1 / 25

Advanced Depth of Field Rendering Using Pyramidal Image Processing Techniques

This presentation explores innovative approaches to depth of field rendering by employing pyramidal image processing methods. Depth of field is crucial in photography and cinematography, providing a realistic focus effect that differs across various optical systems. We discuss related work, present our proposed method that focuses on real-time application without compromising image quality, and share experimental results comparing our technique to state-of-the-art methods. Future work aims to refine these techniques for enhanced performance in interactive environments.

george
Télécharger la présentation

Advanced Depth of Field Rendering Using Pyramidal Image Processing Techniques

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. 07 E U R O G R A P H I C S Depth-of-Field Renderingby Pyramidal Image Processing Martin Kraus (TU München) and Magnus Strengert (Universität Stuttgart)

  2. 00 Outline ofthis Talk 01Introduction 02Relatedwork 03Proposedmethod 04 Experiments 05 Future work

  3. 01 I N T R O D U C T I O N What Is Depthof Field? • The depth in front andbeyondthefocus plane whereobjectsappeartobe in focus • A propertyof all realopticalsystems • onlyvirtualpin-hole camerashave infinite depthoffield • Usedforimportantphotographicandcinematographictechniques

  4. photoby Jon Sullivan (http://pdphoto.org/PictureDetail.php?mat=pdef&pg=8202)

  5. photoby "che" (http://commons.wikimedia.org/wiki/Image:Dandelion_clock_detail.jpg)

  6. 02 R E L A T E D W O R K ClassificationofTechniques: • Splatting [Potmesil & Chakravarty 1982] • Stochasticsampling [Cook et al. 1984] • e.g., in distributedraytracingor in REYES • stateoftheart in offline rendering • Pre-filtering [Rokita 1993] • lots ofartifacts [Demers 2004] • stateoftheart in real-time rendering • Blurringof sub-images [Barsky 2004]

  7. 02 R E L A T E D W O R K High Quality in Real Time? • Real-time performancerequires: • independenceofscenecomplexity • excludesstochasticsampling • thus: image post-processingof a pin-hole colorimageanddepthmap • independenceofimagesynthesisforfree • convolutionfilteringistoo expensive • evenwith FFTs on GPUs • thus: blurringusingpyramidalgorithms

  8. 02 R E L A T E D W O R K High Quality in Real Time? • High imagequalityrequires: • smooth blurringwithoutinterpolationartifacts • pyramidblurring [Kraus & Strengert 2007] • noincorrectcolorbleeding • excludespre-filtering • separate blurringof sub-images • disocclusionof semi-transparent pixels • inpaintingofcolorsanddepthswithpyramidalgorithm [Strengert et al. 2006]

  9. 03 P R O P O S E D M E T H O D Overview pin-hole image & depthmap decompositioninto sub-images 3 1 4 disocclusion matting blurring 2 blendingof sub-images 5 resultingimage

  10. 03 P R O P O S E D M E T H O D Decomposition • decomposeinto sub-imagesandcullforegroundpixelsaccordingtodepthmap 1

  11. 03 P R O P O S E D M E T H O D Disocclusion • foreach sub-image: disoccludeculledforeground (using pyramidal inpainting) 2

  12. 03 P R O P O S E D M E T H O D Matting • foreach sub-image: computealpha-mattingaccordingtoeachpixel'sdepth 3

  13. 03 P R O P O S E D M E T H O D Blurring • foreach sub-image: blurcolorandalpha (using pyramidal blurring) 4

  14. 03 P R O P O S E D M E T H O D Blending • back-to-front blendingof all sub-images • resultcomputed in 70.4 ms • (12 sub-images,hardware: • NVIDIA GeForce 7900 GTX) 5

  15. 03 P R O P O S E D M E T H O D Main Features • independentofscenecomplexity • andindependentofimagesynthesis • interactiveperfomance on GPUs • real-time forsmallcirclesofconfusion • highimagequality • avoidsartifactsofpre-filteringtechniques • let's do someexperiments …

  16. 04 E X P E R I M E N T S OurMethod vs. pbrt • pbrtusesstochasticsampling • Whichiswhich?

  17. 04 E X P E R I M E N T S OurMethod vs. pbrt pbrt our method

  18. 04 E X P E R I M E N T S OurMethod vs. pbrt • Whichiswhich?

  19. 04 E X P E R I M E N T S OurMethod vs. pbrt pbrt our method

  20. 04 E X P E R I M E N T S Can We Break OurMethod? • Yes, with a very largelensradius. pbrt bleeding gray too opaque our method

  21. 04 E X P E R I M E N T S Can We Break OurMethod? • Video withvery large lensradii:eg07.mov • also availableat: http://wwwcg.in.tum.de/Research/Publications/DepthOfField

  22. 05 F U T U R E W O R K Are Wethereyet? • No, but weavoidtypicalrenderingartifactsof real-time techniques • Specializedvariantsforbetterperformanceandimagequality • Alternative blurfilters • High-potential application: • gaze-directedfocus

  23. 06 A C K N O W L E D G M E N T S • The photosappearcourtesyof "che" and Jon Sullivan. • The dragon model appearscourtesyofthe Stanford University Scanning Repository. • The pbrtsceneappearscourtesyof Gregory Humphreys and Matt Pharr.

  24. 07 E U R O G R A P H I C S Thankyou! Andhave a safetriphome! Questions?

  25. 07 E U R O G R A P H I C S

More Related