Animating emotions in ECA’s for interactive applications
Author
Zilmer, Lars
Term
4. term
Education
Publication year
2012
Submitted on
2012-05-23
Abstract
Denne afhandling undersøger, hvordan følelser kan udtrykkes gennem kropssprog i animationer af en embodied conversational agent (ECA) til interaktive applikationer på mobile enheder, hvor begrænsninger i blandt andet polygonantal og knogler i skelettet udfordrer udtryksmulighederne. Med udgangspunkt i en visuel tilgang (uden lyd og tale) gennemgås ECA-begrebet, relevante niveauer af detaljer for mobile platforme samt praksis fra spil, herunder principper for spilanimation og inspirationskilder som The Sims. På baggrund af forskning i nonverbal kommunikation vælges tre følelser med høj genkendelighed—vrede, tristhed og glæde—og der afledes konkrete kropslige indikatorer (fx kropsholdning, tempo, armføring) til brug i animationsdesignet. Der designes og implementeres low‑poly animationer med fokus på loopbarhed og glatte overgange, hvorefter en test med deltagere vurderer, hvordan ECA’ens følelsestilstand opfattes. Resultaterne peger på, at følelser kan kommunikeres effektivt gennem kropssprog i en ECA på mobile platforme trods tekniske begrænsninger, og studiet tilbyder en praktisk ramme for at vælge og animere følelsesudtryk under sådanne vilkår.
This thesis investigates how emotions can be conveyed through body language in the animation of an embodied conversational agent (ECA) for interactive applications on mobile devices, where limits on polygon count and skeletal complexity constrain expressiveness. Focusing on visual cues only, the work reviews ECAs, level‑of‑detail considerations for mobile, and common game animation practices, including core animation principles and references such as The Sims. Drawing on research in nonverbal communication, three highly recognizable emotions—anger, sadness, and happiness—are selected, and concrete body cues (e.g., posture, tempo, arm movement) are derived to guide animation design. Low‑poly, low‑bone animations are then designed and implemented with attention to seamless loops and smooth transitions, followed by a user perception test in which participants assess the agent’s emotional state. The results indicate that emotions can be effectively communicated through body language in a mobile ECA despite platform constraints, and the study outlines practical guidance for selecting and animating emotional expressions under such limitations.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents