pith. sign in

arxiv: 2605.30570 · v1 · pith:5H4CGNXUnew · submitted 2026-05-28 · 💻 cs.AI

Procedural Generation of First Person Shooter Maps using Map-Elites

Simone de Donato , Pier Luca Lanzi , Daniele Loiacono This is my paper

Pith reviewed 2026-06-29 06:46 UTC · model grok-4.3

classification 💻 cs.AI
keywords procedural content generationmap-elitesfps level designquality diversityevolutionary algorithmsgame ailevel representation
0
0 comments X

The pith

New Point-Line and Spatial-Layout representations let MAP-Elites generate more diverse and higher-quality FPS maps than All-Black or Grid-Graph methods.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper applies the MAP-Elites quality-diversity algorithm to procedural generation of first-person shooter levels. It replaces two established map encodings with Point-Line and Spatial-Layout encodings that better capture layout features. Topological metrics that depend only on geometry and emergent metrics that require simulated play are defined and screened for their usefulness as behavioral descriptors. MAP-Elites with Sliding Boundaries is then run on each encoding. The new encodings produce map populations that score higher on both diversity across the feature space and on the quality dimensions measured by the chosen metrics.

Core claim

When MAP-Elites with Sliding Boundaries is applied to FPS map generation, the Point-Line and Spatial-Layout representations, paired with a screened set of topological and emergent behavioral features, evolve map populations that exhibit measurably higher diversity across the illumination grid and higher scores on the quality metrics than the All-Black and Grid-Graph representations previously used for the same task.

What carries the argument

MAP-Elites with Sliding Boundaries (MESB) whose illumination grid is defined by the best-performing subset of topological layout metrics and emergent gameplay metrics, applied to Point-Line and Spatial-Layout encodings that describe FPS maps more informatively than earlier grid or all-black encodings.

If this is right

  • Point-Line and Spatial-Layout encodings produce map populations with higher coverage of the chosen behavioral feature space.
  • The same encodings also yield maps that score higher on the selected quality metrics than maps from All-Black or Grid-Graph encodings.
  • Screening metrics for suitability before illumination improves the effectiveness of the MAP-Elites run.
  • MESB can be used directly as a generator once the feature dimensions and quality functions are fixed.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same metric-screening step could be reused when adapting the method to other game genres whose levels have both geometric and dynamic properties.
  • If emergent metrics can be replaced by fast heuristics that still correlate with player experience, the whole pipeline could run without any gameplay simulation.
  • The feature-space coverage achieved by the new encodings might serve as a benchmark for future procedural generators that aim for controllable diversity.
  • Designers could inspect the final illumination grid to identify under-represented map styles and then bias the search toward those cells.

Load-bearing premise

The chosen topological and emergent metrics, after analysis, are the right features to drive MAP-Elites illumination and that simulated gameplay reliably indicates map quality for players.

What would settle it

Run the same MAP-Elites configurations on a fresh set of maps and collect blind player ratings or logged play statistics; if the new representations no longer show statistically higher diversity or quality scores, the central claim does not hold.

Figures

Figures reproduced from arXiv: 2605.30570 by Daniele Loiacono, Pier Luca Lanzi, Simone de Donato.

Figure 1
Figure 1. Figure 1: Heatmap of the archives generated with the [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Experiments with area and maxSymmetry features: (a) maximum entropy, (b) CCDF of the fitness of elites, (c) QD score of the archive, (d) size of the archive over time. which achieves the most elites in the archive. These results suggest that Spatial-Layout can generate more variety although Point-Line achieves similar results with higher area and maxSymmetry values. Analysis of Maps in the Archive. The map… view at source ↗
Figure 3
Figure 3. Figure 3: Best performing All-Black maps in the archive obtained with the area and maxSymmetry features. 6.4 pace and averageEccentricity In the second set of experiments, we applied MESB illuminating the search space using pace and averageEccentricity. The heatmaps in [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Best performing Point-Line maps in the archive obtained with the area and maxSymmetry features. variety it can generate. In contrast, All-Black, Point-Line and Spatial-Layout cover the archive in similar ways, although Spatial-Layout does it faster as already noted in the previous experiment (Figure 7d). Analysis of Maps in the Archive. Figures 8, 9, and 10 shows examples of the best performing (high entro… view at source ↗
Figure 5
Figure 5. Figure 5: Best performing Spatial-Layout maps in the archive obtained with the [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Heatmap of the archives generated with the [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Experiments with pace and averageEccentricity features: (a) maximum entropy, (b) CCDF of the fitness of elites, (c) QD score of the archive, (d) size of the archive over time. References [1] L. Cardamone, G. N. Yannakakis, J. Togelius, P. L. Lanzi, Evolving interesting maps for a first person shooter, in: EvoApplications (1), Vol. 6624 of Lecture Notes in Computer Science, Springer, 2011, pp. 63–72. [2] P.… view at source ↗
Figure 8
Figure 8. Figure 8: Best performing maps in the archive evolved with the [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Best performing maps in the archive evolved with the [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Best performing maps in the archive evolved with the [PITH_FULL_IMAGE:figures/full_fig_p015_10.png] view at source ↗
read the original abstract

