Just a little something I'm working on.

[c0z3n]

Hey guys,

I just wanted to share a project I've been working on:

Experiments in the remapping of MW3 for predictive modelling
There's a brief description of the project on the page at the link, but I'm also going to mirror the description here for convenience.

The image above is a visualization of Dome, you may recognize it.

Part one - Collection and basic visualization

Because a project such as predictive modelling on a scale such as this will require quite a large amount of data, I've begun collecting it before I've written anything to analyze it on any advanced level. On certain events in the game (such as kills, etc.), as much data as possible about that event is logged to a number of specialized data structures.

In order to better visualize trends and other features of the data being collected and get a general feel for the amount, I've put together a visualizer using WebGL to render only the location of each data point in a small 3D environment.

If you want to jump right to the maps, here's a link for that.

I'm also going to mirror a short post I wrote aboue the project for my own blog:

3D mapping of relevant gameplay data

My recent project has been trying to determine a feasible way to make predictions about players based on the a given state, and a major part of that process has been logging and analyzing lots of data about lots of kills in lots of matches.
To better interpret the data as it was collected, I threw together a couple of visualizers so I could more easily see what was going on, and what we could determine from it.
The first visualizer is written in python using pygame. It creates a simple 3d environment and maps all the data locations collected for a given map. below is a screenshot of the visualizer showing some early data from “Interchange”.

I can also select data points to see more data associated with them, mainly information about who killed who from where (as shown above by the green line). The shading of the dots serves as an aid to interpreting elevation.
The second visualizer is a web based visualizer using the html5 canvas and WebGL.
I export only the locations of each data point (without any associated data) to a JSON object, and then parse and display that in 3D in the browser with javascript and WebGL. I again used shading to aid perception of vertical depth (darker lower).

Besides being a great way to visually inspect the data as I collect it, It makes a nice bit of art.
That’s it for now, I’ll update when I go further.

Again, You can see all the data I've collected so far about the MW3 maps live and 3d at c0z3n.org/mapping.

Thanks for reading, Happy hacking.

.............._____.........._____
............./.....|........|___..\
............/../|..|.........___\..\
.....______/../.|..|________|_____..\______
..../.....|..|..|..|___..../......|..|.....\
.../../|..|..|..|..|.../../..__...|..|..|\..\
..|..|.|__|..|..|..|../../..|..|..|..|..|.|..|
..|..|____|..|__|..|./../___|..|__|..|..|.|..|
...\______|________|/_______|________|__|.|__|

WARNING: These links and pages contain 3D WebGL content and thus require a couple of things to view them. First, you'll need webGL enabled in your browser. On most browsers it's enabled by default but if you're using something odd (mobile browser, tablet, etc.) I can't garuntee support. Also, Your computer needs to be able to do the graphics legwork behind it. If you experience a low framerate or stuttery image, I'm sorry. Some of the maps render upwards of 4000 points and for something not designed to be a 3d engine that starts to take a toll. This is just a demo, it's not perfect and I don't intend it to be.

DISCLAIMER: I know MPGH's policy on external links, which is why I spoke to Master131 via PM before posting this. Also, there are no ads anywhere, and I gain nothing financially or otherwise by posting this here. I just want to share. If any other mods have concerns please PM me or Master131. Thanks!