Generation of Synthetic Digital Image Correlation Images Using the Open-Source Blender Software

With camera equipment becoming cheaper and computer processing power increasing exponentially, optical test methods are becoming ubiquitous in the mechanics and dynamics communities. However, unlike more traditional methods where the measurement response of interest is obtained directly from the sensor (e.g. an accelerometer directly provides an acceleration), image-based measurement techniques often require a non-trivial amount of post-processing to extract displacements and strains from a series of images. Using experimental images to develop and validate these post-processing algorithms can be a challenge; real images have a finite depth of field, they can have poor contrast, they can be noisy, there can be calibration errors, etc. It is advantageous to create synthetic images with which image processing algorithms can be investigated without the need to deal with all the complexity and cost involved in a real experiment. Synthetic images also provide access to a “true” analytical solution, which is typically not available in an experiment. However, many synthetic image generation tools are either bespoke research codes or built into commercial software, which can limit accessibility. Blender is a free and open-source 3D software package that supports scene modeling and rendering, among other features. It runs an underlying Python scripting engine, so activities such as building and deforming a mesh or rendering a series of images can be automated. For these reasons, Blender has the potential to be used more widely than current synthetic image tools, as well as perform more sophisticated analyses. While Blender was not designed for engineering purposes, this work will demonstrate Blender’s suitability for generating synthetic test images for digital image correlation, and show its accuracy is comparable to commercial synthetic image generation software packages.