Computer-Generated Holography (CGH) as a Generic Display...
Transcript of Computer-Generated Holography (CGH) as a Generic Display...
Computer-GeneratedComputer-GeneratedHolography (CGH) as aHolography (CGH) as a
Generic Display TechnologyGeneric Display Technology
Niall Emmerson
Leo Walsh
Jon Etxezarreta
AdvantagesAdvantages
High Optical EfficiencyHigh Optical Efficiency
Ease of tilingEase of tiling holographic displays can be built from multiple holographic displays can be built from multiple SLM'sSLM's without loss of clarity without loss of clarity
Tolerance of Pixel DefectsTolerance of Pixel Defects
Wide colour gamutWide colour gamut use of lasers mean a wider range of colours perceivable by the human eyeuse of lasers mean a wider range of colours perceivable by the human eye
Advantages cont.Advantages cont.
Full depth cuesFull depth cues
High system volume and Image resolutionHigh system volume and Image resolution
Artefact-free binary modulationArtefact-free binary modulation
Aberration, distortion and conformal correctionAberration, distortion and conformal correction
ChallengesChallengesImage width and field of viewImage width and field of view
Image resolutionImage resolution -NCGH = 2NI-NCGH = 2NI
Image qualityImage quality -C -C ∝∝ nxlog[Q]nylog[Q]log(1+S/N) nxlog[Q]nylog[Q]log(1+S/N)
Laser speckleLaser speckle
Algorithms for CGH designAlgorithms for CGH design
Ping-pong algorithmPing-pong algorithm
-- Introduces an obscuration operator, shows hidden-line-removal effects on 3D imagesIntroduces an obscuration operator, shows hidden-line-removal effects on 3D images
-- Easy to codeEasy to code
-- Can only generate images of self-luminous objectsCan only generate images of self-luminous objects
-- Computationally inefficientComputationally inefficient
Interference-based algorithmInterference-based algorithm
-- Implement a 3D scalar diffraction integral, generates high quality imagesImplement a 3D scalar diffraction integral, generates high quality images
-- Can reproduce all human visual depth.Can reproduce all human visual depth.
-- Currently are the benchmark for image qualityCurrently are the benchmark for image quality
-- Problem: high computational loadsProblem: high computational loads
-- CGH require huge amount of pixels CGH require huge amount of pixels crucial the choice of the algorithm crucial the choice of the algorithmUsed to calculate the holographic fringe patternsUsed to calculate the holographic fringe patterns
-- Allow tradeoffs in image quality with computational speed.Allow tradeoffs in image quality with computational speed.
-- They count the hologram spatially (They count the hologram spatially (hogelshogels) and spectrally () and spectrally (hogelhogel vectors), vectors),can vary the computational loads.can vary the computational loads.
-- The algorithms can pre-compute much of the CGH calculation.The algorithms can pre-compute much of the CGH calculation.
-- CL= CL= hogelshogels x image points x (diffraction-table lookup entries) x rows x image points x (diffraction-table lookup entries) x rows
-- High definition TV= 100 x 1024 x 4096 x 1024 = 424 High definition TV= 100 x 1024 x 4096 x 1024 = 424 gigagiga MACsMACs
-- Optimisations can significantly REDUCE the computational loadOptimisations can significantly REDUCE the computational load
Diffraction-specific algorithmsDiffraction-specific algorithms
Computational loadsComputational loads
-- Large size computational loads Large size computational loads appropriate architecture appropriate architecture
-- Candidate architectures include supercomputersCandidate architectures include supercomputers
-- Intel IA-32 systems offer the most cost-effective and high-performance clusterIntel IA-32 systems offer the most cost-effective and high-performance cluster
-- Pixel prioritisationPixel prioritisation
-- Dynamic level of detailDynamic level of detail
-- Wireframe mode forWireframe mode forinteractioninteraction
-increasing the frame rate-increasing the frame rate
-enabling computation at same frame-enabling computation at same frame
rate with lower quality (reducing cost)rate with lower quality (reducing cost)
-- Users should be able to run their applications without modification.Users should be able to run their applications without modification.
Hardware architecturesHardware architectures
OptimizationsOptimizations
TO ACCELERATE CGH COMPUTATION
TO BE COMMERCIALLY SUCCESFUL:
By
Optical Hardware ConsiderationsOptical Hardware Considerations
SLM TechnologiesSLM Technologies−− HolovideoHolovideo
36 million pixels36 million pixels 10 thousand million pixels required for10 thousand million pixels required for
holographic workstationholographic workstation
−− QinetiQ Active TilingQinetiQ Active Tiling Uses both EASLM and OASLMUses both EASLM and OASLM EASLM output tiled onto OASLMEASLM output tiled onto OASLM Shutters used to build up picture on writeShutters used to build up picture on write
side of OASLM. Shutters side of OASLM. Shutters synchronisedsynchronised with withEASLM.EASLM.
Optical Hardware ConsiderationsOptical Hardware Considerations
Active Tiling is modular Active Tiling is modular –– can be stacked to can be stacked toallow for required number of pixelsallow for required number of pixels
Suitable for cluster computingSuitable for cluster computing Pixel areal density of more than 2.2×10Pixel areal density of more than 2.2×1066
pixels cmpixels cm-2-2
−− Pixel count in excess of to 1Pixel count in excess of to 1××101099
Experimental ResultsExperimental Results
This example shows 3This example shows 3××8 thousand million8 thousand millionpixels pixels –– full-colour, full-parallax full-colour, full-parallax
Standard computer graphics techniques canStandard computer graphics techniques canbe appliedbe applied
Large number of pixels requiredLarge number of pixels required Additional computationAdditional computation
−− High costHigh cost
ConclusionsConclusions
Paper was confusing with frequent use ofPaper was confusing with frequent use oftechnical terms specific to the field oftechnical terms specific to the field ofholography.holography.
Due to high cost Due to high cost –– both computational and both computational andmonetary monetary –– makes CGH unsuitable for makes CGH unsuitable forvirtual environments at the present time.virtual environments at the present time.