- 博客(6)
- 收藏
- 关注
RSS订阅原创 CLRS 2nd Round - reading notes 1
<br />I TOTALLY skipped Maths part of CLRS in this 2nd round. You know what, I can finish nearly 100 pages per day, for the first 200 pages of this book. (Practice Problem not included of course... I guess I will tackle them in the 3rd round)<br /> <br />I
2010-09-25 13:33:00
505
原创 One critical issue in Instant Radiosity
<br />I just realized: for each individual pass of Instant Radiosity, only one bounce of diffuse illumination can be evaluated. (At least true in my current implementation.) So to have a relative complete GI effect, many passes are necessary.<br /> <br />N
2010-09-20 10:27:00
784
原创 My renderer update – Instant Radiosity moved on GPU
With ray-tracing as the framework, Instant Radiosity is mainly about point lights record and replacing. In my application, point lights are collected in the iterative ray recursion process, and they will be passed into the second pass. No tricks here. Th
2010-09-09 20:55:00
849
原创 My renderer update – Tone Mapping (Global Op.)
I tried Tone-mapping a little bit. The first try I took is the simple “Spatially Varying Nonlinear Scale ” which is a global tone-mapping operator. GLOBAL, means no local contrast information will be handled, each pixel will be processed according to globa
2010-09-08 01:20:00
922
原创 My renderer update – Blooming
Blooming is a basic image processing operation, which is an lerp between the blurred(convoluted) values and the original values. Not complex but looks interesting: Another two more interesting image processing techniques are Tone-Mapping and Gamma C
2010-09-07 00:03:00
748
原创 My renderer update – GPU Acceleration on Intersections (more than 2x faster)
According to many papers, about 70%-80% of rendering time is spent on intersections. So the first target to optimize for performance is to accelerate intersections. Due to the parallelism essence of rays, GPU is of course the best choice. After copying a
2010-09-05 14:16:00
1091
5
空空如也
空空如也
TA创建的收藏夹 TA关注的收藏夹
TA关注的人