Circa early 2000s, I noticed a somewhat obvious issue with Circle of Confusion calculations pertaining to the actual impact of depth of field. This was something I noticed on both motion picture film scans, still film scans, and digitally acquired imagery when viewed in a theatrical setting on both projected and emissive displays, which is also visible on modern in home displays as well. However, there is a practical versus technical answer when it comes to applying this into creative efforts, which is why I am supporting both concepts with this tool.
Conventional depth-of-field math uses a print-era circle of confusion (CoC): the largest blur spot still seen as sharp in an 8×10″ print at ~25 cm, scaled to the recording format (often image circle ÷ 1500).
The photosite-based option ties the limit to the sensor instead: CoC = pixel pitch × a sampling factor. The factor is the circle of confusion expressed in photosites — how many pixels wide a detail must be before it counts as resolved.
That factor depends on the sensor’s color architecture:
- Monochrome, no OLPF — resolves nearly to the sampling limit (~1.4× pitch).
- Bayer CFA — full color is rebuilt by demosaicing, so a clean detail needs ~2 photosites (2× pitch).
- Bayer CFA + OLPF — an optical low-pass filter blurs slightly to suppress aliasing, widening the spot further (cine ~3×, strong ~4×).
Because this is a per-pixel (100%) sharpness criterion, it is stricter than the conventional ~0.03 mm CoC — so hyperfocal distances come out larger and depth of field shallower. For results that match a traditional DoF calculator, pick a fixed CoC from the Circle of Confusion menu instead.
Phil Holland