Controllable Procedural Content Generation via Constrained Multi-Dimensional Markov Chain Sampling

Authors: Sam Snodgrass, Santiago Ontañón

IJCAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test our approaches by generating maps for two classic video games, Super Mario Bros. and Kid Icarus. and 4 Experiments We test our algorithms by sampling maps for two classic video games (Super Mario Bros. and Kid Icarus). and Table 1 shows the results of our experiments.
Researcher Affiliation Academia Sam Snodgrass, Santiago Ontañón Drexel University Philadelphia, PA USA sps74@drexel.edu, santi@cs.drexel.edu
Pseudocode Yes Algorithm 1 Generate And Test(width, height, C), Algorithm 2 Violation Location Resampling(w, h, C), Algorithm 3 Incremental Sampling(Sec W , h, n, C)
Open Source Code No All datasets used in our experiments can be downloaded from https://sites.google.com/site/sampsnodgrass. No explicit statement or link for the methodology's source code was found.
Open Datasets Yes All datasets used in our experiments can be downloaded from https://sites.google.com/site/sampsnodgrass
Dataset Splits No The paper describes training maps and then sampling new maps, but it does not specify explicit training, validation, or test data splits (e.g., percentages or counts for data partitioning) for reproducibility.
Hardware Specification No Note that on a standard laptop the baseline Md MC approach is able to sample 50 complete maps in about one second. This description is too general and does not provide specific hardware models or detailed specifications.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup Yes In our experiments we set those parameters to the following values: for Super Mario Bros. we configure an Md MC with rowsplits = 12 and look-ahead = 3. For Kid Icarus we configure an Md MC with rowsplits = 10 and look-ahead = 3, which were the most promising configurations over some preliminary runs without constraints. and Location-aware constraints experiments return sections of size 10 x 12 for Super Mario Bros. and 16 x 10 for Kid Icarus (width by height).