In the “old” days when high speed cameras used film, there was a very definite period of time in which the event one wished to record had to occur. This was between the point at which the film had ramped up from a standing start to the desired frame rate, and before the film magazine (which held up to a thousand of feet of 16mm, 35mm or 70mm film) would run out.

Figure 1. Example of Digital Image Correlation (DIC) studying stress on a car tire driving over wooden block. (Image courtesy of Correlated Solutions, Inc.)

High speed digital imaging systems have come a long way from their film camera predecessors. For example, today all-digital video high speed cameras are equipped with massive amounts of onboard, solid-state memory which can be continuously overwritten until the event of interest occurs, making them ideal for non-destructive testing applications. In theory, the event could happen days after the cameras are placed in record mode.

Trigger Modes

Modern high speed cameras offer a wide variety of trigger methods often used for NDT applications, including:

START trigger, where the camera can be entered into the record mode using the events trigger, as in the case of an explosive event, with all frames recorded after the trigger signal at T0.

END trigger, with all frames saved before the trigger signal is received. Again, the trigger is defined as T0 and all preceding frames shown as negative, i.e., occurring before T0.

CENTER, where frames are evenly split either side of T0.

MANUAL mode lets the end user select how many frames are recorded before and after T0. This is frequently used in vehicle impact tests, where typically ten or so frames are recorded before the impact to confirm the vehicle’s approach speed.

RANDOM mode enables a user-selected number of frames to be recorded every time a trigger signal is received. Multiple short recordings are usually combined into a single video sequence. This can be very useful when recording cyclical events, such as engine combustion where only one cylinder is recorded with the camera remaining inactive while the other seven or so cylinders operate.

DUAL SPEED – The camera starts recording, as in START mode, but then changes the recording frame rate by a factor ranging from two to eight times (e.g. from 125 frames per second (fps) to 1,000 fps with the application of a second trigger signal being applied. This can be useful to record a subject’s dormant state at a lower frame rate, say 125 fps before application of a current to the subject, then at the 1,000 fps rate when the subject is being tested, then returning to the slower frame rate to record the resultant condition.

High Speed and Resolution

Today’s high speed cameras utilize very specialized CMOS sensors that enable them to operate at full resolution to speeds as fast as 20,000 frames per second. Additionally, reducing the horizontal and/or vertical pixel resolution makes it possible to push the recording speed to over one million frames per second. However, the results at these very high frame rates typically look better on a data sheet than they do when viewed on a monitor or display.

Figure 2. Fastcam SA-X2 with built-in delay generator, SD Cards, and superior light sensitivity. (Photo courtesy of Photron USA, Inc.)

Many commercial digital single-lens reflex (DSLR), or point-and-shoot cameras, are venturing into the high speed arena. In reality, however, this is more of a marketing ploy than a usable tool because there is little to no control over the triggering. This makes capturing any event (other than an event that is guaranteed to occur within a few seconds of the trigger being applied) nearly impossible. In addition, the resolution is often so low that the imagery is unusable when replayed on a display larger than the one on the back of the camera.

Advanced high speed imaging systems provide megapixel resolution at speeds as high as 13,500 fps, and usable resolutions (defined here as being a minimum of 128 pixels wide by 128 pixels high) at speeds of more than 280,000 fps. Many of today’s cameras can provide true high definition (HD) at 1,920 pixels wide by 1,080 pixels high images in full 36-bit color, as fast as 2,000 frames per second. At present, most televised sporting events include at least a few shots recorded with a high speed video camera.

High speed cameras have traditionally been used to record destructive testing. We’ve all seen the wonderful images of a blade coming off a turbine, a missile exploding in glorious slow-motion detail, or vehicle impact testing where every nuance of an innocent dummy’s demise can be studied in high definition, slow-motion splendor. But there are several areas where high speed cameras are used to great effect in non-destructive testing (NDT) applications, particularly in digital image correlation (DIC).

Digital Image Correlation

Over the years, digital image correlation has found widespread popularity in many applications due to its robustness, flexibility, accuracy, and overall ease of use. Although DIC has conventionally been used mostly for quasi-static or moderately dynamic applications, the availability of affordable and reliable high-speed and ultra-high-speed cameras in recent years has led to a dramatic increase in the use of DIC systems for dynamic applications.

