reattach to the main path/objective, to create the secondary routes as large as the main route would only make things cumbersome.

These 3 categories of path need to be balanced with 3 more considerations. Short, long and medium length paths. Due to the spawn time, players will often favour the shortest path to the objective, even if it is the most dangerous (which it probably will be). Long paths can be employed to balance area's of greatest tactical advantage, whilst short paths might be frequently counter-attacked, spammed and ambushed. In general the main route will be shortest, the high ground of medium distance and the low ground usually the longest, possibly flanking the objective completely.

Progressive difficulty.

A key part of multi-stage A/D (and even single stage), is for the maps difficulty for Blu to progressively become more difficult per stage/control point. The current official maps produce this with various levels of success, but there are several options available to you to achieve this element of A/D gameplay. The easiest and simplest way to make things more difficult is to reduce the number of paths of attack to the next control point. This is a method i would leave until the finale, as its gameplay effects are powerful; early use of reduced alternate paths means its not sensible to use at later points without totally damaging overall map balance and can be seen as generally bad design if the map does not play well throughout. More imaginative ways of increasing Blu's difficulty is to not provide them with areas to regroup or gather before an objective; such an area is present on the last bend of Dustbowl and goldrush before the Finale (the long winding path between the last control points), where although large ammo packs allow Blu engineers to get a forward base setup quickly, a single Red uber or mass counter attack can flush out the Blu team. Other more internal layout choices can be applied, such as more powerful defencive positions that are purposesfully unbalanced in Red's favour. Like more numerous ledges with larger ammo and health drops. Additionally it should be noted that Red are given the use of a "spawn shortcut" to the first CP in the third and final stage of A/D maps, which is not apparent with previous stages before the finale. This is in part due to the long winding path seperating the 2 control points in the final stage and allows Red to travel to the round-gates faster. Narrower paths are also an option, as it does make Blu more vulnerable to splash damage and small scale flanking manouvres.

Defining your routes for a unique environment.


Despite the fact that there are only 2 height levels in this area (excluding stand alone structures) we can consider the left path high ground. It flanks the main path and provides a point where players can gain a greater understanding of the battlefield visually. Players may opt to take cover in the low ground from here or leap to the main route.


The low ground. The low ground is difficult to defend effectively without exposing more vulnerable positions more likely of attack. Limited visibility, reduced number of conflicts with attention on the entrance and exit define this low grounds key tactical features. Taking the low ground as a path to your enemy is always a gamble, handicapping your combat advantages or sacrificing time for the element of surprise.


The middle ground, or “no mans land” may exchange hands in attack and counter attack, and should provide cover and possibly height variations that benefit both teams. The final stage usually includes a winding corridor to the finale. Otherwise, the main route should be wider than the others.


So Blu have used your 3 routes effectively and taken the first CP. How does the first area bridge with the second? Trends vary, but generally consist of the main path and 2 alternates; or 2 equally significant paths (one high and one middle ground). Direct visibility from CP1 to CP2 is nil and for optimisation purposes (although it does inadvertently compliment gameplay by balancing classes). For the main route, the path will often be split into two paths and allow the player to approach the next main area from 2 directions. While the higher and lower grounds flank and penetrate even further into enemy territory.




Here i have demonstrated how Valve generally seperate 2 areas visually/physically. The walls not only prevent excessive lines of sight for better gameplay, but blocks vis, increasing map performance. Textures such as chicken wire and glass when applied to solid world geometry will cause that brush to no longer block vis. They should be avoided if your map suffers from bad performance.

Mixing things up: What makes your map unique.


We can define the layout mechanics here quite easily, despite the complicated geometry and interlocking doors/bridges (which should generally be avoided, in order to focus combat). We can look at this areas design as a worst case scenario in terms of sophistication. The right route (from Blu's perspective) provides limited penetration via high ground. Blu already has easy access and control of the high ground on the left, so this area should not be given extra visibility or firing lines against Red, but has also been made counter-attack-able for Red through a ground route from the main path in addition to the blind corner. Red can seek to control this valuable ground. This balances the left high ground and its one way door that prevents direct counter attacks from Red. The green area can be considered the low ground due to it's limited visibility in and out, narrow route and direct path, with focus on the entrance and exit.

