-------- Original Message --------
Subject: Answers to from Photos to 3D
Date: Mon, 14 Feb 2011 11:16:09 -0500
From: Abby Drake <[email protected]>
To: [email protected]
Hi Everyone -
Awhile back I asked about converting 2D coordinates from photos into 3D
coordinates. I received a few different solutions which I summarize
below. I am testing the conversion of 2D coordinates from photos to 3D
coordinates using a rotation algorithm that matches homologous points in
different views. When I've tested this thoroughly I would be happy to
provide the template to others looking to do the same.
In the meantime here are the various answers I recieved. Thank you all
for your suggestions - they are greatly appreciated!
Best Wishes,
Abby
_Using rotation via homologous landmarks in separate views:_
From Andrea Cardini and /Carlo Meloro/
( I have the PDFs for both of these if anyone wants them)
Corti M, Fadda C, Simson Sh, Nevo E. 1996. Size and shape variation in
the mandible of the fossorial rodent /Spalax ehrenbergi/: a Procrustes
analysis of three dimensions. In: MarcusLF, CortiM, LoyA, NaylorGJP,
SliceDE, eds. Advances in morphometrics. New York, NY: Plenum Press,
303–320.
Fadda, C.; Faggiani, F.; Corti, M., 1997: A portable device for three
dimensional landmark collection on skeletal elements. Mammalia
61, 622±627.
_Using Excel:_
From: Patricia Pilatti Alves
Hi Abby,
I've been using 2d photo to capture 3D coordinates for landmarks...
Actually I use two photos, one in dorsal view and another in lateral
view (of a mammal skull, for example). And then, I just add the "y"
coordinates of the lateral view's landmarks set as "z" coordinates of
the dorsal view. Very simple.
But, before that, the position and orientation of the photos must be
standardized.
I do a translation of the landmarks set (to put the first landmark on
the origin - (0;0) of the cartesian plane) and after a rotation of the
landmarks set (to put the second lardmark on the "x" axis).
All of this I do at Excel.
If you have already had an answer, that's ok. If I can help you more,
just let me know.
Best,
Patricia.
_Software Packages: Photomodeler & Blender_
From Leandro Monteiro:
I have used photomodeler (http://www.photomodeler.com
<http://www.photomodeler.com/>) in the past.
They used to have a student version that was free, but that doesn't
seem to be available any more. I supervised a PhD student that
captured 3D landmarks for ~500 rodent skulls using this software, and
it worked quite well (time consuming though). Hope this helps, good
luck.
From Rudyard W. Sadlei:
Hi Abby,
I've used Blender to do this with some crocodylian skulls. However, I
haven't had
the time to fully explore the extent of any methodological artifacts and
how they
would manifest in a geometric morphometric analysis.
From what I've done, you will have to be very careful with the initial
photographic
set-up and capture images from orthogonal planes around the specimen at
the same
scale in each image. Ideally, there should be 6 images, but I've done
alright with 2
images of developing chick limb buds. From that you can sculpt the 3D
object using
the images as a background template in Blender (it is freeware, very
powerful, and
with a large online community to help if you hit a wall). Once you're
done, export
the 3D file in a compatible format for something like Landmark, and you
good to go.
If you're a glutton for punishment, then you can manually transcribe the
landmark
coordinate data from Blender to Excel using a surface node as a proxy
for the
landmark position.
Blender has a steep learning curve if you're new to it, so let me know
if you might
like a short intro to get you started. I'll let you know if this becomes
a proper
functional method, where I get solid data out of it.
Again, I've not had the time to run a proper validation analysis (e.g.
replicate
reconstructions from different individuals from same and different
species) to
determine the relative variance in the method at biological scales.
However, the 3D
model I sculpted clustered with the same species group in a broad-brush
PCA for the
crown group crocodylians using 30-40 landmarks.
Cheers,
Rudd
--
Abby Grace Drake, PhD
Biology Department
College of the Holy Cross
Phone: 508.981.2783
Skype: abby.drake
Website: Aves3D.org
