The rapid evolution of immersive media has brought increasing attention to six degrees of freedom (6DoF) virtual reality (VR) content, which allows users to experience translational and rotational movement within synthetic or captured environments. Unlike conventional two-dimensional or 360° (3DoF) video, 6DoF content poses unique challenges for real-time streaming, including high bandwidth requirements, stringent latency constraints, and the need for advanced compression to ensure perceptual quality. While existing research has primarily investigated codecs and quality assessment methods in 2D and 3DoF contexts, systematic evaluation tailored to 6DoF remains underexplored. This thesis addresses this gap in collaboration with V-Nova Ltd., a key contributor to MPEG-5 Part 2 Low Complexity Enhancement Video Coding (LCEVC). It proposes and implements a subjective evaluation framework for assessing visual quality in 6DoF VR scenarios, with emphasis on the integration and performance of LCEVC. A dedicated test pipeline was developed to deliver immersive video clips in head-mounted displays (HMDs), enabling controlled studies that collected both quantitative scores (MOS) and qualitative feedback from expert and non-expert participants. The results indicate that LCEVC provides measurable improvements in perceptual quality at reduced bitrates compared to baseline encodings, though limitations persist in handling motion smoothness, artifact visibility, and content heterogeneity. The study also highlights methodological challenges in VR-specific subjective testing, including environmental control, participant pool diversity, and the lack of standardized protocols for volumetric or omnidirectional content. Overall, this thesis—developed in collaboration with V-Nova—contributes a reproducible framework for evaluating immersive video quality in 6DoF VR, offering insights into the role of LCEVC in next-generation streaming pipelines and identifying directions for future research, including refined test environments, wireless setups, and closer integration of subjective and objective assessments.
La rapida evoluzione dei media immersivi ha posto crescente attenzione ai contenuti di realtà virtuale (VR) con sei gradi di libertà (6DoF), che consentono all’utente di sperimentare movimenti sia traslazionali che rotazionali all’interno di ambienti sintetici o acquisiti. A differenza del video bidimensionale o a 360° (3DoF), i contenuti 6DoF presentano sfide uniche per lo streaming in tempo reale, tra cui elevati requisiti di banda, vincoli stringenti di latenza e la necessità di tecniche di compressione avanzate per garantire la qualità percettiva. Sebbene la ricerca esistente abbia principalmente analizzato codec e metodi di valutazione della qualità in contesti 2D e 3DoF, una valutazione sistematica specifica per il 6DoF rimane ancora poco esplorata. Questa tesi affronta tale gap in collaborazione con V-Nova Ltd., attore chiave nello sviluppo dello standard MPEG-5 Part 2 Low Complexity Enhancement Video Coding (LCEVC). Il lavoro propone e implementa un framework di valutazione soggettiva per la qualità visiva in scenari VR 6DoF, con particolare attenzione all’integrazione e alle prestazioni di LCEVC. È stata sviluppata una pipeline di test dedicata per la distribuzione di clip immersive su visori VR (HMD), che ha permesso studi controllati basati sia su punteggi quantitativi (Mean Opinion Scores, MOS) sia su feedback qualitativo di partecipanti esperti e non esperti. I risultati mostrano che LCEVC offre miglioramenti significativi della qualità percettiva a bitrate ridotti rispetto alle codifiche di base, pur presentando limitazioni nella gestione della fluidità dei movimenti, della visibilità degli artefatti e dell’eterogeneità dei contenuti. Lo studio evidenzia inoltre le sfide metodologiche legate ai test soggettivi specifici per il VR, tra cui il controllo dell’ambiente, la diversità del campione di partecipanti e l’assenza di protocolli standardizzati per contenuti volumetrici. Complessivamente, questa tesi—sviluppata in collaborazione con V-Nova—contribuisce a un framework riproducibile per la valutazione della qualità video immersiva in scenari 6DoF VR, fornendo indicazioni sul ruolo di LCEVC nei futuri sistemi di streaming e identificando direzioni di ricerca futura, tra cui ambienti di test più accurati, configurazioni wireless e una più stretta integrazione tra valutazioni soggettive e metriche oggettive.
Subjective video quality assessment for 6DoF immersive VR experiences
Spezzapria, Alessandro
2024/2025
Abstract
The rapid evolution of immersive media has brought increasing attention to six degrees of freedom (6DoF) virtual reality (VR) content, which allows users to experience translational and rotational movement within synthetic or captured environments. Unlike conventional two-dimensional or 360° (3DoF) video, 6DoF content poses unique challenges for real-time streaming, including high bandwidth requirements, stringent latency constraints, and the need for advanced compression to ensure perceptual quality. While existing research has primarily investigated codecs and quality assessment methods in 2D and 3DoF contexts, systematic evaluation tailored to 6DoF remains underexplored. This thesis addresses this gap in collaboration with V-Nova Ltd., a key contributor to MPEG-5 Part 2 Low Complexity Enhancement Video Coding (LCEVC). It proposes and implements a subjective evaluation framework for assessing visual quality in 6DoF VR scenarios, with emphasis on the integration and performance of LCEVC. A dedicated test pipeline was developed to deliver immersive video clips in head-mounted displays (HMDs), enabling controlled studies that collected both quantitative scores (MOS) and qualitative feedback from expert and non-expert participants. The results indicate that LCEVC provides measurable improvements in perceptual quality at reduced bitrates compared to baseline encodings, though limitations persist in handling motion smoothness, artifact visibility, and content heterogeneity. The study also highlights methodological challenges in VR-specific subjective testing, including environmental control, participant pool diversity, and the lack of standardized protocols for volumetric or omnidirectional content. Overall, this thesis—developed in collaboration with V-Nova—contributes a reproducible framework for evaluating immersive video quality in 6DoF VR, offering insights into the role of LCEVC in next-generation streaming pipelines and identifying directions for future research, including refined test environments, wireless setups, and closer integration of subjective and objective assessments.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_10_Spezzapria_02.pdf
non accessibile
Descrizione: Executive Summary
Dimensione
673.23 kB
Formato
Adobe PDF
|
673.23 kB | Adobe PDF | Visualizza/Apri |
|
2025_10_Spezzapria_01.pdf
non accessibile
Descrizione: MSc Thesis
Dimensione
3.35 MB
Formato
Adobe PDF
|
3.35 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/243747