PhotoModeler automatically decides which of the two sub-pixel markers to use based on a Preference setting ( "General" page, "Sub-pixel method threshold size" field). When the detected target has a diameter, in pixels, smaller than this number, LSM is used and conversely if it is larger than this size, centroid method is used. You can force LSM always by making this number very large (300 pixels for instance) or you can force centroid method always by making this number zero.
Note that LSM is used always for Template Targets. Centroid is used for Circular Targets only.
LSM uses an iterative template matching process that is more robust, handles smaller and lower contrast targets better, and tends to be more accurate than the centroid method. The reason it is not used all the time is because it is slower and its speed falls with the square of the size of the target. It can be very slow with big targets and that is why we have the preferences setting described above.
Note: You can set what RGB channel to use for sub-pixel marking using this ini setting (default is 1 which means all sub-pixel marking is done on the green channel for color images):
[General]
SubPixelMarkingColourChannel=1
(where 0=RED, 1 = GREEN, 2 = BLUE, -1 = convert color to grey scale)
Marking in just one channel may improve marking precision because there will be less negative effects of chromatic aberration of the lens. Most digital cameras have more green sensors than red or blue and this is why the green channel is the default.