I have an issue where increasing the size of the ROI by 33% results in a poor and distorted strain field.

The parameters used are:

• analysis mode: subset-based full-field
• initialization method: feature matching
• subset size: 21 px
• step size: 5 px
• ssig: 200
• shape functions: all cases activated

The frame size is 1456 px × 1088 px. Each speckle is about 4 px, with a distance between speckles of about 8 px. Each subset contains roughly 3 speckles on both axes, giving over 50% gradient. I have a series of 4000 grayscale images from which I am trying to obtain the strain field.

This is a single-camera 2D DIC setup. According to various credible sources, the camera, lighting, and specimen setup are close to ideal.

Could you please help me understand what is going on here? I have other test specimens where a similar problem occurs—that is, it works well on smaller ROIs but when I try to run with a larger ROI, the results show a poor and distorted strain field.

Thoughts: Could this be a computer memory issue?