Masked Diffusion Models (MDMs) provide an efficient non-causal alternative to autoregressive generation but often struggle with token dependencies and semantic incoherence due to their reliance on discrete marginal distributions. We address these limitations by shifting the diffusion process into a continuous sentence-level semantic space. We propose CRoCoDiL -- Continuous and Robust Conditioned Diffusion for Language --  a unified fine-tuning approach that jointly trains an encoder–demasker architecture, grounding the MDM demasking in continuous latent representations.

Frametime spikes can disrupt gameplay in first-person shooter (FPS) games, affecting both performance and player experience. This paper examines how spikes during specific game events impact players. We developed a custom FPS game that maintains a steady 500 frames/s while inducing frametime spikes during weapon reloading, fast mouse movement, or targeting. Thirty-eight (38) participants played the game in a user study, providing both performance data and user-reported visual smoothness.

User studies are a cornerstone of human-computer interaction research, including measures of user performance and quality of experience (QoE) – particularly important for games where frame rates and frame timings can impact performance. Unfortunately, commercial games have limited options for customization and do not log player performance data with sufficient detail for use in such studies. This paper introduces Lead Rush, a first-person shooter game designed for conducting user studies on the effects of frame timing and frame rate.

Frametime spikes and graphical fidelity both matter for the feel of first-person shooter (FPS) games, yet their combined effects are not well understood. This paper examines how graphics settings and frametime spikes during aiming interact with player performance and experience. We developed a custom FPS game with configurable textures, lighting, and visual effects, and induced frametime spikes of 0 ms, 225 ms, or 675 ms during play.

Camera-based physiological signal estimation provides a convenient and non-contact way to monitor heart rate, but it also raises serious privacy concerns because facial videos can leak sensitive information about a person’s health and emotional state. We present a learned framework for editing physiological signals in videos while preserving visual fidelity. Our method first encodes an input video into a latent representation using a pretrained 3D Variational Autoencoder, and embeds a target heart-rate prompt through a frozen text encoder.