CROSSING
CROSSING
CROSSING
MICHAEL KIRBY
TRICKS USED IN LEVEL DESIGN
Level design is something that takes place in every video game from 2D to 3D, side scrollers to RPGs (Role Player Games). With level design there are ‘tricks’ being used that affect player’s experiences, whether they are aware of it or not. For my final project at Goldsmiths University of London, I researched the brief history of level design and topics of Architecture, Design Patterns, Illusion of Choice and 10 Level Design Principles. From there I was able to identify some 'tricks' that stood out and looked at case studies to prove there usefulness of them in video games.
If you want to read my finds and final studies then click the button below to read my full dissertation, however on this page I go over the 10 'Tricks' in a brief overview what my overall results were.
THE 10 'TRICKS'
From my research I was able to identify 10 different types of untold 'tricks' being used in level design, they were:
-
Starting Points
-
Signposting
-
Bread Crumbing
-
Lighting
-
Leading Lines
-
Architecture
-
Teaching Mechanics
-
Denial and Reward
-
Soft boundaries
-
Solid boundaries
With these 'tricks' identified I created initially 8 prototype levels and a custom Analytic Tool (from the help of my mentor) to use in Unity to gather data from play sessions and export them out as CSV files. I organised a first batch of play testing to gather data, I found out through my initial data, that the analytic tool worked and that I was gathering too much data at once. So I learnt that I needed to simplify the data I wanted to gather from specific metrics in the levels. This was useful to learn early on and from there I iterated the levels and metrics for preparation for the second batch of tests.
For the second batch of testing I made twenty prototype levels that incorporated all the 'tricks' into the level designs. From there I was able to organise a second batch of A-B testing, this time with twenty players.
The gallery below show the different 'tricks' in the levels I created in Unity.
MY RESULTS
The second batch of tests were successful with more people participating in the testing, because of this I was able to gather more sufficient data for my findings. With the CSV files it was easy to transfer the data into Excel to create graphs and tables to visually showcase the data and see differences from groups 1 and 2.
The first graph above demonstrates how much time players spent over in each of the selected levels. Here the purpose was to see which group stayed in the levels the longest compared to what groups speed run through the levels. The boxes represent inter quartile range (IQR = 75% of players’ time are in the boxes). The lines through the boxes are equal to median (= separated half of players’ time) and the lines from top to bottom are the minimum and maximum values.
We can see straight out of the gate that group 1 in levels 1, 3 and 4 spent a significant amount of time in those levels compared to group 2. Level 1 had the bread crumbing 'trick' in it, level 3 had the soft boundary in it and level 4 had the solid boundary 'trick' implemented. Level 10 there is not a huge amount of change between the groups, this level was about using architecture to influence players movement, but group 1 did spend a small slice of time more in the level compare to group 2. Overall these levels were successful in keeping players in the level rather than finishing quickly.
The second graph above demonstrates again how much time players spent over in each of the selected levels. Here the purpose was to see which group complete the levels the fastest. Again here the boxes represent inter quartile range, the lines through the boxes equal the median and the top to bottom lines are the minimum and maximum values.
We can see group 1 in levels 2, 7, 8 and 9 completed the levels quickly compared to group 2. Level 2 had the teaching mechanics 'trick' in it and I used what I learnt about the Bethesda Loop of Learn, Play, Challenge and Surprise to reinforce this one. Level 7 had the starting points in it and level 8 had the denial and reward 'trick' implemented in it, level 9 had the signposting one. These levels were successful in getting players in group 1 to complete levels quick than group 2. Levels 5 and 6 seemed to have not been as effective as I would have liked. Level 5 was lighting and level 6 was leading lines.
In this third graph it showcases the percentage of players who collected gold orb collectables through out levels 3, 5 and 7. We can see more players in group 1 collected the gold orbs compared to group 2 players who collected significantly less. With level 3 showing a large difference between both groups, level 5 shows a 30% difference and level 7 shows a 50% difference. This graph is used to support the evidence of these 'tricks' being successful in their respect levels.
The last visual to showcase is a heatmap table that I put together in Excel, as through my research I could see heatmaps provided more detailed and in-depth data. The purpose of this table is to show where players from both groups have been throughout the levels, which was used to support my evidence of success and show why in particular levels 5 and 6 were not as successful as I would I have wanted.
OVERALL
In conclusion I believe that this project has strong legs to stand on, it can be further developed to explore more avenues. The ‘tricks’ can definitely benefit from having further testing down the line, as for this project I focused more on the ‘tricks’ being used in a 3D platformer level design structure, but these ‘tricks’ could be tested on different types of level designs. With open world levels, multiplayer maps, strict linear levels and semi open world levels, the possibilities are endless. Overall, my goal was for me to learn more about level design, organise A-B testing, gather player data and produce results. In the end I achieved that goal and I have gained a strong understanding on level design both practically and theoretically, and even though some ‘tricks’ did not turn out the way I wanted them to, I still learnt from them and even gained skills in analysing player data. Which is an area I had little to no experience in.