Signal is the real light you want; noise is random variation that hides it. The metric that matters is signal-to-noise ratio (SNR). Random noise grows with the square root of the number of frames, while signal grows linearly — so stacking N frames improves SNR by √N. Fixed-pattern noise (the same every frame) is removed by calibration instead. Understanding which noise is random and which is fixed tells you whether to collect more (stacking) or calibrate it out (bias/dark/flat frames).