In May 2014 Torn Banner Studios announced a level design contest for their title "Chivalry - Medieval Warfare". Designers were given 4 months to create a Team Objective level containing at least 3 discrete objectives, focusing on gameplay, balance, polish and creativity. After 2 months the top 5 levels received prizes, along with detailed feedback: from 45 entries, Cove achieved 3rd place.
Work on Cove began on May 6th. I split the level into 3 stages each containing one main objective, along with optional objectives designed to keep players moving and thinking rather than heavily focusing on a single objective. My intention was to move players from an open field, into fighting between buildings, and finally close pressed combat, stressing accessibility and flanking. This works well with Chivalry's combat mechanics and class system, encouraging players to adapt to the changing environment. A set amount of time is given for each stage; if the attackers do not succeed in completing all objectives within the alloted time, the defenders win. If an objective is completed early, this time is kept and is added to the time for the next stage.
For the first stage I wanted to keep the play area physically small with open sightlines and a hard minimum completion time, in order to stop players rushing to complete the level within a few minutes instead of creating a more in depth experience. A modified take and hold objective worked well here, utilising a cannon which the attackers must hold for a set time to "load". Once loaded the cannon automatically fires at a portcullis causing damage.
The portcullis required two full shots to destroy. Between shots the cannon needed to cool down before reloading could begin again. This allowed the attackers to roam momentarily, or for the defenders to redouble their attack and force the attackers from the cannon.
During this stage the attackers spawn in a cave system behind the cannons, protected from camping and arrow fire. From there multiple paths spread between houses and light cover to the cannon situated in the centre of the space and lacking in cover. This allowed for flanking, and made it harder to camp.
Defenders spawned above the portcullis and dropped down onto a small ridge, and then again onto the ground. This two level design stopped attackers rushing the defenders' spawn room, and provided some high ground to defend from.
Two watchtowers were placed between the cannon and the defender spawn. These were available to both sides, offering high ground near the cannon. However both sides could chose to burn them down, burning any player caught inside. This added some risk/reward gameplay.
I released the initial test version, limited to the first stage, on 14th May. After receiving some community feedback I made some modifications in order to create some one-sided bottlenecks, to add some ridges to the landscape, and to create a secondary path against a cliff face, allowing the attackers to match height with the watchtowers.
At this point I began work on stage 2, returning to stage 1 as and when changes arose. The most important of such changes was to make the cannon placement dynamic: when the level started the game would now randomly select a cannon to unhide and activate. Players would begin the game without knowing where the cannon was going to spawn, helping to keep the first stage dynamic.
Stage two focused on class play, flanking, and sightlines with tighter space than the first stage. To facilitate this I chose a mobile objective: a battering ram. The attackers must push the ram though the city towards its target; they do this by remaining within a set area of the ram. Defenders must stop this by forcing the attackers away from the ram, taking control of it themselves. If the attackers outnumber the defenders in the set area, the ram moves forward, otherwise it remains stationary.
Height variation was key to this stage: the ram is pushed up a hill covered by two bridges; there are several paths from low to high ground including stairs and a lift; and there are windows and overhangs accessible both from the ground and from spawn points. This mix of heights allowed for the defenders to make skirmish attacks on the ram, and for the attackers to attempt to sneak behind the defenders' lines ahead of the ram.
To keep players near the action, player spawn locations would move forward once the ram passed pre-designated points. For ease of design I used several set spawn buildings, simply swapping which team spawned at that location.
The attackers begin from the defenders' stage one spawn, dropping down from a large gate that opens once the portcullis is destroyed. The ram beings play directly under this area, in front of the destroyed portcullis. From here attackers may head left to flank onto the bridge, or right to counterattack the defenders as the ram rounds the corner.
The defenders begin play slightly ahead of the ram, giving them plenty of time to get to position. The first bridge has decent cover for archers along with two oil pots that can be poured onto the enemy bellow. The defenders may also move down the left slope to meet the attackers as they round the corner; or they may chose to head over the first bridge to defend against flanks.
Once the ram is safely under the first bridge, the player spawns move forward. The attackers spawn from a building in the center of the map: from here they may take the lift up behind the second bridge, chose to head over the bridge (past the old defender spawn) and flank through buildings, or simply meet the defenders head on.
The defenders spawn in a building on the right of the ram's path. They have easy access to the bridge, on top of which are two ballistas that can be used to push attackers from the ram as it moves forward; or they may flank round toward the first bridge.
As the ram passes under the second bridge, the spawns move forward again. The attackers' spawn moves to the previous defenders' spawn, to the right of the ram path. This gives them access to the bridge and ease of access to the back of the ram, however limited flanking options.
The defenders move to a building adjacent to the lift with easy access to the bridge; this facilitates intense fighting on the high ground overlooking the ram. They also gain good access to the front of the ram and very short spawn-to-combat times.
As the ram moves forward, the defenders gain more and more resources to defend, and the attackers lose options, effectively managing the rate of progression as the combat dynamics change.
The second stage was released on 24th May, and in a similar fashion to stage 1, I began work on stage 3 while awaiting community feedback. Most of the changes were minor and related to the balancing of the speed of the ram vs. ease of defense.
Stage 3 focused heavily on internal space, continuing the progression from stages 1 and 2. I wanted to create a small space with limited sightlines to encourage players to swap from long range classes to melee classes. The design of the objective needed to take this into account; and after playing with a few concepts, I decided to use a five-part objective: a well-coordinated attack will be able to overwhelm defenders spread over 5 sub-objectives, while an unplanned assault will likely lack the strength to take objectives. As sub-objectives are completed, defenders need cover less area, trending towards defending a single objective. This can lead to a bottleneck over one objective. To avoid this, I chose to require only 4 sub-objectives out of the 5 to be completed.
I chose to set these objectives within the sacking of a cathedral, reminiscent of the reformation. This matches both the general theme of Chivalry and fits the attackers / defenders narrative Team Objective maps require. The 5 sub-objectives are as follows:
Stage 3 was released on 5th April. Following some testing some major changes were made to the cathedral spawns: moving the defenders' spawn room further back gave attackers more time to filter into the cathedral and helped reduce bottlenecking at the main door. Over several versions the health of individual objectives and the completion requirements were modified until the stage felt balanced.
Phase one of the contest ended 1st May with the winners announced 9th May. Cove attained 1st place in this phase. Developer feedback was provided for all maps entered for phase 1. The main suggestions were:
Additional changes I decided to make, based on community feedback, included adding extra cover via houses, increasing the shots required to destroy the gate, and adding torches to burn towers. I also added two more ballistas, one on the beach and one overlooking the top of the map. These helped the attackers push the advantage and helped push defenders off cannons.
Stage two received fewer changes. The largest and most important was moving a spawn building further from the ram path, helping to reduce the issue of archers attacking from spawn buildings, along with adding to the run time for defenders during the middle of the stage, something which had been problematic in earlier versions.
I chose to chiefly focus on art for the second phase, making gameplay modifications as testing continued. The following are a selection of work-in-progress shots of the initial art pass:
In total 130 meshes were created for Cove. This included an entirely new system of towers, walls, floor segments and trims, along with the entire cathedral set.
The final version of Cove was released 28th June with Phase 2 ending 30th June.
Phase 2 ended 30th June with the winners announced 11th August. Cove attained 3rd place in this phase and was patched into the game on 10th September, following a few small changes by Torn Banner.