[WIP] Generating of TVB compatible datasets #38
Conversation
| # A nicer solution would be to detect degenerate triangles and ignore their | ||
| # contribution to the vertex normal | ||
| vert_norms[k, :] = vertices[k] / np.sqrt(vertices[k].dot(vertices[k])) | ||
| bad_normal_count += 1 |
There was a problem hiding this comment.
We shouldn't, in principle, have to handle this case. Did you encounter it?
There was a problem hiding this comment.
No, but I haven't even run it yet. The computations are mostly Tim's code that I've left as it was without changes
|
|
||
| region_surface_area = np.zeros(regions.size, 1) | ||
| avt = np.array(vertex_triangles) | ||
| # NOTE: Slightly overestimates as it counts overlapping border triangles, |
There was a problem hiding this comment.
I wonder how FS handles that? I guess their triangles are so small, they can exclude borders and be fine.
There was a problem hiding this comment.
You can quite easily scale the triangle contribution to the region area by the proportion of the vertices belonging to the region. Is it worth it? I don't know.
| areas[regions_subcort] = areas_subcort | ||
| areas[regions_cort] = areas_cort | ||
|
|
||
| dataset = StructuralDataset(orientations, areas) |
There was a problem hiding this comment.
should we be doing something with this dataset object?
There was a problem hiding this comment.
Yeah, it is WIP for a reason. But there's StructuralDataset.save_to_txt_zip() to be used eventually.
There was a problem hiding this comment.
👍 I'm brain dead today and will refrain from further comments
|
Looks like a good start; do you have more commits on the way, or would you like me to get this finished? |
|
Same as in #33 - works on its own, but it is not integrated. |
|
Outdated. Functionaly had been integrated already with similar code |
Issue #37.