In order to view these light fields, you must first download lifview. Binaries are available for the SGI (Irix 5.0 or higher) and PC (Windows 95) platforms. Each light field listed below is labeled with the number of light samples it contains (u x v views x s x t pixels) and its compressed size (megabytes gunzipped).
If you want to play with larger light fields, see our new archive of light fields produced using the Stanford Multi-Camera Array, a simple Lego Mindstorms gantry, and a light field microscope.
(32 x 32 x 256 x 256, 9.5 MB gunzipped)
Synthetic images drawn from a polygon model.
The model was made from multiple range images using a
volumetric method.
(8 x 8 x 256 x 256, 2.0 MB gunzipped)
A lower-resolution version of the same light field.
(16 slabs x 6 x 24 x 256 x 256, 18.8 MB gunzipped)
The dragon, with 16 slabs that allow full 360-degree sideways rotation.
(32 x 32 x 256 x 256, 8.8 MB gunzipped)
Synthetic images drawn from a polygon model.
This model was shaded to simulate a
metallic patina.
(16 x 16 x 256 x 256, 2.8 MB gunzipped)
The same model buddha4c, but with half the resolution in u and v
(assembled from 1/4 as many images, with twice the synthetic aperture).
(16 x 16 x 128 x 128, 2.1 MB gunzipped)
The same model buddha3c, but with half the resolution in s and t
(generated from images 1/4 the size used in buddha3c and buddha4c) .
(64 x 64 x 128 x 128, 9.5 MB gunzipped)
Volume-rendered light field of two kidneys, spine, and blood vessels.
(64 x 32 x 256 x 256, 18.0 MB gunzipped)
Synthetic images from U.C. Berkeley
Soda hall
walkthrough.
(4 slabs x 32 x 16 x 256 x 256, 16.8 MB gunzipped)
Digitized images captured with a video camera and computer-controlled
gantry. The camera aperture was too small
for the view spacing, producing aliasing.
crayons2.lif.gz (910 K) (16 x 16 x 256 x 256, 8.1 MB gunzipped) Digitized images captured with the same gantry used for the lion above. In this lightfield, the st-plane passes near the magnifying glass. This makes the frame of the magnifying glass sharp, but the crayons appear out of focus. Click on the icon at left to see the full-res result. |
crayons3.lif.gz (980 K) (16 x 16 x 256 x 256, 8.1 MB gunzipped) The lightfield from crayons2, but sheared and resampled so that the st-plane passes through the crayons. The visual effect is the same as changing the plane of focus in a camera having a wide aperture. Now the crayons are sharp, but the frame of the magnifying glass appears out of focus. Click on the icon at left to see the full-res image. |
Historical note about synthetic aperture focusing: The ideas of extracting 2D photographs from a 4D light field, and of integrating across a synthetic aperture while performing this extraction to ensure that the extracted photograph is correctly antialiased, were first described in Levoy, M., Hanrahan, P., Light Field Rendering, Proc. SIGGRAPH 1996. The idea of shearing the light field before extraction in order to effect a change in focus was first demonstrated (on this web page) in 1996 using the two light fields shown above. However, this functionality was not described in Levoy and Hanrahan's paper. The idea was first published, further explored, and nicely analyzed in the frequency domain, in Isaksen, A., McMillan, L., Gortler, S.J., Dynamically reparameterized light fields, Proc. SIGGRAPH 2000. (An alternative frequency domain analysis of light field sampling is given in Chai et al. in Proc. SIGGRAPH 2000.) More recent work on synthetic aperture focusing is reported in papers by Vaish et al. (Proc. CVPR 2004) for imagery from a multi-camera array, Levoy et al. (Proc. SIGGRAPH 2004) for imagery from a single camera and an array of planar mirrors, Ng et al. for imagery from a handheld camera containing a microlens array, and Levoy et al. (Proc. SIGGRAPH 2006) for imagery from a similarly modified microscope. In the popular literature (and as commercialized by Lytro), synthetic aperture focusing is sometimes called digital refocusing. Click here for our current research on light fields and computational photography.
Here are a few sets of the original images, before they were compressed into a .lif file. Each tar.gz file contains the original .rgb images and the .lid (light field description) file that specifies the geometry and general information about the light field.
Note that these raw lightfields can be viewed by running the SGI version of lifview on the .lid files. You can run side-by-side comparisons with the compressed .lif files above to evaluate the effects of the VQ compression.
(8 x 8 x 256 x 256, 8.4 MB gunzipped)
The raw images and .lid file for the low-res version of the dragon
light field. Lifview can display this light field on systems with
64MB of memory.
(32 x 32 x 256 x 256, 100 MB gunzipped)
The raw images and .lid file for
the high-res version of the dragon light field. Lifview requires more than
256MB of memory to view this light field.
(Note: Three images in this archive appear to be corrupt: dragon.2.0.rgb, 4.0, and 8.0. Unfortunately, the originals are no longer available. Our apologies.)
(32 x 32 x 256 x 256, 45 MB gunzipped)
The raw images and .lid file for the high-res version of the buddha
light field. Lifview requires more than 256MB of memory to view this
light field.
You are welcome to use these light fields and images for research purposes. They are not to be used for commercial purposes, nor should they appear in a product for sale without our permission. If you use them in a publication, please acknowledge the Stanford University Computer Graphics Laboratory.
back to the LightPack home page.