Thank
you all for your kind comments on my animation (Wojtek, Marshall, John, Brian,
Duke, Flash, MikeB).
Some
more information on how it was done.
As I
was saying animation needs precision: each position of the (virtual) camera must
be precisely computed to have a smooth movement.
For
making a stereo picture in Google Earth, you can move sideway with a few
keyboard strokes to make your stereo base. For an animation you will not be able
to repeat the same base for each frame.
Fortunately with GE you can record the current position
in a "Placemark" file containing all the information (.kml or .kmz file) and
Google is kind enough to provide full information on the format of these
files:
With
this you can design a simple program (or even an Excel spreadsheet) to compute
the positions for you and generate as many placemarks files as needed for the
video.
I
could have chosen a lazy approach by generation only a "circle" of pictures and
taking pairs a few degrees apart to make stereo pairs.
That
would have introduce convergence and I wanted to have "real" stereo there. In
fact it wasn't harder to do "real" stereo as in the KML file you specify the
"heading" of the camera. Two camera with the same heading and placed on a line
perpendicular to this heading will make a good stereo camera. Some will object
that only East and West headings will produce truly parallel lens axis while
North heading will produce a convergence around the North Pole! Anyway the size
of the scene compared to the size of the Earth itself has enabled me to make
further simplifications: spherical Earth with a 40 000km circumference, locally
flat Earth surface and meridians considered as parallel.
With
some sines, cosines and interpolation you can generates the needed KML files for
the animation with reasonable precision.
Well
now you have 2880 individual KML files that you need to load in turn in GE, wait
for the display to stabilize and be fully loaded and save the image in a new
JPEG file.
Making
animation also needs perseverance! But nobody will found the task of manually
doing this 2880 times enjoyable!
But
the computer can do it for you. I've found a marvelous little freeware called
AutoIt3. Its main purpose is to send keystrokes and mouse clicks to other
applications to mimic a human interaction.
AutoIt
is much more than just this, in my case it's a AutoIt script that is doing all
the cosine things, create all the KML files and remote control the GE
application.
For
the later you simply:
- send
Ctrl-O keystrokes to GE to display the "Open" dialog box
- send
the name of KML file #i
- send
Enter key
- wait
a specified delay for the image to stabilize (about 10 s)
- send
Ctrl-Alt-S to open the "Save" dialog box
- send
the name of the JPEG file #i to save in
- send
Enter key
- loop for frame i+1
And of
course do the same thing for the right viewpoint.
It
didn't took me long to write the first working script (about 2 evenings) but it
was still needing improvement to send the keystrokes at the right moment and not
losing synch with GE.
One
problem is that you cannot know when GE has fully loaded the current image, it
may depend on your internet access for instance. But small increment in position
(here a quarter of degree rotation) means less changes in the picture and less
to download from the internet so I've settled to a 10s wait for each frame
(that's already 8 hours for 2880 pictures!) and with the other operations the
computer is busy during 20 to 24 hours. Google Earth is running full screen to
have the largest image : 1600x1175 on my 1600x1200 monitor.
A
little more time four mounting the stereo pairs in X views (a simple 100 pixel
horizontal shift was applied to place the window) and build the
video.
Here
I'm using the Adobe Flash 8 movie format as it is widely used on the internet
(YouTube for instance). The On2 VP6 codec used is not perfect for anaglyph (I'm
waiting for Flash 9 with h.264 compression) but is doing a decent
job.
Pierre
Meindre.
Wonderful.
Mike Beech
Pierre MEINDRE
<pierre.meindre(-at-)free.fr> wrote:
Hello,
Making stereo pictures with Google Earth is quite easy
but making movies is a
little more challenging.
To make a movie
that will play smoothly you need a lot of pictures (here 24 fps)
and for
the video not to be jerky, the positions (in the 3D space) of
each
picture must be computed precisely.
Well that's what computers
were invented for: perform complex and tedious tasks
while you
sleep!
Here is a circular view around Mt Saint-Helens, composed of
2880 individual
pictures, 1440 stereo pairs (rotation by quarter of
degree) for a one minute
animation. Stereo base is 650
m.
Anaglyph:
http://www.stereoscopie.fr/?79e3ceCrossed-eye
:
http://www.stereoscopie.fr/?f6c45cNote:
Videos need Flash plugins v8+ installed in your web browser.
Pierre
Meindre.
Be a better Heartthrob.
Get
better relationship answers from someone who knows.
Yahoo! Answers -
Check it out.
