Schedule

Towards Field Experiments in Esports Competition

Arjun Madhusudan, Josef Spjut, Ben Boudaoud, Joohwan Kim, Benjamin Watson

Presenter: Arjun Madhusudan

Esports is having a significant societal impact. Yet scheduling competitive group play during experiments is difficult, and recreating competitive environments in the lab is challenging. To address these problems, we adopt the field experiment methodology, to create hybrid tournament-experiments — experimentation integrated into the tournament itself. We run two Rocket League tournament-experiments and discuss their methodological and experimental implications.

What Do Competitive Players Do With Their Settings? A Multi-Game Esports Configuration Survey

Ashish Rajpurohit, Arjun Madhusudan, Benjamin Watson

Presenter: Ashish Rajpurohit

Competitive PC games typically expose twenty or more graphics settings to players, each setting representing a tradeoff between visual quality and interactive performance. How players actually work through making these tradeoffs is not well documented. We report findings from an ongoing survey of players across 14 esports titles spanning five genres, combining self-reported configuration preferences with parsed configuration files uploaded from participants’ machines. Across 44 game-level responses, 75% of players prioritized frame rate over visual quality, and 86% of the configuration files uploaded had VSYNC disabled. This preference is not uniform across genres: every tactical shooter and battle royale response favored performance, while MOBA and sports game responses were more divided. We also asked players to rate their own understanding of the graphics settings and the usability of each game’s settings interface. More than half rated themselves Advanced or Expert, though this varied by genre; players reporting a stronger understanding were more likely to prioritize performance. For rendering engineers and game designers, these patterns argue against treating “competitive defaults” as a single target. Responses to the usability question suggest that the complexity of current settings menus may affect different player populations differently.

Tactical Telemetry: Using Replay Files for Gameplay Analytics

Peter Xenopoulos

[Invited]

Presenter: Peter Xenopoulos

Details to be announced.

AI Level of Detail: Distance-Aware ML Model Precision Selection for Real-Time Human Motion Prediction in Games

Mathew Varghese

Presenter: Mathew Varghese

Modern game engines spend significant compute animating NPCs with learned motion models. This paper proposes AI Level of Detail (AI LOD), a framework in which machine learning inference precision is adapted based on the distance between each NPC and the player camera. The core idea mirrors classical geometry LOD [2, 4]: substitute a cheaper approximation where the difference is imperceptible. Here, the approximation is a lower-precision quantized machine learning model rather than a lower-polygon mesh. The contribution of this work is the AI LOD concept itself: that inference-time quantization can serve as the LOD axis for AI-driven character animation—and more broadly, for any AI-based runtime system where perceptual sensitivity varies with context. The convolutional sequence-to-sequence model of Li et al. [7] is used as a representative example to demonstrate the concept, with its trained checkpoint exported into three ONNX Runtime variants (FP32, FP16, and INT8 per-tensor), intended to be routed by a distance-based selector at runtime. Evaluation on the CMU Mocap dataset [1] provides initial evidence that each precision tier can be served at its assigned distance range with negligible perceptible degradation, supporting the broader premise that distance-aware ML model precision selection is a viable LOD strategy for AI-based character animation

Improving Interaction by Transforming the Render-Display Pipeline

Aaron Fulmer, Evan Jonson, Hee-Jin Choi, Hyoungsik Nam, Benjamin Watson

Presenter: Aaron Fulmer

Over fifty years ago, rendering and display systems were designed for latency-free, recorded media. Yet today’s interactive systems — which suffer significantly from latency — are still based on the old latency-free design assumptions. We propose an alternative: close integration of input, render, and display that reduces interactive latency far below delays in current systems. By combining just-in-time input sampling; frameless, image-order rendering; and random-access displays supporting flexible scanning; we eliminate frame and synchronization delays, mitigate refresh delays, and reduce click-to-photon latency below 2 ms.

Understanding Emergent Non-Verbal Communication in the Delta Force Competitive Video Game through Multimodal AI Analysis

Jinyuan Guo, Josef Spjut

Presenter: Josef Spjut

Non-verbal communication plays a critical role in multiplayer games, especially in environments where verbal communication is limited or constrained by game mechanisms. Players often rely on gestures, movement patterns, item interactions, and UI signals to communicate intent, negotiate cooperation willingness, and avoid conflict. In our work, we use gameplay video as the input, and analyze it to extract non-verbal communication. We propose a multi-modal pipeline for detecting and interpreting non-verbal communication in gameplay videos. Our research combines pose estimation, visual-textual extraction, and audio analysis to capture diverse behavioral signals, which are then integrated using a large language model to infer player intent. Using gameplay data collected from publicly available online videos, we conduct preliminary analyses suggesting that a wide range of meaningful non-verbal interactions can be systematically extracted and interpreted. Through this study, we aim to develop an empirical understanding of emergent non-verbal communication in modern multiplayer games and a practical framework for analyzing such behaviors using AI.