Brian,
The jpgs will lose additional info on each save.
The simple solution is to minimize the number of saves
in jpg format. My guess is that you don't have the option
of saving the original in something other than jpg, and
you want the final to be jpg (but, even using movie3,
you could save *.png with no additional lose). But if you
plan to do mouse adjustments from files, the intermediate files
should not be *.jpg. By the way, in theory I think it is possible
to do a lossless |. type rotate of a jpg that was compatible with
the jpg blocksize (often 16). However, the only versions given by
movie3 are as in rotate around the center type motions and managing this
would probably require significant effort at unwrapping jpg format.
Not to discourage you, but if you think about how many hours
you would need to create your mouse editor, and the equivalent $,
you could probably buy a much nicer camera with better movie
capabilities. Still, you have a very interesting project going.
By the way, I did not find it easy to identify the black
edges. It seems they are either antialaised or, most likely,
they show significant non-black artifacts from the jpg-ing.
Happy movie making,
Best,
Cliff
Brian Schott wrote:
Here is another take on guessing the frames; maybe
the guessing is adequate and the edge detection is
unnecessary if the "Edit" mode is enhanced. What I mean is,
if the guesses are kept crude and the editting windows are
designed to overlap the guessed frames a little, then maybe
the mouse drags can overcome the crudeness of the guess.
This is especially true if there is some control over
rotation of the individual frames (or maybe of all the
frames en masse), and in changing the size of the actual
frame viewing windows.
I anticipate problems with rotating .jpg's because
of image detail loss. Is the amount of loss proprotional to
the amount of rotation, or what?
Reactions?
On Thu, 26 Apr 2007, Brian Schott wrote:
+ Cliff,
+
+ Your suggestion is great. I do get a nice movie, but
+ the problem is that the guess of frames being 80 by 60 is
+ crude. I was thinking there was a way to detect the frames
+ using something like edge detection, especially with photos
+ that don't line up as well as this one. Even this one has
+ unequal width frames, but at least they are (almost?)
+ parallel to the edges of the whole picture.
+
+ The frames were recorded in the order 4 5 6 7 3 2 1
+ 0 , so a resequencing that even I can handle is required.
+
+ So, the edge detection is still a problem. And when
+ I have that working I want to be able to "Edit" the frames
+ with mouse drags to position each frame in its viewing
+ frame, because often the photographer moves the camera
+ between frames and the frames need realignment inside a
+ slightly smaller viewing frame.
+
+ Thanks to all,
+
+
(B=)
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
--
Clifford A. Reiter
Mathematics Department, Lafayette College
Easton, PA 18042 USA, 610-330-5277
http://www.lafayette.edu/~reiterc
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