At the end of the day it is not a requirement that a particular area requires all of the above 3 routes to be complete. The alternate paths can exist in varying forms as long as they all balance out. 2 routes of high ground and moderate territory penetration can still equal the tactical advantage of one high route with low penetration and one low route of higher penetration. The choice of what alternate routes to use are often influenced by the available space and is not a bad reason to apply a low ground over some high ground even if that is what you intended. As a designer compromises will always have to be made, and it is not worth making things difficult. If it doesn't work, cut it out and think of this as a template to consider other possibilities.

Final thoughts: Considering the static defence.

The immediate thought that comes to mind when the majority of people consider making a custom assault style map is for Blu to be attacking a rediculous fort, with all sorts of amazing, elaborate sentry nests and sniper towers. Directly considering the role of individual classes over others generally leads to class favourable maps and imbalances.

Meet the engineer.

If a sentry spot is too powerful, but you don't want to change your layout, reduce the sizes of local ammo drops. This way it is difficult to set up a sentry gun, but if it is achieved, the engineer is rewarded with many kills. A level designer can influence sentry positions by ammo drop locations and their ammo size, along with the available cover such as corners or waist height props. If a sentry covers a large number of exits it allows the engineer to acquire many points, have it outranged somewhere or directly flanked to reduce its mortallity rate.

Meet the sniper.

We all love to consider the perfect sniper position: An area that covers a key route with a lot of player traffic. Local ammo and health supply. Cover for protection. Height and great distances to keep yourself out of reach. One path to dream location to watch your back all the more easier. If your map has a location like this then i would seriously recommend removing it. A sniper always needs to be vulnerable, as vulnerable to its anti-classes as all other classes are to the sniper behind his scope. If a sniper has great distance for protection reduce his field of vision, if it covers the main route where a sniper can acquire many points, have it flanked by an alternate route or make sure the distance and/or height advantage is severely limited. Because we all know what a sniper wants it's hard not to accomodate for it, but we still need to balance him by allowing him to be killed as much as any other class.

Summary and additional information:
3 basic routes of attack: High, low, and middle ground. As a designer you will negotiate these variables.
A stage: Spawn > arena > objective > dividing wall > arena > objective.
High ground may penetrate territory, but should not provide Blu with area control.
Low ground may penetrate territory and even flank an objective.

Approximate area of play sizes in Hammer units.
Dustbowl area sizes:
Stage1 Area1: 2432 long 1344 wide
Stage1 Area2: 1920 long 1344 wide
Stage2 Area1: 2368 long 1488 wide
Stage2 Area2: 1984 long 1536 wide
Stage3 Area1: 1792 long 1024 wide
Stage3 Area2 consists of 2 winding corridors 1408 long and 1472 long
Stage3 Finale: 1536 long 1664 wide

Goldrush area sizes:
Stage1 Area1: 2432 long 1408 wide
Stage1 Area2: 1408 long 1792 wide
Stage2 Area1: 2688 long 1664 wide
Stage2 Area2: 2176 long 1664 wide
Stage3 Area1: 2304 long 1280 wide
Stage3 Area2 consists of 2 winding corridors 2304 long and 2176 long
Stage3 Finale: 1280 long 1792 wide

Sentry sight distance: 1100 hammer units.

Please note that this is not a comprehensive guide to Attack / Defend maps, the possibilities are quite extensive. I have discussed the introduction of map limitations that focus and control gameplay in order for you to produce a map with working-complimentary gameplay for TF2, for standard 24-32 player servers.
Last edited:


Old Man Mutant Ninja Turtle
Mar 4, 2008
I intended to make this back in May this year, when it would have subsequently made a nice addition to the A/D competition announcement. However i was covering this aspect as part of my dissertation (originally) and figured i'd just wait and publish the dissertation here instead. Unfortunately i have had to change my subject focus half way through and so i have nothing to show for it.

The contest entry's that are currently being evaluated are all of a high quality and almost seemed to make the notion of such an article redundant, however i still see some basic and common mistakes being made. So i went ahead and formed this guide to A/D maps anyway as it still seems that some of the basic features remain overlooked.

Also see:
[ARTICLE] Stairs and your map
[ARTICLE] Immersion and Your Map
[ARTICLE] Roofs, Skybox Brushes and Your Map
[GUIDE] Base Facades and your map map
[GUIDE] Scale and Your Map
Last edited:


L7: Fancy Member
Mar 22, 2009
Great post, however IMO you shouldn't have used dustbowl as the main example because I don't think its such a great map for TF2.


