룩업테이블기반비선형렌즈플레어실시간렌더링방법 (Real-Time Nonlinear Lens-Flare Rendering Method Based on Look-Up Table) 조성훈 정유나 이성길 (Sunghun Jo) (Yuna Jeong) (Sungkil Lee) 요약.,,. /.,,. :,,,, Abstract In computer graphics, high-quality lens flares have been generated using costly offline rendering. A recent matrix-based approximation has enabled generation of high-quality lens flares suitable for real-time applications, but its quality degrades due to the lack of nonlinear patterns of lens flares. This paper introduces a method for high-quality lens-flare rendering, which includes blending of both nonlinear as well as linear patterns. The nonlinear patterns are pre-rendered or photographically captured offline and stored in a look-up table. The online stage reads only the pattern by looking up the table using a light angle, hence making its performance drop negligible while greatly improving the quality. Key words : lens flare, rendering, look-up table, blending, non-linear patterns 1. 서론 이연구는미래창조과학부의재원으로한국연구재단의 < 실감교류 인체감응솔루션 > 글로벌프론티어사업, 중견연구자지원사업, 정보통 신기술진흥센터대학 ICT 연구센터육성지원사업으로수행된연구임. (2012M3A6A3055695, 2015R1A2A2A01003783, IITP-2016-R2718-16-0017) 정회원 : 성균관대학교전기전자컴퓨터공학과 shjo@skku.edu 비회원 : 성균관대학교전기전자컴퓨터공학과 jeongyuna@skku.edu 종신회원 : 성균관대학교소프트웨어학과부교수 sungkil@skku.edu 논문접수 : 2016 년 11 월 25 일심사완료 : 2016 년 12 월 31 일 CopyrightC2017 한국정보과학회 ː 개인목적이나교육목적인경우, 이저 작물의전체또는일부에대한복사본혹은디지털사본의제작을허가합니 다. 이때, 사본은상업적수단으로사용할수없으며첫페이지에본문구와 출처를반드시명시해야합니다. 이외의목적으로복제, 배포, 출판, 전송 등모든유형의사용행위를하는경우에대하여는사전에허가를얻고비용 을지불해야합니다. 정보과학회논문지 : 정보통신제권제호 (2017. 4) /,.,.,..,.
정보과학회논문지 : 정보통신제권제호 (2017.) 1 (a) (b). Fig. 1 Results of physical simulation based lens flare rendering (a) and proposed real-time lens flare rendering based on look-up table (b).,.... (topology) (, ),,. / /. ( ), (, 2, 4, 6 ),.,,.,.,.,, [10].,.,.,.,,..,,..
룩업테이블기반비선형렌즈플레어실시간렌더링방법 2. θ Φ. Fig. 2 Positioning of a virtual light ray. Tilting angle of incoming light ray (θ) and rotation angle of light source revolving around the entrance plane (Φ).,,,,.,.. 2. 관련연구, [1,2].. [3], [4], (area light) [5],. (path tracing) [6] (photon mapping) [7].,. (sparse) (bundle) [8]. (polynomial representation) [9]. (entrance pup- 3. Fig. 3 Example of ghost generated by two reflections within an optical system. il),., [10].,.,... 3. 알고리즘.,.. 3.1 비선형렌즈플레어기록을위한광원파라미터화,,,. 2, 2 θ Φ. θ
정보과학회논문지 : 정보통신제권제호 (2017.) 4 θ. Fig. 4 Example of look-up table for nonlinear lens flare ghost using the parameter θ., Φ (projection). θ Φ 2,. /. θ.,., Φ, Φ (, ). 3.2 렌즈플레어패턴의나열,.., (interface). (ghost).. 3. 3.3 비선형렌즈플레어의기록. (, 5. Fig. 5 Nonlinear lens flare according to the angle. 6. Fig. 6 Example of a cell in a look-up table that synthesizes ghosts. )., (look-up table). θ, 4.. 2D, 5..,,. (cell) 6.. 1D.
