Week 11 Testing

I did some playtesting with 2 of my friends and the results were particularly interesting.

Participant 1 notes:

Started game

Knows to avoid enemies

Walks into 2 enemies

Ran into another enemy

Got caught

Created a “train” of enemies”

Had half the enemies in the level follow him around

Got caught

Got caught again

Says he enjoys exploring

Says he can easily get lost

Expressed the fact the door appearing makes the camera bug out a little bit.

-Mouse sensitivity too high, needs to be fixed for next test

Wants enemies to go back to patrol

Participant 2 notes:

Started game

Understands movement

Understands to avoid enemies

Wants to know where to go

Got seen by 2 enemies.

Escaped.

Got seen by another enemy

Is running around

Got caught

Is being careless, but isn’t really getting punished

Made an enemy train for fun

Dies

Follows enemy around for fun

Enemies stuck on walls

Survey Results

Both participants expressed that they enjoyed the prototype, both rating the system 4/5. Both admit they were caught off guard by the enemy. Both participants were disoriented by the level system which indicates it has met the desired effect. One participant though the enemy speed was fine while another thought it was slightly too slow. In terms of the enemy view cone, 1 participant thought the range was fine while another thought it was far too short. both participants thought the view cone field of view was too narrow. Both participants thought the difficulty was too easy.

Additional notes

Whenever a participant entered a new room, the game would lag for half a second. I suspected it could be the fact I'm generating up to 3 rooms at once, however, the profiler shows that the cause is actually the pathfinding system and the maths adds up when you consider I can generate 3 rooms with 90x90 grids with probably about 10 enemies who each have to use pathfinding and on top of that, generally inefficient coding that I could have done better had I planned a little more and had more experience with pathfinding. Luckily, its not my job to make the game performance good, and for this prototype its a minor inconvenience. Next time I do this, I might take a look at different graph representations so that I'm not using a grid which is very expensive, especially if it were 3 dimensional. I will also look at multithreading so that I can run the pathfinding for each AI in parallel and that should increase performance.

Conclusion

The results show that both participants (funnily enough) exploited the fact the enemy was slower than them to create "enemy trains" as one participant calls it. I initially made the enemy slightly slower than the players so that the player could be given a chance to get away, however this has caused the the AI to be too easy and this is backed up by the survey. Both participants also understand that enemies need to be avoided which is one the desired effect of the view cones. Both participants were captured by the enemy, normally because they were caught off-guard by an enemy around a corner, which is a desired effect.

Another issue participant 1 expressed comes with the camera. When a door appears, the camera could still be in the other room which causes the camera to quickly jolt into the next room.

One of the more interesting parts of participant 1 was that he expressed his enjoyment of exploring which was an aspect I didn't initially consider. While the scope of the project would be too large to further implement upon this, I would love to design ideas on what I could do to further emphasise that aspect if I were to work on this in the future which is one of the things I will try to work on in the coming weeks.

Testing the game with other people has definitely paid off as I probably wouldn't be able to identify these design issues myself.

Adjustments that need to be made

I will speed up the enemy so that the enemy is slightly faster than the player. I think I can code a sort of line of sight system for the AI where they follow a set number of "breadcrumbs" from the player until they go back to patrolling and I think it will make the difficulty more balanced because the level design has lots of line of sight breaks I think players will be able to use to balance the fact that enemies are faster. This should also stop the problem of AI getting stuck at walls.

To stop the camera from jolting, I will probably adjust the room trigger box to be further away from the door which should hopefully fix the problem.

I will adjust the view cone have a slightly longer range and to be wider.