Thesis
Evaluating ray tracing, path tracing, and AI denoising in Unreal Engine for cinematic rendered sequences: a combined image analysis and user perception study
Master of Science (M.S.), Drexel University
Jun 2026
DOI:
https://doi.org/10.17918/00011440
Abstract
This thesis evaluates the visual quality and structural similarity of Ray Tracing (RT), Path Tracing (PT), and denoised Path Tracing in Unreal Engine for cinematic rendered sequences. Cinematic rendering relies on accurate lighting, material response, and frame consistency to achieve perceptual realism, yet most existing studies focus on static images rather than animated sequences. This study addresses this gap by examining how different rendering pipelines perform over time and how these differences influence viewer perception. The study generates identical cinematic sequences using RT, high-sample PT, and low-sample PT with denoising. It applies quantitative image analysis using metrics such as Peak Signal-to-Noise Ratio (PSNR) and Learned Perceptual Image Patch Similarity (LPIPS) to measure structural and perceptual differences across frames. In parallel, a user-perception study collects viewer responses regarding realism, clarity, lighting, and visual stability under controlled conditions. This combined approach enables direct comparison between measured image quality and human judgment. The results show that denoised PT closely approximates high-sample PT in low-light conditions, with minimal measurable differences. As scene brightness and lighting complexity increase, differences become more noticeable in both quantitative metrics and user perception. RT produces stable results with faster rendering but exhibits limitations in edge quality and anti-aliasing, with viewers reporting visible shimmering and instability. PT produces higher visual fidelity, with smoother edges and more accurate lighting, but at a significantly higher computational cost. Denoised PT strikes a balance between efficiency and quality but introduces temporal artifacts, such as flicker under certain conditions. The findings establish that temporal stability, lighting conditions, and anti-aliasing behavior strongly influence perceived image quality in rendered sequences. This thesis provides a structured evaluation of rendering pipelines and identifies conditions under which denoised PT serves as a practical alternative to high-sample PT. The results support informed decision-making in cinematic rendering workflows that require trade-offs among quality, stability, and render time. Keywords: Ray Tracing, Path Tracing, Denoising, Unreal Engine, Image Quality, User Perception.
Metrics
1 Record Views
Details
- Title
- Evaluating ray tracing, path tracing, and AI denoising in Unreal Engine for cinematic rendered sequences
- Creators
- Kameswara Naga Sai Siva Dheeraj Mantha
- Contributors
- Nicholas E. Jushchyshyn (Advisor)
- Awarding Institution
- Drexel University
- Degree Awarded
- Master of Science (M.S.)
- Publisher
- Drexel University
- Number of pages
- xv, 83 pages
- Resource Type
- Thesis
- Language
- English
- Academic Unit
- Digital Media; Drexel University; Antoinette Westphal College of Media Arts and Design
- Other Identifier
- 991022189596104721