Blu Hatte, Greyscale Backdrop.
Dec 28, 2008
Thanks for this, as I am currently working on an a/d cp map. I never noticed how maps curl around themselves, I'll keep that in mind!


L3: Member
Sep 16, 2009
Good tutorial, now if only someone would make a tutorial on how to create buildings correctly.

You know, make them valve-ie


Jul 14, 2009
I personally avoid making major exits on purpose. The players will find their preferred way anyway, and in the examples you showed, nobody uses the major ways because they're too dangerous.
Jan 31, 2008
I wouldn't use dustbowl as an example, some places are really lacking alternate routes and is very tight in general.

Bermuda Cake

L9: Fashionable Member
Feb 20, 2009
wow, thanks for this.
I always find that analysing the VALVe maps helps me immensely, might put up some of my own thoughts on the maps up some time


Old Man Mutant Ninja Turtle
Mar 4, 2008
I wouldn't use dustbowl as an example, some places are really lacking alternate routes and is very tight in general.

You are very correct. I had another section all written up about improvements to Dustbowl but it wasn't supposed to be an analyse of Dustbowl like you might see a competitive player ranting on his pug forums about. Instead i just left out the most flawed area's of Dustbowl and concentrated on the more relavent area's that demonstrate my points.

Great post, however IMO you shouldn't have used dustbowl as the main example because I don't think its such a great map for TF2.

I didn't really want to use Dustbowl for the very fact that it is a port. I made a point in the article originally, but later removed it, that Dustbowl has a lot of flaws because it was a design for similar but ultimately different game with differing game mechanics. I removed it because dustbowl has had 3 major modifications since TFC that improve its gameplay for TF2. The fact that there are 24/7 Dustbowl servers out there, and before the release of the class achievements this was the most popular map, it seemed justified.

I didn't really want to extend this already lengthy article including the flaws of particular A/D maps when it was about taking the basics of these layout designs and applying it to your own A/D map. Badwater is considered widely to be the best of all A/D maps to date and is considered so because of its wider open spaces and secondary route that almost helixes around the payload route. This made me want to include points on Badwater, but i only ended up reitterating all the basic points made previously. I figured there's a place here for a single stage gameplay mechanics article while this can focus on multi-stage A/D.

If anyone has any additional thoughts feel free to drop them in. This tutorial is for everyone, especially the mapping newbie. I'm thinking of updating a couple additional images but there are loooads already.

Good tutorial, now if only someone would make a tutorial on how to create buildings correctly.

You know, make them valve-ie

I intend to cover a few points involving buildings in future, but to make a single comprehensive article might be a bit axcessive. For now the best advice is red have angled roofs, blu have flat roofs, anything else is worth opening Hammer and loading up an sdk_* map for reference. One of the best pieces of advice i ever heard was "an artist looks, he doesn't paint", the point she was making was that painters and illustrators of reality spend the majority of their time observing their subject, minute amounts of time are spent applying the stroke.
Last edited:


Old Man Mutant Ninja Turtle
Mar 4, 2008
It's good to see people are already benefiting from this articles insight.

I included a chapter log at the start after including additional comments on the sniper and engineer as a static defence (after all). Hopefully people can gather the more specific content within this article without having to read the whole thing this way. As well as an additional paragraph on mixing up alternate routes beyond the high, medium and low ground template. Although i'm finding it hard on what information to include that couldn't better be covered in more detail in another article altogether, despite its relavence.
Last edited:


L420: High Member
May 24, 2009
Really what people should do with this is not take is as a completely necessary set of rules, but a framework that supports the gametype well. Adapt it to each given case and use it to help when you are trying to figure out what an area is missing.

(this is said as if grazr was saying it, not me saying it to grazr)


L13: Stunning Member
Feb 20, 2008
nice, i fully support anything that will encourage more multi-stage attack and defend maps ;)

2 things I would have liked to see though:
Going through the methods you can use to create increasing difficultly as the stages progress would be very useful for people to know.
And...pics from hammers are a bit...difficult to view, especially the overhead screenshots, the other ones are technically fine.

