Preserve data scales during chart animation

For an illustrative count chart, use values 40, 50 and 60 items and an axis from 0 to 100. Record the axis minimum, maximum, tick spacing, pixel bounds and category order. This is the chart’s scale contract. Each scene should either preserve it or explicitly declare a replacement.

A fixed height is not sufficient if the numeric labels change invisibly. A bar that occupies half the plotting area can mean 50 on one axis and 500 on another. List both the data values and their rendered positions when reviewing transitions so that the relationship remains inspectable.

Storyboard defect A begins with a 40-item bar on a 0–100 axis, then crops the baseline upward to 35 as the 50-item bar arrives. The difference appears larger even though the quantities have not changed much. Correct the scene by keeping zero and the original domain visible throughout the comparison.

If the legitimate question concerns small differences around a reference, a different chart can show that explicitly. For example, label a deviation plot as “Difference from 45 items” and display signed differences. Do not silently reuse the original bar geometry with a new meaning. The reference and unit belong beside the chart.

Storyboard defect B uses a 0–60 domain for the first scene and a 0–600 domain when a later value of 500 appears. Existing bars shrink while new ones grow, and viewers may infer that the original quantities fell. Correct it by choosing a domain covering the whole intended comparison before animation, or split the material into separately labeled views.

A deliberate change can use a neutral hold reading “New scale: 0–600 items.” Show the new axis before revealing marks, and retain the earlier scale label in a small comparison panel if needed. This costs screen space and time but prevents the change from masquerading as data movement.

At the beginning, midpoint and end of each transition, record domain, baseline, units, visible ticks and the position of one reference value. Also inspect every frame around an actual axis change; sparse samples can miss a brief jump. Compare the selected marks with the source values after export.

Ask whether the viewer is comparing quantities, changes or deviations. Those tasks may need different encodings. If preserving one domain makes small values unreadable, use labeled small multiples or a separate detail view with its scale stated. Do not exaggerate the small values to keep the visual exciting. Finish with a stable frame whose labels explain the complete comparison.