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.

随着相机设备变得越来越便宜，计算机处理能力呈指数级增长，光学测试方法在力学和动力学领域变得无处不在。然而，与直接从传感器获得感兴趣的测量响应的更传统方法不同（例如，加速度计直接提供加速度），基于图像的测量技术通常需要大量的后处理来从传感器中提取位移和应变。一系列图像。使用实验图像来开发和验证这些后处理算法可能是一个挑战；真实图像的景深有限，对比度可能较差，可能有噪点，可能存在校准错误等。创建可以研究图像处理算法的合成图像是有利的，而无需处理实际实验中涉及的所有复杂性和成本。合成图像还提供了对“真实”分析解决方案的访问，这通常在实验中是不可用的。然而，许多合成图像生成工具要么是定制的研究代码，要么内置于商业软件中，这可能会限制可访问性。Blender 是一个免费的开源 3D 软件包，支持场景建模和渲染等功能。它运行底层 Python 脚本引擎，因此可以自动执行诸如构建和变形网格或渲染一系列图像之类的活动。由于这些原因，Blender 有可能比当前的合成图像工具得到更广泛的使用，以及执行更复杂的分析。虽然 Blender 不是为工程目的而设计的，但这项工作将证明 Blender 适合生成用于数字图像相关的合成测试图像，并表明其准确性可与商业合成图像生成软件包相媲美。