Earticle

For the 3D game visual effects, the deferred rendering can only consider only the scene image resolution regardless of the scene complexity. Therefore, it is effective in processing realistic visual effects using many geometric buffers as multiple render targets. This paper uses the deferred rendering for the 3D game visual effects such as dynamic lights, specular, shadow, motion blur, and water shading. The 3D game supporting deferred rendering is developed to evaluate various 3D rendering effects with variation of the screen resolution in terms of the rendering speed and PSNR image quality. The performance results show that the rendering speed of the 3D visual effect techniques with 1280x960 resolution is average 56.0% better than that with 640x480 resolution. Also, the average PSNR is getting better as the screen resolution gets higher. The PSNR of the 3D visual effect techniques with 1152x864 resolution is average 30.9% better than that with 640x480 resolution.

In 3D games, the deferred rendering is an effective way in processing realistic visual effects using multiple render targets regardless of the scene complexity. In this paper, based on deferred rendering using multiple render targets, the 3D game visual effect techniques such as dynamic lights, specular, shadow, motion blur, and water shading are compared and analysed. The 3D game supporting deferred rendering is developed to evaluate various 3D rendering effects with variation of the size of the render target memory in terms of the rendering speed. The performance results show that the rendering speed of the 3D visual effect techniques with 4bytes render target memory is average 1.4 and 1.9 times better than those of 8bytes and 16bytes memories, respectively. Also, the shadow mapping with 2-pass plays the biggest role on the performance. Other techniques with 1-pass cause a negligible speed degradation.

In three dimensional (3D) computer games, the deferred rendering is an effective way to process realistic visual effects such as dynamic lights, shadow, depth of field (DOF), and high dynamic range (HDR) using multiple geometric buffers(G-buffers) regardless of the scene complexity. However, the G-buffer only stores information about a single pixel in each texel, so transparency with alpha blending on deferred rendering is difficult compared to forward rendering. The conventional way to process transparency on deferred rendering is to separate opaque and transparent renderings. But the object sorting required in it causes the speed degradation. Nevertheless the transparency techniques without object sorting such as screen door, interfaced, and stochastic can reduce the rendering time, but those cause the image quality degradation. This paper compares transparency techniques on deferred rendering. The Game Institute 3D game engine is modified to measure the rendering speeds of transparency techniques. The deferred rendering can speed up the rendering time of about 1.48 times than the forward rendering. The transparency techniques without object sorting can also speed up the rendering time of about 1.1 times than the conventional technique. Plus, six 4K ultra high definition (UHD) game images are used to measure the extent of quality degradation of the transparency techniques without object sorting in terms of peak noise signal ratio (PSNR). Consequently, the interlaced technique can get better PSNRs of about 124.8% in the same than both the screen door and the stochastic techniques.