{% extends "base.html" %} {% block title %}Results · {{ project.name }}{% endblock %} {% block body %} {% set active_tab = 'results' %} {% include '_project_tabs.html' %} {% if not files %}
No CCM data (greyscale sensor or CCM disabled).
Delta E (CIE2000) per Macbeth patch after matrix optimisation, at the selected colour temperature — each bar is the patch's reference colour; lower is better. Colour ΔE removes the overall-brightness offset (set at runtime by exposure/tone), so it reflects the hue/saturation accuracy the matrix actually controls; the badge is graded on it.
Each Macbeth patch on the CIE 1976 u′v′ chromaticity diagram at the selected colour temperature: the open circle is the reference colour, the filled dot is the camera's corrected (post-CCM) colour, and the line between them is the residual hue/saturation error (shorter is better). The triangles are the sRGB/Rec.709 gamut (solid — they share primaries) and Rec.2020 (dashed); ◇ is the D65 white point and the curved outline is the spectral locus. This shows the direction of colour error that the ΔE bars only give as a magnitude.
White-balance locus — R/G against B/G across the calibrated colour-temperature range (hover for the CT).
Predicted luminance non-uniformity of the flat-fields after this correction (worst capture across colour temperatures). With strength < 1 some residual is by design — corners are deliberately left slightly darker to limit noise amplification.
Luminance per lux (Y / lux·exp·gain) for each capture against colour temperature — ideally flat. The dashed line is the value the lux calibration uses. A sloped/scattered response means the sensor's luminance-per-lux is colour-temperature dependent, so a single lux reference will read off away from the anchor.
The template gamma curve (solid) against the target transfer function (dashed), with the measured Macbeth greyscale patches (points) pushed through the actual curve. Gamma is not tuned — it is a fixed curve — so this is a check, not a calibration. The greys are normalised so white sits at 1:1; deviation in the shadows/mids means sensor non-linearity or a curve that differs from the target. Linearity R² near 1 means the sensor itself is linear, so any tone error is the curve's shape rather than the sensor.
Per-channel black level of each dark frame against total exposure (shutter × gain) — ideally flat lines at the calibrated value (dashed). A slope means the pedestal drifts with exposure (dark current); diverging R/G/B channels suggest a light leak during the dark captures.
All dark frames share one total exposure (shutter × gain), so there is no trend to plot — values are the per-channel means across the dark frames. Capture darks at several shutter/gain settings to map black level against exposure. Diverging R/G/B values suggest a light leak during the captures.