日本語フィールド
著者：Yoshiyuki Yamashita題名：Implementing a Rasterization Framework for a Black Hole Space-Time発表情報：Journal of Information Processing 巻: 24 号: 4 ページ: 10キーワード：Rasterization, Black hole, Graphics processing unit概要：曲がった4次元時空内で高速CG描画を達成するために、計算メッシュを用いた光線の湾曲の事前計算と線形補間、およびテッセレーションを組み合わせた描画プログラムをOpenGL/GLSLを用いて完成させ、様々な被写体例についてデスクトップPCおよびノートPCのGPUで実際に描画速度を計測し、約100万ポリゴンの性能を達成した。また線形補間によるCG画像の誤差を1画素以下に抑えることに成功した。抄録：The theory of general relativity predicts that the strong gravity of a black hole bends the trajectories of light rays. Calculating their bendings numerically, we can obtain a 3D CG image when the view point is set in the black hole spacetime. The existing researches adopt the ray tracing method for rendering while we adopt the rasterization method in this paper.
In order to achieve fast perspective projection in the curved spacetime, we calculate more than thirty million light trajectories on an optimally constructed computational mesh in advance and let a GPU interpolate them when rendering.
Furthermore, in order to render the lines and triangular polygons of CG objects accurately, we apply the dynamic subdividing technique (tessellation). Various types of CG programs can be easily written in the same way as in the conventional 3D CG programming with a common graphics API. Utilizing the recent computing power of the GPU, the rendering performance of nearly one million polygons per second is achieved even on a notebook PC.英語フィールド
Author：Yoshiyuki YamashitaTitle：Implementing a Rasterization Framework for a Black Hole Space-TimeAnnouncement information:Journal of Information Processing Vol: 24 Issue: 4 Page: 10Keyword：Rasterization, Black hole, Graphics processing unitAn abstract：The theory of general relativity predicts that the strong gravity of a black hole bends the trajectories of light rays. Calculating their bendings numerically, we can obtain a 3D CG image when the view point is set in the black hole spacetime. The existing researches adopt the ray tracing method for rendering while we adopt the rasterization method in this paper.
In order to achieve fast perspective projection in the curved spacetime, we calculate more than thirty million light trajectories on an optimally constructed computational mesh in advance and let a GPU interpolate them when rendering.
Furthermore, in order to render the lines and triangular polygons of CG objects accurately, we apply the dynamic subdividing technique (tessellation). Various types of CG programs can be easily written in the same way as in the conventional 3D CG programming with a common graphics API. Utilizing the recent computing power of the GPU, the rendering performance of nearly one million polygons per second is achieved even on a notebook PC.