ImageReasonMethodology
This page documents how ImageReason arrives at an answer, not why it should be trusted. Where a specific tool's method differs from a general convention described here, that tool's own worked example and limitations take precedence.
| Convention | Default |
|---|---|
| Circle of confusion | Sensor-format-derived default, overridable per tool |
| Reference viewing distance (print) | 25 cm / 10 in unless a tool states otherwise |
| Diffraction limit basis | Rayleigh criterion against the sensor's pixel pitch |
| Rounding | Stated precision shown next to the value it applies to |
Optical geometry and system equivalence
Field of view, equivalent focal length, and depth-of-field-equivalent comparisons are derived from documented sensor dimensions and standard optical geometry formulas, not estimated from a marketing spec sheet. A system-equivalence result states the geometry it matched (framing, depth of field, or both) rather than implying a single universal "equivalent" number exists.
Exposure and dynamic range
Exposure relationships (shutter speed, aperture, ISO, and their stop-for-stop tradeoffs) follow standard exposure-value math. Dynamic-range and noise behavior are described in relative, comparative terms rather than as an absolute guarantee, because sensor performance varies by generation and manufacturer in ways a formula alone cannot capture.
Focus, diffraction, and resolution
Depth of field, hyperfocal distance, and diffraction-limited aperture are calculated from circle of confusion, focal length, subject distance, and pixel pitch. A result states the circle-of-confusion assumption it used, because that single input changes the answer more than any other, and treats its own output as a model approximation, not a claim about a specific lens's real-world sharpness.
Local image analysis
Where a tool inspects a selected image (dimensions, EXIF, histogram, clipping, or a labeled sharpness or noise estimate), it measures what the file itself contains and does not infer facts the file cannot support. A heuristic measurement, such as an estimated blur or noise level, is always labeled as an estimate rather than presented as a certified reading. No image-analysis tool ships until its measurement method, supported formats, and uncertainty language are locked in a written specification; see the Privacy page for what does and does not leave your browser.
Validation approach
Deterministic calculations are checked against known reference examples and edge cases before release, and a tool's worked example is checkable against its own default, loaded values, not a separate hidden formula. Where a competitor's tool produces a different number, that alone is not treated as ground truth; a result is validated against documented optical and imaging conventions, not against another site's output.