Patent Application
20060044552
Motion picture visual effects often require the photography of multiple individual images, which must be combined to form one integrated image. If that composite image is intended to appear as from a spatially moving viewpoint, then all the aforementioned individual images may also be required to be imaged from that identical moving viewpoint. The process of repeatably imaging a physically moving viewpoint forms the basis for what is commonly known as “motion-control” photography, in which motorized devices move the viewpoint in predetermined or recordable trajectories or spatial pathways. In some cases, a viewpoint's spatial path may need to be recorded for other purposes, such as acquiring the data necessary for computer generated imagery, a process commonly referred to as “motion capture”.
Many types of devices have been designed to carry cameras with varying degrees of spatial freedom. Most of these devices, from simple pan-and-tilt camera heads to more complex camera cranes, are comprised of mechanically linked elements, each element moving relative to the next. The motion of each of these elements relative to its adjacent element may need to be measured in order to determine the resultant spatial position of a camera held at the end of such a series of linked elements.
One of the more difficult of these axes of motion to accurately measure has traditionally been the movement between a camera cart, or “dolly”, and the track it rides upon. Factors contributing to this difficulty include the fact that track is built from individual sections which must be joined, the variable and potentially long distances over which the dolly may travel, and that some sections of track may need to be curved.
Several types of approach have been taken in the past to attempt accurate measurement between dolly and track. The approach most frequently taken consists of measuring the motion of a motor driving the dolly. Whether the drive mechanism to which the motor is attached is based on mechanical principles such as rack-and-pinion gearing, timing belt, friction drive or other such means, these drives may ensure repeatable measurements at the drive itself but, due to issues of backlash, compliance, or slippage, measurements taken at the motor in such mechanisms may not necessarily provide an accurate representation of movement between the dolly and the track.
In cases where a drive was unnecessary or undesirable, various devices have been used for measurement between dolly and track. A simple friction-contact rubber-wheeled capstan on a common shaft with an encoder, mounted to the dolly and rolling in contact with the track has commonly been used but, due to slippage between capstan and rail, requires frequent recalibration to some reference point on the track. Another approach has been to attach one end of a cable to the dolly, with the cable's opposite end wound on the shaft of a rotary encoder attached to one end of the track, while maintaining constant tension on the cable. Dolly movement in this configuration results in a corresponding encoder shaft rotation, allowing a resultant encoder signal to be generated. Measurement inaccuracy in such applications can result due to elasticity in the cable, and applications requiring the use of curved track sections have prohibited use of this approach.
In film and video production, minimizing setup time is considered to be of great importance. These aforementioned track drive measurement implementations share common drawbacks in such an environment: they may require considerable time to assemble mechanical and electronic components, to align these components, and to calibrate the system to known reference points and in meaningful units of measurement.
The goal of this new approach is a system that, among other advantages, is quick and easy to set up and calibrate, maintains its accuracy in situations where there may be backlash in the drive, can operate with combinations of straight and curved track, works with our without an external drive, and is compatible with most existing track and drive systems.
This new approach places a sensor on the dolly and markings along some surface of the track. The markings are either pre-applied to the track, or applied to the track after the track sections have been assembled. The sensor detects and decodes the markings to produce information about the dolly's motion in real-time, whether that information is position, velocity or acceleration. The information may be used simply to record the motion, or even used to close a servo loop. This approach works well whether a person pushes the dolly or a motor drives the dolly.
One may argue that the advantages of this new approach may be realized by using ordinary linear encoders. But linear encoders fail in aforementioned application because of the following deficiencies:
A. Only short lengths are practical. Linear encoders are expensive, especially in long lengths, and linear encoders are not available in sections that can be joined. They are certainly not available in the arbitrary lengths required for film and video production.
B. Though straight encoder scales are readily available, curved scales are not. The curved scales that do exist are short, and only curve in one direction. They also are very expensive.
C. Linear encoder scales are difficult to align with their sensor, and often require special enclosures to maintain alignment. Alignment can also be time-consuming.
A typical embodiment shown in
When two detectors 6 are used, they may be positioned such that their outputs are phased in the same way as a standard industrial quadrature encoder.
Lenses 9 may or may not be used at the light source 5 and/or the detectors 6 to better focus the light.
A specific distance between the optics 4 and the surface 1 may be required for proper detector performance.
In the currently preferred embodiment, the measured surface consists of a tape with one reflective side and an adhesive coating applied to its opposite side. A multitude of individual non-reflective markings are continuously printed on the tape's reflective side by a seamless-plate flexographic process. By its nature, the adhesive tape can be applied to a surface of any desired length or curvature.
In the preferred embodiment, the emitter 5 and detector 6 are both contained within a commercially available surface-mount electronic encoder chip package, the Agilent HEDR-8000, which includes a single LED light emitter, a pair of lenses, and a photodetector IC. The chip package is connected to a differential line driver on a printed circuit board with additional components, which allow it to behave as a standard industrial quadrature encoder.
Optical characteristics of the emitter and detector may require a specific range of separation between the optics 4 and the surface 1 that the sensor is to read.
