Blender for scientific purpose

As mentioned in the general forums, I’m reposting my question here:

Hi guys,

a friend of mine is biologist and we are trying to find a way to compare 3D geometry.

The purpose is to compare two 3D volumes against each other and find out which are the differences.

  • First goal would be a graphical approach were we would highlight the differences with different colors. eg. green means equal and red means differates
  • Second and long term goal would be a mathematical approach were every difference is calculated as a volume.
    We can not just take the two whole volumes and calculate it’s differences(in that case we would just need to compare the weight of the real-life counterparts).
    Because we are looking for differences in shape and not in the whole volume.

I’ve asked a few engineers so far who said it would be possible in commercial engineering applications such as CATIA. But they were not willing to pass any information along how to do it and a license for such an app is not affordable for us.

As background information:
The geometry is (re-)constructed out of hundreds of fine cutted prawn brain samples and extrapolated to one closed 3D volume. Purpose is too differate species by the differnce in shape of their brain. In the past this has been done by just looking at it under a microscope and simply “telling” the differences.

If anyone can come up with something like:
“Oh boy, simply connect material node x to y and there you have your graphical comparsion…”
or even
“Cmon dude, I write you a Python script while asleep and feeding my 6 headed python, that will compare that ‘crap’…”

…we’d be really grateful!

PS: If anything of this is hard to get… i’m not a native english tongue and will be happy to explain any further details.

In essence do you mean like using boolean operations?
e.g. intersection and difference?

Yes, like a boolean operation. But from my experience they won’t work very well in most applications and the reconstructed model will most likley contain non-manifold geometry. Although I’m not even sure what non-manifold exactly means(polygons with more than 4 verticies/edges, right?), I know that it is tricky to fix it.

http://wiki.blender.org/index.php/Doc:Manual/Modifiers/Mesh/Booleans
(right click & view docs on a Blender menu drive you to the online doumentation)
http://www.blender.org/documentation/blender_python_api_2_58_release/bpy.ops.object.html#bpy.ops.object.modifier_add

on manifold has to do with convex and concave faces!

and right now boolean is not a super tool yet in blender
but is the shape is simple it might work

can you gave some picture or even a file sample as example for this so we can have a look at the differences?

happy 2.5

there’s a build with a different boolean engine around, could try one that also…

and for getting a volume the simplest way i can think of is using blender grid particles to fill objects volume, see file for a simple example… got some limitations -max resolution- but easier than having to program your own volume calculator :slight_smile: see example here http://db.tt/sx4S8Kv

Do you have a sample dataset to work with?
I’d like to try to do sort of a ‘voxel boolean’ and see if that would help.

Thank you for the replys! I’ll give your tips a try as soon as I find time again(im busy freelancing now…).

I’ll ask how long it will take them to provide first samples. You know, first they have to take hundreads of pictures and then reconstruct them by drwaing curves on every single picture.

I’ll get back here as soon as I can.

If you’d like to have a look, here’s a test blend including the original dxf output:
http://file.ilikeviecher.com/TestBlend.7z

We do not have a second model for comparsion yet but I think this should make clear what kind of messy geometry we’re getting :-]
This is a work in progress reconstruction but the final output will most likley look very similar.

Edit: Blend is 2.57.1; imported with DXF importer v0.1.5