Listed below are the light fields in our archive. For each light field, there is a link to the imagery - sometimes in several forms, and sometimes accompanied by calibration information. Following this is a link that allows you to view the light field in your browser using our Flash-based light field viewer.

Notes about the light field viewer

In most cases you shouldn't need to download any software to use our viewer; just click on the indicated links below. Be warned though, this involves loading the entire light field (usually at slightly reduced spatial resolution) into memory. For the largest light fields, this involves downloading about 30MB of data, and will cause your browser to use up to a gigabyte of RAM. Firefox 3 uses notably less RAM than Firefox 2 when viewing these light fields, as does Internet Explorer 7. Safari and Opera should also work, with the appropriate flash player plugin.

You're free to take the viewer and use it for your own light fields. The source code and instructions for the using or modifying viewer are available.

Instructions for the viewer

The online viewer was written by Andrew Adams. The idea for the autofocus interaction was taken from Ren Ng's cool demos at Refocus Imaging.


Light Fields from the Lego Gantry

The light fields in this section were acquired by Andrew Adams.

Chess

289 views on a 17x17 grid, image resolution 1400x800
Original camera images, Rectified and cropped images

A chess board with pieces. Chess boards are great for demonstrating refocusing.

View light field online


Lego Bulldozer

289 views on a 17x17 grid, image resolution 1536x1152
Original camera images, Rectified and cropped images

A Lego Technic bulldozer. Very complex geometry.

View light field online


Lego Truck

289 views on a 17x17 grid, image resolution 1280x960
Original camera images, Rectified and cropped images

A Lego Technic truck. Very complex geometry.

View light field online


Eucalyptus Flowers

289 views on a 17x17 grid, image resolution 1280x1536
Original camera images, Rectified and cropped images

Some traditional Australian flowers. Lots of fine geometry.

View light field online


Amethyst

289 views on a 17x17 grid, image resolution 768x1024
Original camera images, Rectified and cropped images

A chunk of amethyst with interesting specularities and some translucency.

View light field online


Bracelet

289 views on a 17x17 grid, image resolution 1024x640
Original camera images, Rectified and cropped images

A bracelet with translucency and complex specular geometry. Jewelry courtesy of Elena Adams.

View light field online


The Stanford Bunny

289 views on a 17x17 grid, image resolution 1024x1024
Original camera images, Rectified and cropped images

Diffuse bunny with well-known geometry. The angular spacing is quite narrow, so this light field appears to be correctly antialiased (no ghosting when using a large aperture).

View light field online


Jelly Beans

289 views on a 17x17 grid, image resolution 1024x512
Original camera images, Rectified and cropped images

Jelly beans of various colors at various depths. Simple geometry with some texture and specularity.

View light field online


Lego Knights

289 views on a 17x17 grid, image resolution 1024x1024
Original camera images, Rectified and cropped images

A lego scene. Contains relatively simple geometric objects at various depths. The lego is slightly specular, and untextured in many places, making depth reconstruction more challenging than you would expect for this scene. The regular grid of studs on the base plate provides a good demonstration of synthetic focus.

View light field online


Tarot Cards and Crystal Ball (large angular extent)

289 views on a 17x17 grid, image resolution 1024x1024
Original camera images, Rectified and cropped images

A scene with a crystal ball resting on tarot cards. The ball acts as a lens. The cards act as diffuse textured objects at many orientations and depths. This light field is captured at larger angular extent than the next one, and hence contains more angular aliasing - this is good for depth reconstruction but bad for refocusing.

View light field online


Tarot Cards and Crystal Ball (small angular extent)

289 views on a 17x17 grid, image resolution 1024x1024
Original camera images, Rectified and cropped images

The same scene as above, captured with smaller camera spacing. This has less angular aliasing, and is hence better for refocusing and worse for depth reconstruction. The camera is at F/8 at 22mm, so the size of the aperture was 2.75mm. The spacing between two camera positions was just under half of this (roughly 1.33mm).

View light field online


Treasure Chest

289 views on a 17x17 grid, image resolution 1536x1280
Original camera images, Rectified and cropped images

A chest with gemstones, coins, and jewelry spilling out. Contains highly complex specular geometry at a fairly high resolution. Also contains some translucency. All jewelry courtesy of Elena Adams.