We investigate the application of MAP-Elites (a well-known quality diversity algorithm) to design levels for First-Person Shooter (FPS) games. We consider two well-known map representations (All-Black and Grid-Graph) and introduce two novel representations (Point-Line and Spatial-Layout) that improve the characterization of FPS maps. We define a series of metrics to describe maps' topological properties (which solely depend on maps' layout), and emergent properties (which must be evaluated through actual gameplay). We perform an in-depth analysis to identify the most suitable features to guide MAP-Elites illumination process. We apply MAP-Elites with Sliding Boundaries (MESB) to evolve populations of FPS maps. Our results show that the new representations can generate maps with higher diversity and quality than the representations previously used for evolving FPS maps.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The manuscript applies MAP-Elites with Sliding Boundaries (MESB) to procedural generation of FPS maps. It evaluates two prior representations (All-Black, Grid-Graph) against two new ones (Point-Line, Spatial-Layout), defines topological metrics (layout-dependent) and emergent metrics (gameplay-dependent), performs an in-depth analysis to select illumination features, and claims the new representations produce higher diversity and quality than baselines.

Significance. If the empirical results prove robust and reproducible, the work would advance quality-diversity methods in game content generation by demonstrating improved map representations and the value of combined topological/emergent feature selection. The explicit analysis step for metric choice and the use of gameplay evaluation are positive elements that could support broader adoption in PCG research.

major comments (2)
  1. Abstract: the central claim that new representations yield higher diversity and quality rests on reported positive results from MESB, yet the text supplies no experimental details, sample sizes, statistical tests, raw data, or quantitative comparisons; this prevents verification of the claim and is load-bearing for the paper's contribution.
  2. Results (implied by abstract): the assumption that the selected topological and emergent metrics are optimal after in-depth analysis, and that gameplay evaluation reliably captures quality, is stated but not accompanied by the analysis data or validation steps needed to support the metric choice as load-bearing for the illumination process.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. The two major comments correctly identify areas where additional transparency is needed to support the central claims. We address each point below and commit to revisions that will incorporate the requested details without altering the core contributions.

read point-by-point responses

  1. Referee: Abstract: the central claim that new representations yield higher diversity and quality rests on reported positive results from MESB, yet the text supplies no experimental details, sample sizes, statistical tests, raw data, or quantitative comparisons; this prevents verification of the claim and is load-bearing for the paper's contribution.

    Authors: We agree that the abstract as written does not contain the experimental details needed to make the central claim self-contained. The manuscript body reports the use of MESB with multiple representations and metrics, but does not embed sample sizes, statistical tests, or quantitative comparisons directly in the abstract. We will revise the abstract to include concise quantitative results (e.g., relative improvements in QD-score and coverage) and will ensure the main text explicitly states run counts, statistical procedures, and data availability so the claim can be verified. revision: yes

  2. Referee: Results (implied by abstract): the assumption that the selected topological and emergent metrics are optimal after in-depth analysis, and that gameplay evaluation reliably captures quality, is stated but not accompanied by the analysis data or validation steps needed to support the metric choice as load-bearing for the illumination process.

    Authors: The manuscript describes performing an in-depth analysis to select illumination features but does not present the supporting data (e.g., correlation tables, ablation results across metric subsets, or validation against gameplay outcomes). We acknowledge this gap and will expand the relevant section to include the full analysis, including the rationale for the chosen topological and emergent metrics and any validation steps performed. This will make the metric-selection process reproducible and directly support its role in the illumination. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents an empirical study applying MAP-Elites with Sliding Boundaries to FPS map generation. It introduces two new representations, defines topological and emergent metrics after explicit in-depth analysis, and compares results against prior All-Black and Grid-Graph baselines. No equations or claims reduce by construction to fitted inputs, self-definitions, or load-bearing self-citations; the central result is a direct experimental outcome from standard QD algorithm application to novel inputs rather than any internal circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only abstract available; no explicit free parameters, axioms, or invented entities are described beyond standard evolutionary algorithm assumptions and the new map representations themselves.