,. 룩업테이블기반비선형렌즈플레어실시간렌더링방법 3.4 비선형렌즈플레어의재생. ( ; [10]),.,.., θ ( ) Φ. θ, Φ... 8. (additive blending), / (linear/nonlinear interpolation).,. 4. 결과,.,,,. 7, 1. Canon 3.99ms, 0.01ms. Nikon Angenieux 그림 7 행렬기반근사방식과제안한룩업테이블기반 방식그리고물리기반렌더링방법의성능비교와룩업테 이블기반방식에서의오프라인비선형패턴생성시간. Fig. 7 Performance of matrix-based approximation, proposed look-up table based approach, nonlinear pattern generation time in offline stage, and physics-based rendering method. 1,. Table 1. Performance of matrix-based approximation method, proposed look-up table based method, and physical-based method. Canon Nikon Angenieux Linear Only (Lee, 2013) 3.99 ms 1.50 ms 1.25 ms preprocessing 14774 ms 26.67 ms 50.03 ms Ours linear + nonlinear 4.00 ms 1.509 ms 1.301 ms nonlinear 0.01 ms 0.009 ms 0.051 ms Reference (Hullin, 2011) 140.0 ms 69.43 ms 14.50 ms 1.50ms, 1.25ms 0.009ms, 0.051ms., [8] 46. 8, 9..,, PSNR (peak signal-to-noise ratio) SSIM (structural similarity) [11]...
정보과학회논문지 : 정보통신제권제호 (2017.) 8 (Canon 70 200mm, Nikon 80 200mm, Angenieux 100mm). Fig. 8 Lens-flare rendering results of various camera models (Canon 70 200mm, Nikon 80 200mm, Angenieux 100mm).. (, ),.. 512 512, 128.96 MB, 2 20, 2.5 GB. 5. 토론 /..,.
룩업테이블기반비선형렌즈플레어실시간렌더링방법 9 (90, 120, 150 ) (a) (b), (c) (d). Fig. 9 Deformation of linear patterns (a) and nonlinear patterns (b) according to camera angles (90, 120, and 150 ), and comparison of blending results (c) with physically-based results (d)..,., 9,,.. 참고문헌 [ 1 ] Pixar, The imperfect lens : Creating the look of Wall-E. Wall-E Three-DVD-Box. 2008. [ 2 ] Wenzel C, (2005) Far Cry and DirectX [Online]. Available : http://developer.amd.com/media/gpu_asse sets/d3dtutorial08_farcrynanddx9.pdf [ 3 ] Kilgard J, (2000) Fast OpenGL-rendering of lens flares.[online]. Available : http://www.opengl.org/res ources/features/kilgardtechniques/lensflare/ [ 4 ] King Y, Game Programming Gems 2. "Charles River Media ch. 2D Lens Flare", 2001 [ 5 ] Maughan C, Game Programming Gems 2. "Charles River Media ch. Texture Masking for Faster Lens Flare", 2001 [ 6 ] Chaumond. J, (2007), Realistic camera lens flares. [Online]. Available : http://graphics.stanford.ed u/wi kis/cs348b-07/julienchaumond/finalproject [ 7 ] Keshmirian, "A physically based approach for lens flare simulation" Master s thesis, University of California, San Diego, 2008 [ 8 ] Hullin M, Eisemann E, Eisemann, Hans-Peter Seidel and Lee S, "Physically-Based Real-Time
Lens Flare Rendering" ACM Trans. Graphics (Proc. SIGGRAPH 11), Vol. 30, No. 4, pp. 108:1-9, Aug. 2011. [ 9 ] Hullin M, Matthias B, Hankia J, and Heidrich, W, "Polynomial Optics: A Construction Kit for Efficient Ray-Tracing of Lens Systems" Computer Graphics Forum,(Proc. EGSR`12) Vol, 31, No. 4 pp. 1375-1383 June. 2012. [ 10 ] Lee S and Eisemann E, "Practical Real-Time Lens-Flare Rendering" Computer Graphics Forum (Proc. EGSR`13), Vol. 32, No. 4, pp. 1-6, July. 2013. [ 11 ] Wang, et al "Image quality assessment: from error visibility to structural similarity" IEEE transactions on image processing 13.4 600-612, 2004 정보과학회논문지 : 정보통신제권제호 (2017.) 조성훈 2016 년한세대학교전자소프트웨어졸업 ( 학사 ). 2016 년 ~ 현재성균관대학교전자전기컴퓨터공학과석사과정. 관심분야는 Real-time rendering, high-performance GPU computing 정유나 2011 년한국산업기술대컴퓨터공학부졸업 ( 학사 ). 2012 년 ~ 현재성균관대학교전자전기컴퓨터공학과석박통합과정. 관심분야는 Real-time rendering, virtual reality. 이성길 2002 년포항공과대학교신소재공학과졸업 ( 학사 ). 2009 년포항공과대학교컴퓨터공학과졸업 ( 박사 ). 2009 년 ~2011 년독일 Max-Planck-Institut Informatik 박사후연구원. 2011 년 ~2015 년성균관대학교소프트웨어학과조교수. 2015 년 ~ 현재성균관대학교소프트웨어학과부교수. 관심분야는 Real-time rendering, perception-based information visualization, virtual reality, high-performance GPU computing