top of page

Dota2 Maps; Dynamics of its Spatial Layout

Context

Dota2 Minimaps - v 7.23, 7.20, 7.07, 7.00 (Source)

Accessibility VGA - v 7.23, 7.20, 7.07, 7.00 

Patch changes in games are part and parcel of keeping the gameplay relevant. Dota2 has been notorious for having a steep learning curve - being "unnecessarily complex" with in-game mechanics and interactions. As such, match analysis platforms (such as OpenDota, Dotabuff, Stratz, etc) have been providing empirical data and statistics for players of all skill levels to understand the patch changes better - keeping ahead of the "meta".

Traditionally, the Dota2 community has been much more invested in the items/hero updates where it tends to be characterized more prominently in official patch notes as well as 3rd party analysis platforms. Players, after reading a couple of numbers, are quick to understand how the gameplay at large has evolved. On the other hand, spatial updates tend to be more elusive - changes are merely described in the patch notes as "Terrain updated with various changes" and players spend the following weeks or so trying to get used to the revisions.

Granted, the layout changes are harder to capture - there isn't an easy/widely accepted way to conceptualize spatial changes and their implication to the overall gameplay. As such, it's usually thrown to very qualitative forms of analysis in the form of resulting experiences. Players have been using numerous coined terminologies such as "high ground advantage", "chokepoints", and "fog of war" to describe common strategies.

 

I personally feel that the community's comprehension of the unique game space is rather limited as compared to the advancements in hero/items/match analysis. This could be because of its reliance on spatial knowledge on an empirical or a posteriori basis (from observations or experiences to the deduction of probable causes).

Objective: 

Establishing a means to understand the game space through theoretical deductions

As hinted above, I believe that there is great potential to understand spatial characteristics of the game a priori - using theoretical mental frameworks (as opposed to empirical) which I feel could be improved in modern analytics.

Why might this be useful? I thought new spatial concepts could be derived - a stronger knowledge base and vocabulary could be built up within the community to perceive the game in new and more informed ways. Players can also understand the game better at the start of each patch. Also, this could benefit the game developers too, where the tools and ideas explored can inspire new methods to design/prototype the new game space better to align to the intentions of the game patch they intend to introduce.

In this brief experiment, I will build a spatial model to quantitatively analyze the game space, and try to establish objective characteristics which have changed between patches.

 

I am mainly going to model accessibility. Honestly, because visibility in the Dota2 sphere tends to have unorthodox rules like the elevation differences (a player can see elements on this level or lower, but not higher).

2

4

5

7

10

Full.png

Methodology

Technical Pipeline:

> data acquisition (R)

1) Download Raster Files from devilesk's Interactive Map

2) Crop Raster Files and Combine Layers (Trees and Boundary)

3) Convert Rasters into Polygons/Polylines

> data processing (DepthmapX)

4) Import Data into DepthmapX

5) Run VGA - Visual Graph Analysis

> data preprocessing and visualization (R)

6) Import VGA as SpatialPixelsDataFrame

7) Plot SpatialPixelsDataFrame

8) Build Time-Series Plot Animations

9) Resize SpatialPixelsDataFrame to fit map image

                                                            (260x260) -> (4096x4096)

 

> analysis (QGIS)

10) Visualize in QGIS alongside full Map Images

* You probably could do this manually, but because of the large number of map files- I did it as a batch process mostly in R.

Preliminary Observations

- In Progress-

So...What has changed so far?

- In Progress-

bottom of page