This program lets you use a digital camera from your computer. It has a live preview window and a full set of manual controls.
RTIAcquire uses the libgphoto2 library to control the camera, so you must have this library on your system and your camera must be supported by libgphoto2. They have a list of supported models.
libgphoto2 does not currently work on Windows without some tinkering.
It can also control a lighting system for doing Reflectance Transform Imaging, though you'll need to customise it for your exact dome and lighting hardware. If you don't have a dome and lighting system, you can just use the program to take pictures.
The whole thing is in Python so it should be very easy to customise. There are about 600 lines of C for the live preview.
libgphoto2 does not work on Windows, so you need a unix-like system of some sort. We've used several Linuxes but OS X should work too. On Debian-family systems this package is called libgphoto2-2.
RTIAcquire uses IJG libjpeg to decompress preview frames. You must have the headers for this library available: jpeglib.h and jerror.h. On Debian-family systems this package is called libjpeg-dev.
The GUI is done in gtk2, so you need the python-gtk2 package.
RTIAcquire communicates with the lighting system over USB using python-serial. You need to have this package installed too.
The program as it starts up. The main area is a live preview running at about 20 fps. My laptop will go up to about 50 fps, but most cameras can't supply frames that quickly and anyway we don't want to flood the poor computer, so the frame rate is throttled. 20 fps seems fine for manual focussing.
If you mouse over the live preview, a 'pause' button appears allowing you to stop frame grabbing.
The camera is autodetected on startup.
The buttons along the bottom of the window let you control the dome and lighting system, set camera controls, take a single photo, take an RTI preview, and do a full RTI capture. If no dome is found on startup, the lighting controls and the RTI controls do not appear.
The window you get if you click the camera control button. This is generated by interrogating the camera for the controls it supports. The screenshot is for a Nikon D3X.
The controls along the bottom let you refresh the GUI from the camera (in case you change one of the camera controls on the camera body), switch between presets, add a preset, and remove a preset. Presets are remembered between sessions. A special preset called 'startup' records the state of the camera when the program was started.
During RTI capture.
- Install with something like:
$ python setup.py install --prefix=/my/install/prefix
And run with
Make quicklaunch links in the usual way for convenience etc.
- Run with
$ RTIAcquire --debug &> log
to produce a lot of debugging output in the file "log", handy for testing.
When you plug the camera in you may get a desktop camera icon -- right-click and select "unmount" so this program can open it.
Interrupting the program or unplugging the camera during an operation while the view preview is showing may leave the camera in a strange state and make it unresponsive. Unplug, turn the camera off and on again, and reconnect. Sometimes even that doesn't work, argh. Try pulling out the camera power cord if desperate.
Holding down the two green dot buttons on the lower back of the camera for two seconds resets the camera to a sane state.
With the Nikon D3X, put the lens into AF-S mode (the control on the front of the camera body, to the right of the lens as you face it). This will let you autofocus using the program controls, but not refocus before each shot.
In image preview, click Edit / Preferences and turn off image smoothing (smoothing makes checking focus difficult).
This software was funded by the UK Arts and Humanities Research Council "Reflectance Transformation Imaging System for Ancient Document Artefacts Projects" under the Digital Equipment and Database Enhancement for Impact Scheme. Details are available.
stop mousewheel changing combo boxes
managed this in nip2: intercept events in combo menu, block scrollwheel
in camera.py, try ignoring the return value from
gp.gp_camera_capture(), do we still get a sensible value for
cam_path? perhaps only fail if
if this works, we might be able to get rid of the preview() at the beginning of
break main out into a separate program
name private members with a leading
Major new features
how do we do light calibration? a separate program?
build and bundle the ptm fitter and viewer
see the OS X bundler, investigate prebuilt pygtk on OS X packages
Phil wrote a tiny gui wrapper for doing the fitting - so an extra Build button could be added to yours which runs his script (as the path is needed)
fitter notes here
they need to be able to batch fit
also, crop before fit to speed the process up
can we get the preview to reflect actual exposure? it seems to always do auto-expose
nope, doesn't seem to be possible
can we drive the camera focus ourselves
nope, doesn't seem to be possible, see:
but I'll watch for replies just in case