semper
Glossary
Every term semper and this site use, in one place — in plain language, as they mean it here.
| Term | What it means here |
|---|---|
| Reference | The undeformed frame everything is matched against. |
| Deformed frame | One load step — a photo taken while the object is being deformed. |
| Subset | The small pixel window matched at each point. The unit the correlation actually tracks. |
| Step | Spacing between grid points. Smaller step = denser field, longer run. |
| Strain window | How many neighbouring points are fitted together to get strain from displacement. |
| VSG | Virtual strain gauge — the distance one strain value actually covers, in pixels. VSG = (strain window − 1) × step + 1. |
| ROI | Region of interest — the area you correlate, optionally with erased holes over grips or markers. |
| SSSIG | Sum of squared subset intensity gradients — a measure of how much pattern a subset contains. Drives the subset-size recommendation. |
| ZNSSD | A match-quality score for each point (lower is better), unaffected by overall lighting changes. Appears in the CSV so you can filter out weak points. |
| ICGN | Inverse-compositional Gauss–Newton — the sub-pixel matching method the engine uses to locate each subset to ~1/100 of a pixel. |
| mε (millistrain) | The unit strain is reported in. 1 mε = 0.001 strain = 1000 microstrain. |
| Decorrelation | When the deformed image no longer resembles the reference well enough to match — from too much deformation, motion out of view, or pattern damage. A run stops itself when it happens. |
| Sweep | One frame solved across a lattice of subset × strain-window settings, to show where the answer stops changing. |