mirror of git://repoz.nekodune.net/exposit.git
C++ C Makefile Objective-C
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
AUTHORS
ChangeLog
Makefile
NEWS
README
TODO
chunkio.cpp
chunkio.h
draw.c
draw.h
exposit.cpp
gp_imagergbl.cpp
gp_imagergbl.h
graphutils.c
graphutils.h
interact.cpp
interact.h
jeuchar.c
jeuchar.h
show.cpp
show.h
simplechrono.cpp
simplechrono.h
starsmap.cpp
starsmap.h
testchunkio.cpp
vstar.cpp
vstar.h

README

Notes for installing / testing exposit

- early-beta-0.03

    at this stage exposit can
     - take a list of jpg to be processed (from the command line only)
     - take a reference jpg for correcting fall-off (or related aberations)
       (command-line only)
     - propose an initial noise-cut and gain in order to reveal nebulous objects
       on the first frame (interactive)
     - display 
	- the intermediate resulting stack with current cut-off and gain proposition
	- the current frame with same cut-off and gain (corrected in regard of dynamic)
        - a customizable zoom on some part of the intermediate stack
        - the orginal last frame (without gain or cut-off)
     - use the first frame as a stacking reference for the following frames
     - find some vectorial correlation and apply a rapid translation+rotation
       on each frame according to the reference frame
     - try within a set of rotation/translation shift which combination gives
       the supposely best result at aligning each following frames
     - offer a rough noise level-cutoff and gain adjustment before saving to a png
     - quit


    exposit is slowly evolving from a command-line only to
    a user-interactive application. yet most things are
    command-line only, and it cannot be used without some
    mandatory command-line inputs :

	-debug=X	sets the debug level (0 none)

	-dcraw="full_command_line_for_dcraw"
			tune the invoking command for dcraw (special path for
			testing a dedicated/new release of dcraw, etc ...)
			defaults to "dcraw -c -4"

	-adddcrawmatch="ext1 [... extn ...]"
			add file-extensions matching for dcraw decoding. useful
			for handling raw files with esoteric extensions.
			the default list only has "nef NEF cr2 CR2"

	-load-unbayer
			turns on un-bayer algorithms for subsequent images load.
			computed luminance is used for extracting RGB from a
			bayer-filtered mono pic.
	
	-load-normal	resumes normal behaviour, color or monochrome are treated
			as-is (opposite of -load-unbayer)

	-falloff=falloffpic
			sets the fall-off reference pic. this pic is supposed
			to contain ani, as optimal as it can, pic showing the
			brightness variation one can observe from within one part
			of a frame to another with a given optical combination.
			each frame will be leverage in order to flatten those
			variation, it might increase noise at some part of the
			stack, but will help at corner fall-off  correction.
			some would refer this pic as "white(s)"
			I got mine by shooting some random parts of a great blue
			sky with the same aperture settings as used for astro, and
			mixing them all (in gimp) in order to increase uniformity.

	-finetune	enables trying a set of rotation/shift around the vectorial
			alignement proposed in order to find a better alignment result.
			long, and most of the time useless. mostly a remain of older
			algorithms.

	-doublescale	uggly double-sampling mode
			try to double sample (4 times more memory) each frame
			before stacking. one could expect some subsample
			alignment could be detected this way, but a better
			resampling is needed at the first place.

	-noise=noisepic	add a noise reference pic (blacks) in the noise pic list.
			the noise reference would be substracted to each frame.
			noise removal did not give any good result this way yet.

	-watch=[dirname/]fileregexp
			a directory path name + file regexp match combination in order
			to wait for pic to appear in the said directory and stack them
			on the fly. the purpose is to stack pics as soon as they are
			shot. real field use as proven it's difficult to use because
			the first pic must be the reference, so it has to be rather
			clean, and usually the first shot is never clean enough ...
			a much better user interface is needed here !
			(not tested for a while, might be broken)

	interactive keys
	    the mouse is useless, for instance.
	    arrows should navigate the menu, and escape go back to previous menu.
	    enter/space enters/apply the functions.

	    now, the menu part isn't finished, so most function are still simply bound
	    to keys :
		U I     fast decrease/increase the noise cut-off
		H J	slow decrease/increase the noise cut-off

		O P	fast decrease/increase the upper cutoff level
		K L	slow decrease/increase the upper cutoff level

		G	(go !) toggles to the non interactive mode, all subsequent
			frame additions will be performed without waiting for the
			user to press "add frame" until no more pic are available. 

		S	saves a png of the stack with gain/cutoff applied

    dependencies :
	on debian system those package are needed for the build
	    libsdl-image1.2-dev
	    libsdl1.2-dev
	equivalent packages should exist on most distros


    example of use :
	exposit -falloff=falloff_180mmf2.8.png capt0166.jpg *.jpg	
			  white                 ref          pics

	press "G" then "add frame" the progam will add all pic
	to the stack and finally wait for final level adjustments
	"S" will save a png at the end