A virtual reality experimental apparatus for studying visual fidelity in scene perception
Su, Chengyi (2025-05-12)
C. Su
12.05.2025
© 2025, Chengyi Su. Tämä Kohde on tekijänoikeuden ja/tai lähioikeuksien suojaama. Voit käyttää Kohdetta käyttöösi sovellettavan tekijänoikeutta ja lähioikeuksia koskevan lainsäädännön sallimilla tavoilla. Muunlaista käyttöä varten tarvitset oikeudenhaltijoiden luvan.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:oulu-202505123251
https://urn.fi/URN:NBN:fi:oulu-202505123251
Tiivistelmä
The thesis outlines the design, implementation, and evaluation of a virtual reality application for investigating visual fidelity's effect on scene perception. The project's overall goal is to create an experimentally sound and technically solid framework for future investigations.
The application was created with Unreal Engine 5 and utilizes two contrasting visual states: high-fidelity with high-end lighting, complete shading models, and rich material details, and low-fidelity with intentionally degraded methods of visual compromise such as color banding, unlit rendering, and shallow shadows. These contrasts were designed to simulate differences users may experience due to hardware limitations or stylistic rendering in VR systems, while also ensuring clear perceptual differentiation suitable for controlled experimental conditions.
The virtual space contains two large areas, an urban area and an outdoor forest, each with 100 randomly accessible target locations. Participants are transported between the sites, with each site rendered under an independent sun direction to create variability and prevent repeated lighting cues. A neutral "black screen" with a fixation cross is shown prior to each trial to minimize visual carryover. Implementation of all scenes was done with blueprint scripting with addition of experimental timing control and possible EEG synchrony with Lab Streaming Layer (LSL).
The experiment was evaluated by an experienced neuroscientist, who assured that the visual fidelity manipulation is accurately achieved and the inter-trial timing is compatible with EEG study protocols. The evaluator further provided feedback, suggesting adjustments like use of a fixed menu interface and incorporated rest pauses among trial sets. These are the directions in which future versions can be improved.
In summary, this project succeeds in reaching the technical objectives and offers an adaptable, high-accuracy basis for examining the perceptual and cognitive impacts of visual realism in VR via EEG and behavioral evaluations.
The application was created with Unreal Engine 5 and utilizes two contrasting visual states: high-fidelity with high-end lighting, complete shading models, and rich material details, and low-fidelity with intentionally degraded methods of visual compromise such as color banding, unlit rendering, and shallow shadows. These contrasts were designed to simulate differences users may experience due to hardware limitations or stylistic rendering in VR systems, while also ensuring clear perceptual differentiation suitable for controlled experimental conditions.
The virtual space contains two large areas, an urban area and an outdoor forest, each with 100 randomly accessible target locations. Participants are transported between the sites, with each site rendered under an independent sun direction to create variability and prevent repeated lighting cues. A neutral "black screen" with a fixation cross is shown prior to each trial to minimize visual carryover. Implementation of all scenes was done with blueprint scripting with addition of experimental timing control and possible EEG synchrony with Lab Streaming Layer (LSL).
The experiment was evaluated by an experienced neuroscientist, who assured that the visual fidelity manipulation is accurately achieved and the inter-trial timing is compatible with EEG study protocols. The evaluator further provided feedback, suggesting adjustments like use of a fixed menu interface and incorporated rest pauses among trial sets. These are the directions in which future versions can be improved.
In summary, this project succeeds in reaching the technical objectives and offers an adaptable, high-accuracy basis for examining the perceptual and cognitive impacts of visual realism in VR via EEG and behavioral evaluations.
Kokoelmat
- Avoin saatavuus [38618]