pith-pipeline@v0.9.1-grok · 5666 in / 1090 out tokens · 29034 ms · 2026-06-29T06:46:37.049277+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

32 extracted references · 23 canonical work pages · 4 internal anchors

  1. [1]

    Cardamone, G

    L. Cardamone, G. N. Yannakakis, J. Togelius, P. L. Lanzi, Evolving interesting maps for a first person shooter, in: EvoApplications (1), V ol. 6624 of Lecture Notes in Computer Science, Springer, 2011, pp. 63–72

    2011

  2. [2]

    P. L. Lanzi, D. Loiacono, R. Stucchi, Evolving maps for match balancing in first person shooters, in: 2014 IEEE Conference on Computational Intelligence and Games, CIG 2014, Dortmund, Germany, August 26-29, 2014, IEEE, 2014, pp. 1–8.doi:10.1109/CIG.2014.6932901. URLhttps://doi.org/10.1109/CIG.2014.6932901

    work page doi:10.1109/cig.2014.6932901 2014

  3. [3]

    Cachia, A

    W. Cachia, A. Liapis, G. N. Yannakakis, Multi-level evolution of shooter levels, in: A. Jhala, N. R. Sturtevant (Eds.), Proceedings of the Eleventh AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, AIIDE 2015, November 14-18, 2015, University of California, Santa Cruz, CA, USA, AAAI Press, 2015, pp. 115–121. URLhttp://www.a...

    2015

  4. [4]

    Loiacono, L

    D. Loiacono, L. Arnaboldi, Fight or flight: Evolving maps for cube 2 to foster a fleeing behavior, in: 2017 IEEE Conference on Computational Intelligence and Games (CIG), 2017, pp. 199–206, iSSN: 2325-4289. doi:10.1109/CIG.2017.8080436. URLhttps://ieeexplore.ieee.org/abstract/document/8080436

    work page doi:10.1109/cig.2017.8080436 2017

  5. [5]

    P. T. Ølsted, B. Ma, S. Risi, Interactive evolution of levels for a competitive multiplayer FPS, in: 2015 IEEE Congress on Evolutionary Computation (CEC), 2015, pp. 1527–1534, iSSN: 1941-0026. doi:10.1109/CEC. 12 Procedural Generation of First Person Shooter Maps using Map-Elites (a) (b) (c) (d) Figure 8: Best performing maps in the archive evolved with t...

    work page doi:10.1109/cec 2015

  6. [6]

    Illuminating search spaces by mapping elites

    J.-B. Mouret, J. Clune, Illuminating search spaces by mapping elites, arXiv:1504.04909 [cs, q-bio] (Apr. 2015). doi:10.48550/arXiv.1504.04909. URLhttp://arxiv.org/abs/1504.04909

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1504.04909 2015

  7. [7]

    M. C. Fontaine, S. Lee, L. B. Soros, F. De Mesentier Silva, J. Togelius, A. K. Hoover, Mapping hearthstone deck spaces through map-elites with sliding boundaries, in: Proceedings of the Genetic and Evolutionary Computation Conference, GECCO ’19, Association for Computing Machinery, New York, NY , USA, 2019, p. 161–169. doi:10.1145/3321707.3321794. URLhttp...

    work page doi:10.1145/3321707.3321794 2019

  8. [9]

    Ballabio, D

    M. Ballabio, D. Loiacono, Heuristics for placing the spawn points in multiplayer first person shooters, in: IEEE Conference on Games, CoG 2019, London, United Kingdom, August 20-23, 2019, IEEE, 2019, pp. 1–8. doi:10.1109/CIG.2019.8848121. URLhttps://doi.org/10.1109/CIG.2019.8848121

    work page doi:10.1109/cig.2019.8848121 2019

  9. [10]

    de Donato, Quality diversity in procedural generation of first person shooter maps, Master’s thesis, Politecnico di Milano (Oct

    S. de Donato, Quality diversity in procedural generation of first person shooter maps, Master’s thesis, Politecnico di Milano (Oct. 2024)

    2024

  10. [11]

    Ballabio, Project Arena, https://github.com/MarcoBallabio/ProjectArena, [Online; accessed 17- March-2026] (2018)

    M. Ballabio, Project Arena, https://github.com/MarcoBallabio/ProjectArena, [Online; accessed 17- March-2026] (2018)

    2026

  11. [12]

    Tjanaka, M

    B. Tjanaka, M. C. Fontaine, D. H. Lee, Y . Zhang, N. R. Balam, N. Dennler, S. S. Garlanka, N. D. Klapsis, S. Nikolaidis, pyribs: A Bare-Bones Python Library for Quality Diversity Optimization, arXiv:2303.00191 [cs] (Apr. 2023).doi:10.48550/arXiv.2303.00191. URLhttp://arxiv.org/abs/2303.00191

    work page doi:10.48550/arxiv.2303.00191 2023

  12. [13]

    D. L. Simone de Donato, Pier Luca Lanzi, Quality diversity in procedural generation of first person shooter maps repository, https://github.com/SimoDedo/MAPElites_FPS_Maps, online; accessed 17-March-2026 (2025)

    2026

  13. [14]

    Togelius, G

    J. Togelius, G. Yannakakis, K. Stanley, C. Browne, Search-Based Procedural Content Generation, 2010, pp. 141–150.doi:10.1007/978-3-642-12239-2\_15

    work page doi:10.1007/978-3-642-12239-2 2010

  14. [15]

    CUBE2: Sauerbraten,http://sauerbraten.org, [Accessed 2025-01-18] (2004)

    2025

  15. [16]

    Bhojan, H

    A. Bhojan, H. W. Wong, ARENA - Dynamic Run-Time Map Generation for Multiplayer Shooters, in: Y . Pisan, N. M. Sgouros, T. Marsh (Eds.), Entertainment Computing – ICEC 2014, Springer, Berlin, Heidelberg, 2014, pp. 149–158.doi:10.1007/978-3-662-45212-7\_19

    work page doi:10.1007/978-3-662-45212-7 2014

  16. [17]

    Wikipedia contributors, Flood fill — Wikipedia, the free encyclopedia, https://en.wikipedia.org/wiki/ Flood_fill, [Online; accessed 24-March-2026] (2024)

    2026

  17. [18]

    Intentional Computational Level Design

    A. Khalifa, M. C. Green, G. Barros, J. Togelius, Intentional Computational Level Design, arXiv:1904.08972 [cs] (Apr. 2019).doi:10.48550/arXiv.1904.08972. URLhttp://arxiv.org/abs/1904.08972

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1904.08972 1904

  18. [19]

    V . R. Warriar, C. Ugarte, J. R. Woodward, L. Tokarchuk, PlayMapper: Illuminating Design Spaces of Platform Games, in: 2019 IEEE Conference on Games (CoG), 2019, pp. 1–4, iSSN: 2325-4289. doi:10.1109/CIG.2019. 8848118. URLhttps://ieeexplore.ieee.org/document/8848118

    work page doi:10.1109/cig.2019 2019

  19. [20]

    M. C. Fontaine, R. Liu, A. Khalifa, J. Modi, J. Togelius, A. K. Hoover, S. Nikolaidis, Illuminating mario scenes in the latent space of a generative adversarial network, Proceedings of the AAAI Conference on Artificial Intelligence 35 (7) (2021) 5922–5930.doi:10.1609/aaai.v35i7.16740. URLhttps://ojs.aaai.org/index.php/AAAI/article/view/16740

    work page doi:10.1609/aaai.v35i7.16740 2021

  20. [21]

    Alvarez, S

    A. Alvarez, S. Dahlskog, J. Font, J. Togelius, Empowering Quality Diversity in Dungeon Design with Interactive Constrained MAP-Elites, in: 2019 IEEE Conference on Games (CoG), 2019, pp. 1–8, arXiv:1906.05175 [cs]. doi:10.1109/CIG.2019.8848022. URLhttp://arxiv.org/abs/1906.05175 16 Procedural Generation of First Person Shooter Maps using Map-Elites

    work page doi:10.1109/cig.2019.8848022 2019

  21. [22]

    S. O. Kimbrough, G. J. Koehler, M. Lu, D. H. Wood, On a Feasible–Infeasible Two-Population (FI-2Pop) genetic algorithm for constrained optimization: Distance tracing and no free lunch, European Journal of Operational Research 190 (2) (2008) 310–327.doi:10.1016/j.ejor.2007.06.028. URLhttps://www.sciencedirect.com/science/article/pii/S0377221707005668

    work page doi:10.1016/j.ejor.2007.06.028 2008

  22. [23]

