Considerations for High-Magnification Tests
Posted by Micah Simonsen, Last modified by Micah Simonsen on 13 October 2016 01:04 PM
Considerations for High-Magnification Tests
Some of the most useful results in Digital Image Correlation come from high-magnification, small field of view tests. At the same time, these tests can be increasingly difficult, depending on the magnification. This article is an overview of some of the challenges and solutions in this kind of testing.
Defining "high magnification"
While there's no strict line, any test with magnifications of around 1:4 or 1:2 or greater can be considered high magnification. This ratio refers to the size of the sensor compared to the size of the field of view, so for a high-speed camera with a very large sensor, a 3-4" field of view could be high mag; for a camera with a 2/3" sensor, fields under 1-2" are high mag. The maximum magnification we will be covering here is around 1:1; higher magnifications will require a stereomicroscope.
Testing at high magnification creates a lot of minor issues which can come together to make this kind of test very frustrating. The limited depth of field may make it hard to even find your specimen, and a tiny bump of the cameras may require you to start over. Be patient! Just before starting the test, double check your focus and setup, and be prepared to start over if necessary.
Lens selection and setup
For small fields of view, we have to select a lens that focuses close enough to image the specimen, and that also physically allows a stereo rig. So, while some lower focal length lenses (i.e., 12mm or 17mm) may allow the magnification to image a 10mm object, this setup normally wouldn't permit two cameras to be in the right position. So, a longer lens will typically be chosen - often, in the 75mm or 100mm range.
Lenses referred to as "macro" typically allow close focus, and lenses are available which focus as close as 1:1. For other lenses, it may be necessary to use extension tubes to focus closer. These tubes are simple mechanical extenders that place the lens farther from the sensor, and can be threaded together to add more extension. At some point, the image will become vignetted (dark corners), so there's a limit to just how much we can extend a lens. Note that extension tubes don't change the focal length of the lens, only the close focus limit.
Once we have lenses on both cameras and a good stereo rig, we'll have to address focus and depth of field. At high magnification, you will find your depth of field very limited. For highly shaped objects like cylinders, it may be very difficult to get the whole specimen in focus. Stopping the lens down (higher F/numbers) will help, but very small apertures will cause a decrease in the sharpness of your image. For a typical lens and camera, F/numbers above around F/8 or F/11 will start to cause blur; at this point, you will have to make a compromise between depth of field and image quality. For flat objects, using a small stereo angle will make the specimen less oblique and alleviate the depth of field issue somewhat.
With the small field of view and the typically small apertures, you will need quite a bit of light for your test. Bright incandescents can be used, but spot LED lights or fiber-optic illuminators will give the best concentration of light without heating the specimen as much.
Note: before calibrating, be sure your rig is very snug, and that any cables are securely tied down. At these magnifications, jarring the rig or pulling on a cable can have a very big effect.
Obviously, a small grid will be required for calibration.
For this calibration, you will almost certainly have to move/remove your specimen. The limited depth of field will make it impossible to calibrate directly in front of the specimen, and with the glass grid, an even background must be present. If the specimen absolutely cannot be moved, you can rotate the entire rig - carefully - and calibrate next to the specimen. If you do this, rotate the rig very smoothly, and be especially sure that your cables are tied down.
Hand-holding the grid will be problematic here - motion blur can be a problem, and simply positioning the grid can be frustrating. If you have flat surface to brace the grid on, that will be a big help; better still will be to use a pan/tilt fixture or third-hand type rig
Because your depth of field is so limited, it will be difficult to get appreciable tilt angles. Still, it's important to tilt the grid as much as possible out-of-plane; be aggressive and take images even when the edges of the grid are defocused. This won't hurt your score although the image may not be usable.
If the tilts are still not sufficient, the result will be very poor estimates for the Center (X) and Center (Y) values. In this case, you can either try calibrating again, or as a last resort, use the High magnification option in the calibration dialog. This will force the values to the geometric center of the sensor - i.e., on a 2448x2048 sensor, the values will be forced to 1224,1024. This isn't quite as good as getting a good true center estimate, but it's much better than having a poor estimate.
Before testing, be sure to take a static image and run a shape analysis! This will be the time to find out if there are any problems with your focus or if your calibration has been disturbed.
Once you have a good calibration, testing will be comparatively straightforward.