De-rendering and Re-rendering the World скачать в хорошем качестве

De-rendering and Re-rendering the World

Chris Careaga, final-year PhD, Computational Photography Lab, Simon Fraser University Decomposing real-world images into renderable graphics representations, a process known as inverse rendering, is a long-standing challenge in computer vision. Solving this problem enables physically-based image manipulations, such as material editing and relighting, granting artists and creators unprecedented control over visual appearance. Despite the problem's inherent difficulty, the field has seen remarkable progress in recent times: in just five years, we've advanced from oversimplified light transport models to physically accurate, in-the-wild scene relighting. This talk highlights four SIGGRAPH papers that have helped drive this transformation and discusses promising directions for the future. https://yaksoy.github.io/PhysicalReli... Chris Careaga is a final-year PhD student in the Computational Photography Lab at Simon Fraser University, led by Dr. Yağız Aksoy. He received his MSc and BSc in computer science from Western Washington University. His research focuses on modeling and estimating illumination in images for physically-based editing applications. His work has been recognized by the graphics community in the form of a SIGGRAPH Asia Best Paper Honorable Mention Award, and has been featured in various outlets such as Two Minute Papers, PetaPixel, and the SIGGRAPH Technical Papers Trailer. https://ccareaga.github.io/ https://www.meetup.com/sv-siggraph/ev... Rough Chapters 0:00 Chapter Intro 2:22 Speaker Intro 3:10 Presentation 4:00 Intrinsic Image Decomposition vis Ordinal Shading 16:57 Intrinsic Harmonization for Illumination-Aware Compositing 29:00 Colorful Diffuse Intrinsic Image Decomposition in the Wild 43:30 Physically Controllable Relighting of Photographs 58:07 Q&A