Binning

In hardware binning, charge from adjacent pixels is combined on the sensor before the analog-to-digital conversion. For N×N binning, the effective pixel size increases by N, the signal increases by N², and the spatial resolution decreases by N in each dimension. Importantly, read noise is incurred only once per super-pixel (not once per constituent pixel), which is why binning improves SNR more effectively than software averaging.

Effective well depth also scales with N², meaning 2×2 binning on a 20,000 e⁻ sensor gives an effective 80,000 e⁻ well — dramatically extending dynamic range.

Binning is a practical tool for low-light imaging scenarios where signal is the limiting factor. For dim fluorescent markers, 2×2 binning can mean the difference between a noisy, unusable image and a clean, quantifiable one. The resolution tradeoff is acceptable when the structures of interest are larger than the binned pixel size.

Imaging a weakly expressed membrane marker at 40× with a 6.5μm pixel camera: at 1×1 binning, the signal is 50 photons/pixel with SNR ≈ 7. At 2×2 binning, the effective pixel is 13μm, signal is 200 photons, and SNR ≈ 14 — a 2× improvement that transforms a marginal image into a reliable measurement.