I wrote something similar to this actually, but I think the way you make it significantly more general makes it far better then mine.


Old Man Mutant Ninja Turtle
Mar 4, 2008
+Going through the methods you can use to create increasing difficultly as the stages progress would be very useful for people to know.

Hm, i suppose i could include something like this, although i have yet to see a map, even by Valve, that really manages to increase difficulty consecutively, properly. My experience is that difficulty generally goes: Hardest>easiest>moderate difficulty. Although i guess we can put a lot of that down to the fualt of the playerbase being mostly Nincompoops.

In terms of making things difficult-er per stage i feel there are too many options available to a level designer, when he can make almost any virtual scenario possible with the given materials. making things difficult-er could merely manifest in unbalancing the area in Red's favour by increasing the advantages for Red's classes, such as wider sniper views, more defencive platforms, higher uphill incline. Actually i'll go and make a note of that in the article if that suffices?

+And...pics from hammers are a bit...difficult to view, especially the overhead screenshots, the other ones are technically fine.

Maybe i spend too much time in hammer :/ I'll see what overhead ingame shots reveal, but the trouble is models fade out and fog fades in. Since people would be working in hammer they get a more accurate view of what they're looking at, whilst ingame textures are distorted (IE rendered correctly). I figured that would probably be more valueble to the beginner.


L13: Stunning Member
Feb 20, 2008
sv_cheats 1
cl_drawhud 0
hidepanel all
r_skybox 0
fog_override 1

My overhead screenshot cfg :p

One thing ive seen with dustbowl anyways is that the number of paths decreases between rounds, from what ive seen this does seem to make a defense easier for red.
I am pretty sure I went into alot more detail here: http://forums.tf2maps.net/showthread.php?t=4989
It was my attempt to train myself to analyze gameplay based on overhead screenshots, my next attempt on a map I only briefly played failed so I abandoned the idea <_<

And actually, the predominance of rocks/ other props fading out and nodraw is what makes it difficult to follow, its alot of yellow
Last edited:


Old Man Mutant Ninja Turtle
Mar 4, 2008
Hm, well one alternative is to turn off nodraw rendering in hammer.. After looking at your ingame shots all that red ground with red overlays, red rooftops and red cliffs is disorientating. Hammer provides greater contrast in the textures so i'll redo them but without nodraw.

The decreasing of the number of pathways is one possible way of increasing difficulty for Blu, although it really only turns things into a spam fest. The start of the third and final stage of dustbowl is statistically the hardest chokepoint, and one can see why when it only has one path, and despite its close proximity to Blu spawn, Red's direct shortcut negates any Blu advantage on the first point (another addition to the difficulty scale that does not exist on the first 2 stages), and often results in a spawn camp on Red's part ever since Valve moved the spawn visualisers back.

As a player this is just frustrating and as a designer it screams poor consideration. Although the additions of airblast and sandman came later, i find reducing paths to the point of one (and not that wide of one either in the case of dustbowl) is just an unimaginative response. Atleast this is my opinion; airblast and sandman only emphasise the point of poor gameplay experiences that this causes, since now allowing players to additionally outright halt any assault, including ubers, in its tracks. There is literally nothing Blu can do with 16 enemies and up to 4 additional sentry guns all firing into one corridor, where class choice really makes little difference. A feeling of helplessness was proven to be a bad experience with the introduction of the sandman. Taking away a player's control is dirty business and not providing alternative choices is either lazy or.. a lack of imagination and critical thinking.

future additions: discussion on one way doors, spawn shortcuts and reducing alternative pathways.

Hopefully get this done on Friday.


Mann vs Machine... or... Mapper vs Meta?
Sep 8, 2008
I disagree with a few images:

In this image you say the right exit for blue is minor while it actualy is used by most of the players. Its the most important route for them so its actualy major. It would also become equal to goldrush as that one had 1 major and 2 minor exits.

In this image you called the lower route a major exit while its actualy 2 minor ones. You cant call it major because actualy the top right exit is used more often then those 2.

Also, i was missing an overview image per stage showing the major routes and the minor routes next it and maybe even mark them with a number of how effective they are (from 1 to 5). Such image will also show how many options a team has which can be helpfull. A thing you will notice in those images is that in any later stage in dustbowl less routes will show up or that there are alot of hard to use routes.