Chapter 4 On Determining the Stability of a Camera System

And how to detect Flicker

By Ted Kinsman

Spring 2010

www.sciencephotography.com





Digital Streak Photography.

This technique was refined by the author of this text for the sole purpose of testing the stability of digital cameras in different setups, and has a rich history in chemical photography. Digital streak photography uses a set of several hundred to several thousand individual digital images to build a single image that composes elements of each of the images and has the potential to hold tremendous amounts of information as compared to traditional chemical streak photographs.

It should be noted that streak photography has a rich and historic past rooted in chemical photography. This technique is often used to visualize high-speed events such as missiles and bullets. It has only been due to the great scientists and photographers that have come before us that we are able to apply these age old techniques to our advantage in the new age of digital photography and digital cinematography.

To test the stability of a camera under a particular set of parameters, a set 258 images will be used as an example. A set of time-lapse images is often referred to as a data set, due to the tremendous amount of data such a set of images contains. One column of pixels is extracted from each of the images, and these columns are then placed side by side. Since each image represents a different point in time, the resulting images is an image graph. When the images of a subject that is not moving in space are combined, the resulting image is called a digital time-lapse streak. The technique itself can often lead to a quite a bit of confusion, so we will “walk through” the technique step by step. There are a number of programs than can be combined to create these patterns.

Three stills were taken off a cove in Maine to examine time-lapse photography of tides. These stills are from a sequence of 258 images taken over a five-hour span of time. The ocean tides become quite visible.

To examine both the stability of the camera and to see how the scene changes from frame to frame, a column two pixels wide is extracted from each image. The extracted images are placed side by side to the image from one image to the next until all the frames are used up. For this example, a column of pixels will be taken from the area of the image that contains the floating dock with a boat tied to it. The resulting combination of columns is shown below.

It can be quickly observed that the tides were just starting to recede when the sequence began. The horizontal direction of the image above represents a duration of five-hours. Using this known scaling for the horizontal axis, the lowest point of the tide can be determined to be two-hours after the sequence began.

As interesting as the image elements are, the problems are more interesting to the time-lapse photographer. The vertical bands that show areas of inconsistent exposure are of particular interest. This digital camera was set to fully automatic settings, although the focus was locked on an infinite distance setting. The columns of exposure problems are mainly due to clouds that cause the color correction to change radically, a somewhat surprising result that the exposure between sequential images could change so much. When these individual images are combined into a time-lapse movie, the image has a noticeable flicker. This flicker will make the resulting time-lapse movie unusable. It is not only important to be able to identify the conditions in which a digital camera will yield inconsistent exposures, but also be able to identify and eliminate the sources of these problems quickly.

It should be noted that the above sequence of images can be salvaged by removing about four dark frames or by using several advanced digital techniques. This technique is now the standard for determining the stability of a digital camera. If you were to print out the images shown above and try to detect the exposure differences, it would be difficult to pick out the offending images. It is only when the images are placed side-by-side, or sequentially in a movie that the problem becomes noticeable.

The digital time-lapse streak image can be used in two particular ways. First, it is important to see if there are any focal changes or exposure problems. The majority of digital cameras tested in our lab in the last few years will have a little bit of wandering exposure even when set on fully manual modes. If you are evaluating a camera for use in time-lapse photography, this test will tell you very quickly how good or bad your camera is.

The second type of data that can be extracted from this image is how fast an event takes place. This technique can be applied to a number of special situations. Several examples are shown on the following pages.

Three veins are photographed using a microscope attached to a digital video camera. Out of each frame a column is extracted. This forms a streak photograph of blood moving in the veins. The more elongated the blood cells are, the slower they are moving. The vein at the bottom of the frame is moving the slowest. The duration of the photograph is six seconds. If it were longer, the image would show the bunching and expanding of the blood cells due to the pulse. Note the uniformity of exposure across the horizontal axis of the image.

This image is also a streak image, although the source images were 35-mm cinefilm shot at 24 frames per second. The resulting footage was digitized, and this streak image shows the transition from laminar flow to turbulent flow in a two dimensional soap film. The soap film was filmed in white light to bring out the color in the optical interference. The duration of the streak image is about 20 seconds.

This shot is a real digital time-lapse streak image: ice crystals frozen under polarized light over 45 minutes’ duration. The center column of pixels was extracted out of each frame and placed side by side with sequential frames. The transition from one crystal to many can be seen in the right side of the picture.

Digital Peripheral Streak Photography

In this form, streak photography has a rich history in film, particularly as a technique to record the whole image always around a vase or other circular object.

A standard situation would be to photograph a bouquet of flowers, in this case a bunch of pink roses. The roses are placed on a computer-controlled turntable that takes an image each time the stage is rotated .3 degrees. The 1200 images have a column of pixels extracted out of the center of the image and placed beside the next column of the next image in the sequence. Strip by strip a record is built of images each representing .3 degrees around the object. The resulting image is a digital peripheral streak – or a record of the whole circumference of an object.

A bouquet of roses shows all sides of the flowers at once, made from a combination of 960 individual images. The horizontal axis is a time duration of almost 16 hours. The camera took one image a minute. Note that the sequential images are quite uniform and free of exposure problems.



The surface of Indian corn can be shown and mapped with this same technique. Again, a fairly stable camera over the 16 hours used to collect the individual images and to test the camera stability.



A small milk pitcher was used for this fast image test. The camera collected images every five seconds and shows that the automatic exposure is fairly good.



Perpendicular Axis Digital Streak Photography.

Oddly enough, this technique also has roots in traditional photography. The author has seen examples dating back over a hundred years in which photographers have explored the technique. In our case the subject (flower) is placed on a computer-controlled rotating stage, and the camera collects an image every .5 degrees. Instead of extracting a column of pixels, the programmer extracts a row of pixels from each image and places them sequentially into a new image.

In this example, a bouquet of pink roses was used. The resulting image is often quite surprising and unique. Time is represented by the vertical axis of the photograph. In this case the sequence of roses took 16 hours to photograph, so the vertical axis starts at the top left and ends 16 hours later at the bottom.

This particular technique is also useful for testing sequential exposures, like all of the techniques shown on the preceding pages, but the images are often confusing and difficult to understand. It should be noted that interesting and dramatic movies can be created by combining sequences of such strip images as shown above. Since each image represents one particular row of pixels a movie can be made by combining all the first rows, second rows, third rows, etc. until the bottom of the image is obtained. The resulting movie has little scientific value but is particularly beautiful to view.

Title Page and Index of the Time-Lapse Photography FAQ

Drop me a note and let me know what you think.

ekinsman@rochester.rr.com

Thank you for your time.

Click Here To Return to Kinsman Physics Productions