Light field display: Difference between revisions - VR & AR Wiki - Virtual Reality & Augmented Reality Wiki

Line 117:

Creating content compatible with LFDs requires capturing or generating directional view information:

* '''[[Light Field Camera|Light Field Cameras]] / [[Plenoptic Camera|Plenoptic Cameras]]:''' Capture both intensity and direction of incoming light using specialized sensors (often with MLAs).<ref name="WetzsteinPlenoptic"/> The captured data can be processed for LFD playback.

* '''[[Computer Graphics]] Rendering:''' Standard 3D scenes built in engines like [[Unity (game engine)|Unity]] or [[Unreal Engine]] can be rendered from multiple viewpoints to generate the necessary data.<ref name="ForbesLightField~~"/><ref name="LookingGlassSoftware~~"/> Specialized light field rendering techniques, potentially using [[Ray tracing (graphics)|ray tracing]] or neural methods like [[Neural Radiance Fields]] (NeRF), are employed.<ref name="ForbesLightField"/><ref name="Mildenhall2020NeRF">Mildenhall, B., Srinivasan, P. P., Tancik, M., Barron, J. T., Ramamoorthi, R., & Ng, R. (2020). NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis. European Conference on Computer Vision (ECCV), 405-421. doi:10.1007/978-3-030-58452-8_24</ref>

* '''[[Computer Graphics]] Rendering:''' Standard 3D scenes built in engines like [[Unity (game engine)|Unity]] or [[Unreal Engine]] can be rendered from multiple viewpoints to generate the necessary data.<ref name="ForbesLightField"/> Specialized light field rendering techniques, potentially using [[Ray tracing (graphics)|ray tracing]] or neural methods like [[Neural Radiance Fields]] (NeRF), are employed.<ref name="ForbesLightField"/><ref name="Mildenhall2020NeRF">Mildenhall, B., Srinivasan, P. P., Tancik, M., Barron, J. T., Ramamoorthi, R., & Ng, R. (2020). NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis. European Conference on Computer Vision (ECCV), 405-421. doi:10.1007/978-3-030-58452-8_24</ref>

* '''[[Photogrammetry]] and 3D Scanning:''' Real-world objects/scenes captured as 3D models can serve as input for rendering light field views.

* '''[[Focal Stack]] Conversion:''' Research explores converting image stacks captured at different focal depths into light field representations, particularly for multi-layer displays.<ref name="WetzsteinTensor"/>

Line 144:

* '''Limited Field of View (FoV):''' Achieving wide FoV comparable to traditional VR headsets while maintaining high angular resolution is challenging.<ref name="Lanman2020NearEyeCourse"/>

* '''Brightness and Efficiency:''' Techniques like MLAs and parallax barriers inherently block or redirect light, reducing overall display brightness and power efficiency.

* '''Content Ecosystem:''' The workflow for creating, distributing, and viewing native light field content is still developing compared to standard 2D or stereoscopic 3D.~~<ref name="LookingGlassSoftware"/>~~

* '''Content Ecosystem:''' The workflow for creating, distributing, and viewing native light field content is still developing compared to standard 2D or stereoscopic 3D, due to no consumer lightfield hardware.

* '''Visual Artifacts:''' Potential issues include [[Moiré pattern|moiré]] effects (from periodic structures like MLAs), ghosting/crosstalk between views, and latency.