fnv / mousetrap Goto Github PK

While the color picker saves information to the config file, it doesn't actually communicate with the algorithm. Suggested implementation: when a hue is selected, the program tracks a range of color (5 or 10 degrees on either side) centered on that hue.

Apparently it's possible for the application to start without writing a proper section to the cfg for color tracking. There's one routine that does this on startup. We must debug it.

When is it appropriate to call the sync() method in the Capture class, if ever? Why do we call it twice in get_image() in color.py? Is there a better way to do this? If we do not call sync(), the frame does not get initialized.

Don't forget to turn that function into valid python. fnv has the source. It is also defined in gtkhsv.c in newer versions of gtk as rgb_to_hsv().

We need to get this twin backend thing working so we can use both algorithms at once.

Perhaps using debug window?

Allow users to set the range considered by the color picker in the config menu.

Turning tracking on and off works, but after tracking has happened successfully once the mouse gets locked into its last coordinate. I suspect this has something to do with calls to get_pointer, but calling the remove("obj_center") method didn't work either.

This either requires multiple color pickers, or an add/remove color-to-track interface.

Currently the color picker sets the hue range when mousetrap starts, so any hue saved to the config file will not take effect unless the user restarts mousetrap. Obviously this is not cool.

Go, Ryan, go!

The color algorithm option in Mousetrap produces very choppy mouse movements.

This means more config information. Basically this allows people to control "mouse speed".

When the color picker is opened, and then closed again, the window stays expanded and looks foolish.

Currently the keyboard input only happens in the get_image method, which is called every .15 seconds. This isn't actually fast enough to have responsive keyboard input. Is there a way to do it that won't have a bit of lag?

Currently set_opencv is called on every mouse point after the first one, from the color.py file. Is this the appropriate time for it?

Currently, in get_image() in color.py, we initialize the IplImage objects only when a boolean flag (first_time) is set to true. There is probably a better way to do this. How?

Currently, a user must select an area to track in the Mask window, and the selection box seems to move the opposite direction of your mouse. This is very bad.

For example, "--idm=color" and "--idm=forehead".

Perhaps through the preferences menu and utilizing the config file!

Should framerate and pointer-update rate be configurable? Why are the defaults set at 150/50? It seems that tracking might be smoothable by tweaking this options.

It needs some refactoring / reformatting. Pleeeze. @_@