check_focus (package) at this time contains two tool:

	    checkfocus & bestfocus

Checkfocus Is the "broad brush" tool, it describes the overall picture and the centre part

Bestfocus: Is a more narrow focus (pun unintended) tool, it can find the "best
small box" of focus.


Which tool you use depends on your style of photography and your workflow. If
your images consist of a large central figure which should be in focus
(portrait) or a wide area all in focus (Landscape) then checkfocus may be enough
and is the faster tool.  If your photos contain lots of nice Bokeh and a small
area in sharp focus, then you might do better with bestfocus. The latter tools
has lots of tuning options you may settle on a "usual suspects" set which suit your
photographic style (e.g. your images are always 4000x6000 and you need an area of
at least 100x150 for it to be a success ....but the location moves around a lot)

Note focus has a very specific meaning WRT this tool (it's contrast based) it
can produce some unexpected results.

This is the TL;DR version. To see how to really use the tools, read the man pages
(or in the extreme "use the source Luke" )

(I've added some blanks to ease human understanding)

This is an image of a drone almost hitting the camera (in another drone) in a
high speed pass '-sharp' is "OK", '-blurred' has been deliberately blurred (in gimp)
and '-focused' has a small focus area just round the drone and the rest is blurred.

                           Overall    Hoiz    Vert        Box     Boxh      Boxv
$ ./checkfocus drone*.jpg 
drone-blured.jpg           22357791   4864803  17492988   988237  268736    719501
drone-focused.jpg          66158661  25155372  41003289   405334  116660    288674
drone-sharp.jpg           320541267  74258203 246283064 13836176 3535124  10301052

So overall '-sharp' is clear winner. Still wins considering just Horizontal Or Vertical (buts closer)
looking at the (default) centre box sharp is way better, so the "in focus" is proably not the centre.


$ ./bestfocus drone*.jpg 
drone-blured.jpg    74539 (492:140-503:146)
drone-focused.jpg 1120411 (456:140-467:146)
drone-sharp.jpg   1136823 (552:406-563:412)

Here a much smaller box is defined (only 11x6) so the "best" focus box in
'-focused' is [456:140-467:146] but '-sharp' has a slightly better box. So the
blurred out bit was the background, so something in that background was in
slightly better focus than the drone. This points out a shortcomming of this
approach; a photograph of your child, where the bush on the right is in perfect
focus and their head is blurred is proably a "fail" but that's a human decision.

So you could envisage a workflow starting:

./checkfocus drone*.jpg | sort -nr -k2   # sorts overall sharpness decending
drone-sharp.jpg 320541267 74258203 246283064 13836176 3535124 10301052
drone-focused.jpg 66158661 25155372 41003289 405334 116660 288674
drone-blured.jpg 22357791 4864803 17492988 988237 268736 719501

$ ./checkfocus drone*.jpg | sort -nr -k5  # sorts centre box sharpness decending
drone-sharp.jpg 320541267 74258203 246283064 13836176 3535124 10301052
drone-blured.jpg 22357791 4864803 17492988 988237 268736 719501
drone-focused.jpg 66158661 25155372 41003289 405334 116660 288674

Here interestingly '-blurred' beats '-focused' . Turns out the drone is off
centre (outside box) so both are looking at the same out of focus area. However
'-focused' has been edited more often and saved at (very slightly) higher
compression so it's a 6% smaller file (it's slightly more blurred).


checkfocus:

Usage:   checkfocus [-v|--verbose]
                    [-d|--debug]
		    [-r|--red|-b|--blue|-g|--green]
		    [-B x:y-x:y|--box x:y-x:y]
		    [-f <filelistname>|--file <filelistname>]
		    <list of filenames>
		    
-v or --verbose produce more verbose output, can be repeated for more verbosity

-d or --debug produce debug output on stderr, can be repeated for more verbosity

-r|--red|-b|--blue|-g|--green

    base the calculations on reg/green or blue channels, default is green, ignored
    for greyscale

-B x:y-x:y|--box x:y-x-y,

   where 0,0 is top LH corner of image, x is column, y is row. Defines a smaller
   'box' to analyse'

[-f <filelistname>|--file <filelistname>]
    filelistname is a file which contains a list of files, one per line. These are
    processed before <list of filenames>


bestfocus:

Usage:   bestfocus [-v|--verbose]
       		   [-d|--debug]
		   [-r|--red|-b|--blue|-g|--green]
		   [-f <filelistname>|--file <filelistname>]
		   <list of filenames>

-v or --verbose produce more verbose output, can be repeated for more verbosity
-d or --debug produce debug output on stderr, can be repeated for more verbosity

-r|--red|-b|--blue|-g|--green

    base the calculations on reg/green or blue channels, default is green, ignored
    for greyscale
	
-f <filelistname>|--file <filelistname>

   filelistname is a file which contains a list of files, one per line. These are
   processed before <list of filenames>

-b <cols>:<rows>|--box <cols>:<rows>
-H <hframes> | -hframes <hframes>
-V <vframes> | -vframes <vframes>

The Image is divided into <hframe> Horizontal frames by <vframe> Vertical frames
A 2nd set of alternate frames is defined midways (horizontally & vertically) between frames
In each frame a focus box is defined <cols>:<rows>
The focus box with the highest contrast store is output

<hframes> defaults to: HFRAMES <vframes> defaults to: VFRAMES 
The box size defaults to the entire frame