When creating a cone-catch image it is fairly common to have some negative values, which are highlighted in red overlay (plus the log lists the frequency of negative values in each image). Negative values can be caused by two main factors:
- Image calibration imperfections. These could be due to incorrect measurement of the reflectance of the grey standards used to calibrate the image. This is particularly important when two or more standards are used. Negative values can also occur where the intensity of light falling on the standards does not match the intensity on the samples being measured (e.g. if the sample is slightly in shade compared to the standards, it could be seen as being negative). This highlights the importance of correct use of grey standards and lighting. These negative values tend to occur in very dark sections of the image, and setting the values to very low values (near zero default) is appropriate.
- Alternatively, when negative values occur in very colourful samples this can imply the colour is out of gamut of either the camera or visual system. In this instance the numbers being produced (both in the channel with negative values and the other channels) should be treated with caution. You should note in your results that these colours are likely to be at or beyond the limit of the equipment’s or cone-mapping model’s colour reproduction abilities.
Any image processing step which requires calculation of RNL-space values (e.g. various aspects of the QCPA framework, JND calculation etc…) cannot accept negative values. In this case, and if deemed appropriate you can have the ‘convert to cone catch’ script change all negative values to a near-zero number which is accepted by the RNL calculations.