    Charity, M

    M. Charity, M. C. Green, A. Khalifa, J. Togelius, Mech-elites: Illuminating the mechanic space of gvg-ai, in: Proceedings of the 15th International Conference on the Foundations of Digital Games, FDG ’20, Association for Computing Machinery, New York, NY , USA, 2020.doi:10.1145/3402942.3402954. URLhttps://doi.org/10.1145/3402942.3402954

    work page doi:10.1145/3402942.3402954 2020

  23. [24]

    General Video Game AI: a Multi-Track Framework for Evaluating Agents, Games and Content Generation Algorithms

    D. Perez-Liebana, J. Liu, A. Khalifa, R. D. Gaina, J. Togelius, S. M. Lucas, General Video Game AI: a Multi-Track Framework for Evaluating Agents, Games and Content Generation Algorithms, arXiv:1802.10363 [cs] (Feb. 2019). doi:10.48550/arXiv.1802.10363. URLhttp://arxiv.org/abs/1802.10363

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1802.10363 2019

  24. [25]

    González-Duque, R

    M. González-Duque, R. B. Palm, D. Ha, S. Risi, Finding Game Levels with the Right Difficulty in a Few Trials through Intelligent Trial-and-Error, arXiv:2005.07677 [cs] (Jun. 2020).doi:10.48550/arXiv.2005.07677. URLhttp://arxiv.org/abs/2005.07677

    work page doi:10.48550/arxiv.2005.07677 2005

  25. [26]

    B. M. F. Viana, L. T. Pereira, C. F. M. Toledo, Illuminating the Space of Dungeon Maps, Locked-door Missions and Enemy Placement Through MAP-Elites, arXiv:2202.09301 [cs] (Apr. 2022). doi:10.48550/arXiv.2202. 09301. URLhttp://arxiv.org/abs/2202.09301

    work page doi:10.48550/arxiv.2202 2022

  26. [27]

    Charity, A

    M. Charity, A. Khalifa, J. Togelius, Baba is y’all: Collaborative mixed-initiative level design, in: 2020 IEEE Conference on Games (CoG), 2020, pp. 542–549.doi:10.1109/CoG47356.2020.9231807

    work page doi:10.1109/cog47356.2020.9231807 2020

  27. [28]

    Talakat: Bullet Hell Generation through Constrained Map-Elites

    A. Khalifa, S. Lee, A. Nealen, J. Togelius, Talakat: Bullet Hell Generation through Constrained Map-Elites, arXiv:1806.04718 [cs] (Jun. 2018).doi:10.48550/arXiv.1806.04718. URLhttp://arxiv.org/abs/1806.04718

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1806.04718 2018

  28. [29]

    Gravina, A

    D. Gravina, A. Liapis, G. N. Yannakakis, Constrained surprise search for content generation, in: IEEE Conference on Computational Intelligence and Games, CIG 2016, Santorini, Greece, September 20-23, 2016, IEEE, 2016, pp. 1–8.doi:10.1109/CIG.2016.7860408. URLhttps://doi.org/10.1109/CIG.2016.7860408

    work page doi:10.1109/cig.2016.7860408 2016

  29. [30]

    Shaker, J

    N. Shaker, J. Togelius, M. J. Nelson, Procedural Content Generation in Games, Springer-Verlag, 2016

    2016

  30. [31]

    org/wiki/Rogue_(video_game), [Online; accessed 9-January-2025] (2024)

    Wikipedia contributors, Rogue (video game) — Wikipedia, the free encyclopedia, https://en.wikipedia. org/wiki/Rogue_(video_game), [Online; accessed 9-January-2025] (2024)

    2025

  31. [32]

    J. Whitehead, Spatial Layout of Procedural Dungeons Using Linear Constraints and SMT Solvers, in: Proceedings of the 15th International Conference on the Foundations of Digital Games, FDG ’20, Association for Computing Machinery, New York, NY , USA, 2020, pp. 1–9.doi:10.1145/3402942.3409603. URLhttps://dl.acm.org/doi/10.1145/3402942.3409603

    work page doi:10.1145/3402942.3409603 2020

  32. [33]

    Hullett, J

    K. Hullett, J. Whitehead, Design patterns in FPS levels, in: Proceedings of the Fifth International Conference on the Foundations of Digital Games, ACM, Monterey California, 2010, pp. 78–85. doi:10.1145/1822348.1822359. URLhttps://dl.acm.org/doi/10.1145/1822348.1822359 17

    work page doi:10.1145/1822348.1822359 2010