View light field online


Lego Gantry Self Portrait

289 views on a 17x17 grid, image resolution 640x1024
Original camera images, Rectified and cropped images

The lego gantry capturing itself, looking in a mirror. The gantry naturally moves as it captures itself, which results in some unusual apparent depths for each object. Furthermore, the gray towers move vertically but not horizontally, giving them one apparent depth from horizontal parallax, and a different depth from vertical parallax. The focus is hence astigmatic. More usefully, panning the scene shows all the positions the lego gantry moves through.

View light field online


Microscope Light Fields

The light fields in this section were acquired by Marc Levoy. More detail on these datasets can be found at the Light Field Microscope page. Some things to be aware of when viewing these light fields using the Flash-based viewer:

You can also view these light fields using LFDisplay, our hardware-accelerated viewer for microscope light fields. This viewer requires downloading an executable, but it won't have the resolution limitations of the Flash-based viewer.

Fluorescent Crayon Wax

400 views on a 20x20 grid, image resolution 170x114
Rectified camera image

The specimen is a 200-micron-tall "tower" of fluorescent crayon wax. The objective is 20x/0.5 (dry), used without a cover slip. The parallax you see in this light field is normally unavailable to the microscopist, who only sees a single, orthographic view of the specimen.

View light field online

(with no magnification)


Golgi Stained Neurons at 20x

256 views on a 16x16 grid, image resolution 289x289
Rectified camera image

A Golgi-stained slice of rat brain (courtesy of Shinya Inoue, Marine Biological Laboratory), imaged with a 20x/0.75NA (dry) objective. Lateral spatial resolution on the specimen is 1.5 microns. Angular resolution is about 5 degrees. In other words, when panning across the light field in pinhole mode using our Flash-based viewer, each step is 5 degrees.

View light field online

(with no magnification)


Golgi Stained Neurons at 40x

256 views on a 16x16 grid, image resolution 289x289
Rectified camera image

The same slice of rat brain as above, imaged with a 40x/1.3NA oil immersion objective. Note the large amount of parallax in this light field. A 1.3NA objective can capture rays leaving the specimen at angles up to 59 degrees on either side of the optical axis. For photographers, this would correspond to a lens having a relative aperture of f/0.58 - far faster than any commercially available photographic lens.

View light field online

(with no magnification)


Light Fields from the Gantry

The light fields in this section were acquired by Vaibhav Vaish.

CD cases and poster behind plants

105 views on a 21x5 grid, image resolution 650x515
Camera images, Keystone corrected images
Camera positions, Homographies for keystone correction

This light field (and the following one) have been used in research for 3D reconstruction of occluded surfaces [5].

View light field online


CD cases and poster (unoccluded)

105 views on a 21x5 grid, image resolution 650x515
Camera images, Keystone corrected images
Camera positions, Homographies for keystone correction

This light field is identical to the previous one except that the plants in the foreground were removed.

View light field online


Toy Humvee and soldier (unoccluded)

256 views on a 16x16 grid, image resolution 650x515
Camera images, Keystone corrected images
Camera positions, Homographies for keystone correction

View light field online

In "Full Aperture" mode, move the focus slider towards "Far Focus" to get the objects in focus.


Toy Humvee and soldier behind dense foliage

256 views on a 16x16 grid, image resolution 650x515
Camera images, Keystone corrected images
Camera positions, Homographies for keystone correction

This light field is identical to the previous one except that dense foliage was placed in front of the humvee and soldier. The exposure time was also increased and lights turned up slightly to avoid dark regions.

View light field online


Light Fields from the Camera Array

The light fields in this section were acquired by Bennett Wilburn and Neel Joshi.

Objects behind dense foliage

88 views, image resolution 640x480. Color calibration performed using [3].
Camera positions, Homographies for keystone correction
Camera images, Keystone corrected images
We also computed mattes for the foliage in the foreground using standard blue screening:
Images with black curtain behind foliage, Images with white curtain behind foliage
Keystone corrected mattes

This light field was used in matted synthetic aperture imaging [4] and 3D reconstruction [5].

View light field online


Students behind bushes

45 views, image resolution 640x480.
Camera images, Keystone corrected images
Camera positions, Homographies for keystone correction
Synthetic aperture sequence (Quicktime movie)

This light field was used in our paper on plane+parallax calibration [8]. Since it was acquired outdoors, we were not able to perform color calibration.

View light field online


© 2008 Stanford Graphics Laboratory
Created by Vaibhav Vaish. Updated by Andrew Adams.
Last update: