Keep the runs comparable
A useful comparison begins at the printer. Keep the build surface, probing setup, bed temperature, and preparation consistent. Allow a similar settling period. Use the same sampled area and grid when possible, and record what you changed between runs.
Label the panels with meaningful names such as Before · 60 °C and After · 60 °C. Bed Mesh can display different grid sizes, but they sample different positions: a smaller range on a coarser grid may simply miss a local feature.
Match scales and height references
Add a second mesh and select Visualize all. Enable Match scales when comparing meshes so heights share a consistent colour and vertical scale. Separate fitted scales can make a small variation look as dramatic as a larger one.
Choose the same Height reference in both panels. Measured height keeps original offsets visible. Centre on mean removes each mesh's average from its display. Remove fitted tilt shows the remaining variation around each mesh's fitted plane. Comparing different references can obscure the change you are trying to inspect.
Check orientation too. Flip controls affect the plot, while the measured-values table stays in original input order. Enter the actual sampled width and depth in both panels if you want physical X/Y coordinates.
Read the change at more than one level
Start with the comparison table: original height range, mean, and standard deviation describe the full measurements. A shift in mean with similar spread may indicate a changed reference between runs; it does not, by itself, demonstrate a flatter surface.
Then inspect the analysis. Fitted X and Y rise describe overall slope in the original column and row directions. Range and RMS without fitted tilt describe what remains after a least-squares plane is removed from each regular grid. Compare original range and residual range together to understand whether slope, remaining shape, or both changed.
Check the heatmaps and individual values for isolated differences. These summaries cannot distinguish real surface change from a changed measurement process. A universal target range cannot account for every printer, probe, surface, and first-layer setup.
Inspect the difference heatmap
In Difference heatmap, choose the before and after meshes. Each cell shows after minus before: blue means the later measurement is lower, warm colours mean it is higher, and white means no change. The colour scale stays centred on zero. The grids and declared sampled dimensions must match; Bed Mesh does not invent intermediate measurements.
The difference view has its own reference control. Use As measured to retain height offsets, or compare relative to each mesh's mean or fitted plane to inspect changes in shape. Check Flip X/Y and the physical probing origin yourself: equal dimensions alone cannot confirm that two cells represent the same spot. RMS point change describes the size of changes, rather than whether they improved a first layer.
Save enough context to repeat it
Use Save workspace for a JSON backup of the meshes and their display settings. Export original CSVs, delta CSVs, and chart images for separate records, and keep your probing notes alongside them. Browser storage is convenient for returning to a workspace; exported files give you another copy. Use a repeated measurement under the same conditions to check whether an apparent improvement is consistent.