Andersen Chen (aychen)

CS129 / Project 5 / HDR

In this project, we automatically combine multiple exposures into a single high dynamic range radius map. Then, we map this radiance map to an image. To construct our radiance map, we solve for a function g that has certain constraints that maps pixel values to the log of exposure values. To do tone mapping, we first have a simple global tone mapping operation. A natural logarithm does fine. To do local tone mapping, we use a bilateral filter and process details and large scale structure differently. We apply a natural logarithm to the intensity values, then apply an offset and scale to the large scale structure. We recombine the large scale structure and the details.

Pipeline of Results

Relationship between exposure and pixel values for garage

A sample of bilateral filtering based decomposition into detail and large scale structure for garage

Final results with local tone mapping