At Home with John and Debbie

What?

Slivr (former working name: “Videorama”) is a technique for producing panoramic photographs from a video source. For example:

More…

Each image is made up of hundreds of slices, each from a different frame from the source video.

Why?

It has recently become popular to create panoramic images by taking a series of still photographs, then “stitching” these together using dedicated software. The software needs to do lots of clever geometry to avoid the seams showing: the lens distorts what it sees as it is projected onto a flat surface; the software has to undo this distortion, work out where the joins are between images, then distort the joined-up images back onto a flat surface.

Here is an example of a stitched panorama:

Stitching is very effective, especially with static scenes. The technique allows the creation of extremely high resolution images, and striking compositions. Images can be stitched in two directions, allowing for the simulation of very wide-angled lenses:

However, stitching does not cope well with movement — for example, it is problematic to have a fluttering flag at the border of two stitched images. In this example, a stitched vertical panorama of a moving crane has an obvious shear in the cable:

Also, careless capturing of source images can result in a stitched image which isn’t even approximately rectangular:

Before computers made stitching possible, panoramic images were created in a different way. Slit-scan panoramic cameras would expose a narrow vertical strip of film. The camera would be rotated, and at the same time, the film would be pulled past the slit. The camera would capture a wide image over a period of time.

Some of you may have ensemble school photographs created using this technique. I do. We were arranged in a large semicircle, but in the photograph we appear as a straight line. The straight lines of the school tennis courts appear as sweeping curves. One pupil shook his head as the camera passed him, and he appears as a distorted gargoyle.

Slit-scanning produces fascinating artefacts when faced with a moving subject. For example, if a slit-scan camera were to pan left-to-right while filming a busy road, then cars travelling left-to-right would appear elongated in the photograph, while cars travelling right-to-left would appear shortened. A slit-scanned version of the crane picture above would have shown the moving cable as a continuous curve, rather than a sheared straight line.

The artist Jonathan Shaw has produced a number of works using slit-scanning to capture sporting events, dancing etc., published in the book “Time | Motion“.

Slit-scanning can accommodate irregular camera movement. If you deviate from a straight line while scanning, the end result will be warped to fit the movement into frame. The non-rectangular lava flow image above would be warped in just such a manner, for example.

Andrew Davidhazy of the Rochester Institute of Technology has written about strip photography and developed home-made digital strip cameras by adapting document scanners.

Slivr brings slit-scanning effects within the reach of anyone with a computer and a digital camera capable of capturing video. Unlike Professor Davidhazy’s technique, no hardware hacking is required. The main compromise is in resolution: a typical digital stills camera today has a video capture mode capable of 320×240 at 15 frames/second. Using VideoRama this translates to images 240 pixels high, with an arbitrary width. The frame rate affects the speed with which we can effectively scan a single scene, and the level of detail with which we can capture movement related effects. An HD camcorder would give better resolution, both in time and space.

Within these constraints, however, Slivr still gives us lots of opportunities that stitching does not:

• Movement can be embraced, rather than avoided.
• Arbitrary camera movements can be employed — e.g. dip while panning, or rotate the camera
• Variations in speed of pan and camera shake result in a pleasing “wonkiness” to images (depending on taste!)
• Pleasing “lo-fi” images, reminiscent of Lomography
• … and it’s fun!

How?

Glad you asked. Here is a tutorial.