Digital Image Correlation utilizes a pair of precisely synchronized high speed cameras to record a random pattern painted onto the side of the object of interest (Figure 1). The recording is then analyzed for shifts in the pixel intensity array subsets on two or more corresponding images. Due to the ease in which DIC can be implemented, it has been widely adopted for micro- and nano-scale mechanical testing applications, and to validate finite element analysis (FEA) models utilized in the development of many products. For example, in the past, in order to validate the computer model, engineers typically placed strain gauges in the predicted high-stress locations, applied a force in the same location and direction as the model, and then analyzed and compared the strain data from the strain gauges to the theoretical data from the FEA model. Typically, the data does not match, and then engineers must determine how to adjust the mechanical testing or the computer model to make sense of the results. Among the many benefits high performance, high-speed cameras provide, it is the ability to precisely synchronize multiple units that has made cameras like the new Fastcam SA-X2 mainstays of the Digital Image Correlation market (Figure 2).

Particle Image Velocimetry (PIV)

Figure 3. Sample model of tomographic PIV measurements. (Image courtesy of LaVision Inc. and Delft University of Technology)

Another non-destructive test in which high speed cameras have gained widespread acceptance is the study of particle image velocimetry (PIV) for the analysis of particle movement within a gas or fluid (Figure 3). Here seed particles are very precisely tracked to ascertain the flow within the image sequence. These particles are often very fast moving, and extremely small, so it is critical to keep the time between images as short as possible. Some cameras minimize the time between subsequent frames to the order of a couple of hundred nanoseconds. Then the cameras are precisely synchronized to the laser to trigger it at the end of the odd numbered frames, i.e., frame 1, 3, 5, etc. and at the very beginning of the even frames, frame 2, 4, 6, etc. The PIV software then tracks the seed particles’ motion to provide a very exact map and analysis of the fluid or air flow being studied.

Critical Light Sensitivity

Low cost LED lighting has greatly helped, but it is the sensor design itself that has made the greatest inroads into light sensitivity. Miniature micro lenses have been successfully employed to ensure the maximum possible amount of light is directed onto the light-gathering area of the pixel.

A pixel’s fill factor can roughly be defined as the percentage of the pixel that is used for collecting light, as opposed to the assorted control circuitry also required. As recently as five years ago, a state-of-the-art high speed camera could boast a light sensitivity of around ISO 8,000 for monochrome and 1,600 for color. The reason for this disparity, typically around two to three times between color and monochrome, is the utilization of a color filter array (CFA) over the monochrome sensor from which the color levels for each pixel are calculated. Today, the most light-sensitive high speed camera available delivers an ISO 12232 Ssat (saturation-based sensitivity) qualified light sensitivity of 25,000 for monochrome, 10,000 for color.

Light sensitivity is a very important factor in NDT and other high speed imaging applications. This is why it is essential to carefully evaluate the high speed cameras available, when choosing an imager for a particular application. At first glance, it may not appear that ISO 25,000 represents the most light sensitive camera available, but this is because not all manufacturers accurately report their cameras’ stated light sensitivity. One manufacturer states their cameras are tested to ISO 12232, but then adds a mysterious letter ‘T’ after the stated levels of 43,700 and 3,900 respectively – neither of which are recognized ISO values.

It is also noteworthy that the ‘rule of thumb’, i.e. a color sensor being half to a third as sensitive as its monochrome counterpart, does not apply here. The light-sensitive standard established using ISO 12232 states that monochrome sensors shall be measured using tungsten illumination with an infrared (IR) cut-off filter in place to remove the otherwise overpowering infrared light. When this filter is removed, the amount of light, most of it being unusable infrared light, is significantly increased, hence the outlandish figures noted above. Assuming a usable figure of 43,700 was available (which it is not), ISO speed values dictate that any figure in between 40,000 and 49,999 shall be listed as ISO 40,000.

When evaluating the light sensitive ISO values, always take a critical look at what the stated claims are from all high speed camera manufacturers. If the monochrome value is in excess of 40,000, it does the end user no good as they cannot realize the claimed benefit; it would be like a car manufacturer claiming 1,000 miles per gallon if you were to use their top secret and commercially-unavailable fuel. So make sure when you compare light sensitivity, you compare like-to-like to reveal a true ISO 12232 Ssat value and not something else. Most importantly, insist on a demonstration at your facility so that you can be absolutely certain you are comparing apples to apples.

This article was written by Andrew Bridges, Director, Sales and Marketing, Photron (San Diego, CA). For more information, contact Mr. Bridges at abridges@ or visit .

Photonics Tech Briefs Magazine

This article first appeared in the November, 2013 issue of Photonics Tech Briefs Magazine.

Read more articles from this issue here.

Read more articles from the archives here.