sampling theorem
A mathematical result showing a continuous signal or image can be perfectly reconstructed from discrete samples on a regular grid, provided frequencies above a limit are removed and reconstruction uses the proper peaked, wavy shape rather than squares.
As defined in 325 | Alvy Ray Smith on Pixar, Pixels, and the Great Digital Convergence
Where it comes up · 1
Most important claims · 3 of 3Search all claims for “sampling theorem” →
- The sampling theorem shows that if you sample a continuous image correctly on a regular grid and reconstruct each sample using the proper wiggly peaked shape (not a little square), you can perfectly reconstruct the original continuous image, seemingly throwing away an infinity of information between samples without actually losing it—provided you remove super-high frequencies first.325 | Alvy Ray Smith on Pixar, Pixels, and the Great Digital Convergence
- Poor digital conversions of old films—producing jerky motion in what should be smooth movement, or spinning wheels appearing to go backwards—result from people using an incorrect 'little square' model and not doing frame conversions and reconstruction correctly according to the sampling theorem.325 | Alvy Ray Smith on Pixar, Pixels, and the Great Digital Convergence
- Claims that analog media like vinyl or film are inherently superior to digital are wrong: if digital representation is done correctly per the sampling theorem, it can be made extremely accurate and could even simulate vinyl's sound or film's graininess by adding that noise back—though it may be cheaper to just buy vinyl if you want it.325 | Alvy Ray Smith on Pixar, Pixels, and the Great Digital Convergence