This dataset was created using the Lytro Illum camera.
The images are organized into various different categories that may be interesting to researchers working with light field images.
Each category consists of images and the corresponding RAW files. We have also included the depth maps and the metadata extracted using the commmand line tool from Lytro.
This database is created by Abhilash Sunder Raj, Michael Lowney and Raj Shah with the help of Prof. Gordon Wetzstein from Stanford University.
Light Field cameras have the ability capture both the intensity and direction of the light striking the sensor.
This gives the user the ability to refocus and shift the view point of the image.
Light Fields have a wide range of potential applications from computer vision to virtual reality.
Currently there are very few light field images available publicly to researchers.
Our goal was to create a dataset of natural images that would contain interesting scenes for others to use.
The "living picture" below shows the refocusing ability and the different perspectives captured by the Lytro Illum. Hover above the image to see different perspectives and click on the
image to refocus different parts of the image -
The collection of light field images is divided into 9 different categories. Each category consists of the processed images with the
corresponding depth maps obtained from a command line tool provided by Lytro. The RAW images can be downloaded by clicking on the link on top of
each image. The metadata for each RAW image is also provided. To browse through category click on the links below: