What are these? Controls for ants?

Empires of the Undergrowth has gone through many iterations in many areas since the beginning. We began building the Unreal Engine version of the game in January this year so soon it will have been in production for a whole year, however the first builds of the game were in our own engine long ago, and the game played very differently then.

 

The Early Concept

The idea started as Dungeon Keeper but with ants for mobile. We went about setting up a game where you would fight subterranean wars with other colonies, tunnelling out food and growing. One of the core gameplay concepts was always to have the player focus on the bigger picture rather than individuals, after all, an ant colony is not about the individual ants themselves.

The question arose: how do you control something that is by its very nature autonomous and numerous?.. managing that would be a nightmare to master. We did not want the player to have direct control as this would defy the point of the game, the creatures should really have their own minds. We decided to automate everything they did, if an enemy appeared they would attack it, if you tunnelled into the enemy colony, your soldiers would head right for it.

Fancying the challenge, we built our own engine… Fast forward a year and a half our buggy engine is running an early build of the game but it was missing something. This subterranean warfare didn’t feel right, it wasn’t doing ants justice.
1eairlybuild

Back to the Drawing Board

The simple fact is ants don’t stay underground, they forage on the surface. Our ambition getting the best of us, we decided that is what we needed. Not a single underground, but one for each colony.

Under this system our automated attacking would play out like a Tug of War. Workers would constantly bring eggs to soldier tiles and new and new ants would be born. These ants would clash with the opponent on the surface in epic battles growing in size;  the player needed to mix up their colony composition to gain an advantage, pushing the opponent back and killing their queen.

In principal this was a tried and tested game mechanic borrowing heavily from a custom map in Starcraft 2 – Desert Strike. This was manageable as well and we could bend it to fit into a campaign idea that had emerged by this point that we had fell in love with.

2tugofwarnew

It was with this concept we launched our first Kickstarter which didn’t completely fall flat, but it did fall short. We were not looking for a huge amount of money, and by the end we had been seen by a lot of people, but the backing was not coming through. Something was wrong.

 

Do people really want a mobile version?

Ask someone who likes our game concept if they want a mobile version as well a PC and they will say yes. Ask them which one they would prefer and the answer seem to heavily favour the PC version. We used the data from the kickstarter the make a judgment on this. Looking at the backers for mobile only and PC only, it is 20 on mobile and 149 on PC. Not huge numbers but it does give you an idea.

So we cut the mobile version, and with it our own engine. We had stuck with our own engine to this point for performance reasons, in order to have so many ants running well on mobile technology, but with that requirement gone we were just making life difficult for ourselves.

And with all this changing, do we really need the controls to be so simplistic? With a mouse we can now give players more control. You could argue that this is feature creep, but until the features match what player will pay for you probably need to add more.

The pheromone marker made its first appearance in the demo we released with our second Kickstarter: A point that soldiers would be attracted to, instantly giving players a level of control but not so much that it destroyed the original bigger picture concept. Players seemed to get on well with this too. This was the first public demo build we had released and streamers had started picking it up. The system was intuitive, responsive and still felt like the ants were in control of themselves. Underground this works very well.

3pheremonemarker1

 

 

Making this mesh with the surface

So you haven’t played the surface have you? We made a test build for the surface just before EGX and we took it along to see what people thought (some will have played it but not many).

The controls worked, but it was not right. With such a wide open space and with groups of ants hovering around a single point it felt a bit strange. Almost too much like a standard strategy game and less autonomous. What was great in the tight corridors of the nest was falling a bit flat above ground.

4singlepointproblem

 

 

Enter trails

In nature ants will wander to and from the nest, feeding other ants on their way if they find a food source thus reinforcing a route that leads to food. Our game being more fast paced than a simulation we wanted a way to replicate this but in a short space of time.

The first iteration of trails is what you see in the 0.0561 build you can download. Ants will head to a point, wander around the point for a few seconds and then head back to the nest. The amount of ants heading to the point is based on the number you have assigned to it and the distance from the colony. It essentially tries to balance the number of ants ant each point of the trail. We built in a mechanism to prevent the player having control where, if you place a new point down, ants must return to the nest before heading to the new point. In principle this should mean when you place a new point, the flow of ants coming from the nest go to the new point so it is still fairly responsive but what we found was when players had few ants (less than about 50) the system felt frustrating. Quite a few liked the system, but quite a few others did not, and we can very much see why. As a developer I have been playing the game constantly as I change things, training myself to think in the new systems and understanding them intrinsically. The updated systems were not intuitive and some people flat out thought they were broken.

5trails

 

 

What is next?

We did a further iteration of the system. We wanted it to feel more responsive but we are trying to avoid something quite specific. Imagine you have a trail of ants going back and forth and you change the end point they are heading to, the whole trail changes direction at once. This looks and feels bad and is not at all like ants.

The latest implementation maintains a number of ants at the end of the trail, and as a new one arrives, one heads home. This keeps a fighting force ready to attack. These ants at the end also respond to new markers being placed and will head to them. Ants on their way or coming back from a trail will remain doing so. In this way we have trails going back and forth, but the system also feels responsive

latestcontrol

As we move on with the game we will be revisiting the controls, and we want to get this new system into your hands as soon as possible so we can get feedback. I think you will agree it is much improved we will continue to improve upon it as the game develops