Two lights can have the same minimum and maximum measured output yet spend very different amounts of time near each level. Percent modulation describes that range. Flicker index also accounts for the area of the waveform above its own mean, so its result changes with waveform shape and duration.
Use a linear optical signal measured under stable conditions, with enough bandwidth and sampling to represent the variation of interest. The values can be in a calibrated optical unit or a common proportional linear scale. Multiplying every value by the same positive factor leaves both ratios unchanged. A nonlinear camera brightness value, an uncorrected detector offset or clipped peaks can change the ratios and needs attention before the calculation.
The ENERGY STAR measurement-method draft gives the percent-flicker and flicker-index formulas on pages 6–7, with measurement considerations earlier in the document. It is a historical 2017 US draft. These descriptive calculations do not turn it into an Australian compliance test or provide a health threshold.
Represent a complete cycle without changing the time weights
For equally spaced samples, each value represents the same time interval. Their arithmetic mean is the time-weighted mean under that discrete representation. Select complete cycles: a cropped interval containing mainly bright or dark parts will change the mean and the area ratio.
For constant segments with different durations, let each optical level be Lᵢ and its positive duration be Δtᵢ. The entered model assumes that level remains constant for its entire segment. Then:
Mean L̄ = Σ(LᵢΔtᵢ) / ΣΔtᵢ
All durations must use the same unit. Seconds and milliseconds give the same ratios when every duration is converted together. This calculation integrates the declared segments exactly; it does not reconstruct an unknown curve between instrument samples. An actual waveform needs adequate sampling and an appropriate measurement method.
Separate amplitude range from area above the mean
The amplitude calculation is:
Percent modulation = 100 × (Lmax − Lmin) / (Lmax + Lmin)
Flicker index uses the same cycle but a different question:
FI = Σ[max(Lᵢ − L̄, 0)Δtᵢ] / Σ(LᵢΔtᵢ)
For an artificial two-level waveform at level 1 for half a cycle and level 3 for the other half, the mean is 2. Percent modulation is 100 × (3 − 1)/(3 + 1) = 50.00%. The area above the mean is (3 − 2) × 0.5 = 0.5, while the total area is 1 × 0.5 + 3 × 0.5 = 2. Flicker index is therefore 0.2500.
For comparison, the ideal continuous waveform L(t) = 2 + sin(t) has the same extrema, 1 and 3, and the same 50% modulation. Over one full cycle its mean is 2, its positive area above the mean is 2 and its total area is 4π. Its analytic FI is 1/(2π) ≈ 0.1592. A finite set of equal-time samples approximates this continuous result; the two-level input does not represent a sine wave simply because the extrema match.
A positive constant and an all-zero signal have different meanings
A constant positive signal has zero modulation and zero FI under these definitions. An all-zero signal has zero denominators, so both ratios are undefined. Zero output is not evidence of a flicker-free operating light. The reported minimum and maximum are also the extrema of the supplied data; insufficient sampling can miss the actual peaks and troughs.
Keep the result with its measurement conditions
Record the operating state, instrument, range, geometry, dark correction, sample rate, bandwidth and selected cycles when comparing lights. The ratios alone contain no frequency weighting and do not calculate PstLM, SVM, a health rating or a camera-safe shutter setting. Camera behaviour also depends on exposure and readout conditions that these inputs do not describe.
A measured dimming curve answers a different question: how steady-state light output and input watts vary with a control setting. A smooth dimming curve does not establish the time waveform at each setting, so retain the waveform measurements alongside any dimming comparison.