On Sat, 2010-10-23 at 00:09 +0100, Ognjen Bezanov wrote:
> #>> pano_modify --center --fov=AUTO --canvas=AUTO --crop=AUTO
> --projection=0 -o project.pto ./project.pto
> 
> Setting projection to Rectilinear
> Center panorama
> Fit panorama field of view to best size
> Setting field of view to 179 x 178

I think this should be lower for scans.

> Caluclate optimal size of panorama
> Setting canvas size to 2086957 x 1043374

This should definitely be much lower.

> How much memory do I need? I have 3GB on my box atm (Running Gentoo
> Linux 2.6.31).

That should be more than enough.

> And why is the image so large? Each scan is 1578x935 and there are 15 of
> them. Assuming they are put end to end (which is not the case, as there
> is like 70% overlap between them) we'd get a maximum size of 1578x14025
>  This is far less than "2086957x1043374". Where on earth is that number
> coming from?

I think your optimised panorama is wrong if this is happening. You need
to optimise roll, x shift and y shift for each image to align scans. The
the yaw and pitch should remain 0 degrees. Since you have set the
images' field of view to 10 degrees, the field of view of the panorama
should therefore be much less than 179 by 178 degrees. The huge size
comes from having images spread out too far.

You'll need to set the right variables to optimise in the pto file, then
run autoptimiser with -n instead of -a.

In PTO file, a line beginning with v indicates variables to optimise.
There is a letter code for the variable, followed by a number for the
image. Images are numbered from 0. The code for roll is r, d and e are x
and y shift respectively, v is the field of view. This is useful if the
scans are different sizes.
So "v d3" indicates the x shift of the forth image should be optimised.
Don't optimise any variables for one of the images, this image will
remain fixed so the other images can be positioned around it.

It also seems automatically setting the field of view ignores the
shifted images, so they don't fit in the output. (Bug?)
Autocrop finds a reasonable crop, but takes longer. So I would set a
slightly too large field of view then autocrop (assuming the first image
has a reasonable rotation).

The attached bash script should stitch scan images.

-James

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Attachment: scan_stitch.sh
Description: application/shellscript

Reply via email to