How to visualise 10000 cell volume

Question

Pierre Lebel 2019-9-25

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/482055-how-to-visualise-10000-cell-volume

评论： Pierre Lebel 2019-10-11

I have tried to view this system in 3D using patch. It is very slow. I've seen other software do this much much faster, even with transparency. What am i missing?

2 个评论
显示无隐藏无

Fabio Freschi 2019-9-25

Any data? I visualize easily millions of patches but it is difficult to help without a simple example of the data

Jan 2019-9-25

Please explain any details. How are the "10000 cell volume" represented? If you post code and data, it is likely, that the speed can be improved.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Fabio Freschi 2019-9-25

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/482055-how-to-visualise-10000-cell-volume#answer_393485

在 MATLAB Online 中打开

Load nodes and faces in a unique structure and plot all patches in a single command

tic
nodes = []; faces = [];
for i = 1:length(PatchWork)
    faces = [faces; PatchWork{i}.FaceVertices+size(nodes,1)];
    nodes = [nodes; PatchWork{i}.VertexXYZs.'];
end
figure
patch('Faces',double(faces),'Vertices',nodes,'FaceColor','red');
view([1 1 1]);
toc
>> Elapsed time is 1.018881 seconds.

11 个评论
显示 9更早的评论隐藏 9更早的评论

Fabio Freschi 2019-10-11

编辑：Fabio Freschi 2019-10-11

在 MATLAB Online 中打开

From the documentation of patch

'flat' — Use a different transparency for each face based on the values in the FaceVertexAlphaData property. First you must specify the FaceVertexAlphaData property as a vector containing one transparency value per face or vertex. The transparency value at the first vertex determines the transparency for the entire face.

So FaceVertexAlphaData is a node-based property. This can complicate what you want to do, because if the face inherited the transparency from its first node.

I don't know if there are alternatives, I would plot independently all faces that share the same transparency. For example, let's assume that the first half of the faces have FaceAlpha = 0.2 and the last half FaceAlpha = 1. there are 60000 faces in your example, so:

nodes = []; faces = [];
for i = 1:length(PatchWork)
    faces = [faces; PatchWork{i}.FaceVertices+size(nodes,1)];
    nodes = [nodes; PatchWork{i}.VertexXYZs.'];
end
figure, hold on
patch('Faces',double(faces(1:30000,:)),'Vertices',nodes,'FaceColor','b','FaceAlpha',0.2);
patch('Faces',double(faces(30001:end,:)),'Vertices',nodes,'FaceColor','r','FaceAlpha',1);
view([1 1 1]);

Hope it helps

Fabio Freschi 2019-10-11

The point is: if there are duplicated nodes, so that each face has its own set of nodes, so you won't have problems in using 'FaceVertexAlphaData'.

There is still the fact that you must address the specific property to the right node. and this may not be simple if your data are generated by an unknown software/function (like an independent mesh generator). I actually haven't check your data in details, but you may know what you really have as input.

I usually work with non-duplicated nodes, like in finite element meshes. In this case, for example, one of your corner nodes is hsared by three faces. Suppose that this corner node is the first node for all 3 faces: so you cannod specify different transparencies for these faces.

Pierre Lebel 2019-10-11

Thanks again,

Those are several good points.

The first is that in my case, there are 10000 cells each with eight vertices and six faces. Vertices are [80000x3 double] and so, are independent for each cell. Faces are [60000x4] and point to the vertices. Each vertex is used three times (3-faces) and only in one cell.
Color and transparency are both assigned by face. The colors appear to be assigned correctly.
Your last point is good, but because I use duplicated nodes, not really applicable.

I think I should ask Mathworks about the 'FaceVertexAlphaData' feature, but your approach (above) will certainly work, although not quite as elegant or versatile as I would like.

请先登录，再进行评论。