** Big speed boost courtesy of Oliver Woodford. Recommended to update to latest version **
stlwrite(FILE, FV) writes a stereolithography (STL) file to FILE for a
triangulated patch defined by FV (a structure with fields 'vertices'
stlwrite(FILE, FACES, VERTICES) takes faces and vertices separately,
rather than in an FV struct
stlwrite(FILE, X, Y, Z) creates an STL file from surface data in X, Y,
and Z. stlwrite triangulates this gridded data into a triangulated
surface using triangulation options specified below. X, Y and Z can be
two-dimensional arrays with the same size. If X and Y are vectors with
length equal to SIZE(Z,2) and SIZE(Z,1), respectively, they are passed
through MESHGRID to create gridded data. If X or Y are scalar values,
they are used to specify the X and Y spacing between grid points.
stlwrite(...,'PropertyName',VALUE,'PropertyName',VALUE,...) writes an
STL file using the following property values:
MODE - File is written using 'binary' (default) or 'ascii'.
TITLE - Header text (max 80 chars) written to the STL file.
TRIANGULATION - When used with gridded data, TRIANGULATION is either:
'delaunay' - (default) Delaunay triangulation of X, Y
'f' - Forward slash division of grid quads
'b' - Back slash division of quadrilaterals
'x' - Cross division of quadrilaterals
Note that 'f', 'b', or 't' triangulations now use an
inbuilt version of FEX entry 28327, "mesh2tri".
FACECOLOR - Single colour (1-by-3) or one-colour-per-face (N-by-3)
vector of RGB colours, for face/vertex input. RGB range
is 5 bits (0:31), stored in VisCAM/SolidView format
% Write binary STL from face/vertex data
tmpvol = false(20,20,20); % Empty voxel volume
tmpvol(8:12,8:12,5:15) = 1; % Turn some voxels on
fv = isosurface(~tmpvol, 0.5); % Make patch w. faces "out"
stlwrite('test.stl',fv) % Save to binary .stl
% Write ascii STL from gridded data
[X,Y] = deal(1:40); % Create grid reference
Z = peaks(40); % Create grid height
% Write binary STL with coloured faces
cVals = fv.vertices(fv.faces(:,1),3); % Colour by Z height.
cLims = [min(cVals) max(cVals)]; % Transform height values
nCols = 255; cMap = jet(nCols); % onto an 8-bit colour map
fColsDbl = interp1(linspace(cLims(1),cLims(2),nCols),cMap,cVals);
fCols8bit = fColsDbl*255; % Pass cols in 8bit (0-255) RGB triplets
The third argument of reshape in line 81 should be size(faces,2) instead of 3 in order to account for a variable number of faces per patch surface. Other than that, good job.
Did the trick for me!
Just two tips I find useful for beginners (like me), and I think would be helpful if added to the documentation:
- You can create de faces-vertices structure from a volumetric 3D array, say V, with your data simply by:
fv = isosurface(X, Y, Z, V);
where X, Y, Z are vectors with the dimension of your volumetric array.
- The string argument 'FILE' or 'filename' (whatever you may call it) MUST INCLUDE the '.stl' extension, otherwise you'll create a file that is not that format. Lost five minutes figuring that out lol, thought it would be added by the script.
nevermind, I found an easy solution to flipping the normals. say you used :
stlwrite([name,'.stl'],con+1,pts,'mode','ascii'); % exports mirror information to STL format
con = [con(:,1),con(:,3),con(:,2)];
this will walk the other direction around each triangle flipping the normal for you.
is there a convenient way to flip the normals as an option. Sometimes I'll build a symmetric region of the mesh, transform (X,Y,Z) to (X,-Y,Z) and the inside will become the outside.
This worked very well for me! Is there a simple way to densify the model? The shape should be the same, but with more triangles/vertices/faces.
I have been able to tweak the permissions command to use to append ascii type stl file. But the same doesn't seem to work with binary format files. Please advise.
Thank you! Works well with Adam A's "STL to ACIS SAT conversion" for facet to solid conversion for many CAD programs.
@Daschulka, yes, an STL file has triangles and faces, so you will need to turn your point cloud into a mesh in order to make an STL file. The solution to do this always depends on your particular problem - some simple shapes can be meshed using Delaunay triangulation but it will all be dependent on the point clouds you have and the meshes you want to make.
I have been looking for a solution for receiving open surfaces after using stlwrite, like it is the case for Alexandra, but could not find a solution in the comments.
What is the way to resolve it? My input data is a point cloud... does it need to be turned into a mesh first? If that is the case, which functions have you been using to resolve the issues?
Many Thanks for the help!
just what i needed to be able to 3d print from matlab surfaces. at least if the object has a well-defined interior. now to find a good solidifier... still waiting on that one.
Great job, very useful!
I was able to figure it out. Your codes work beautifully together. I doubt I could have done this better in any other way.
I've gotten your code to work for my 2D image--making it a surf() and then exporting fully into SolidWorks. I'm wanting to add thickness or full volume to it as in your surf2solid script.
Is there a way to take the surf2solid 'Block' and transfer it to an .stl with stlwrite()?
Thank you for your help! This code is awesome.
Thank you for your help! But it's still creating something like a basis under the solid, like a sheet.
It's difficult to troubleshoot without being able to reproduce your error, but here's a way to take a 2D image and produce an STL of its surf() result. Does it give you a clue on how to fix your issue?
I = imread('rice.png');
I2 = conv2(double(I),ones(5)/5,'valid'); % Make it a little smoother
sH = surf(I2);
% Get XData and YData as grids (not vectors)
[XD,YD] = meshgrid(sH.XData,sH.YData);
Hi Sven, thanks for your really good work!
I'm having some problems with my stl, can you help me?
I have a image, I'm reading it and transforming into a double image. Then I'm selecting one of the 3(RGB) slices of the matrix to surf it with the function surface(), which give me the
propertys XData, YData and YData to use in your writestl(), but it's creating something like a basis under the solid, like a sheet and when I open this stl on the CubePro software, to print, I receive a bad file message.
Can you help me? Am I doing something wrong?
I creat a struct and solve it,Thank you for your code.
I have a vertex p matrix:8x3
and a triangualtion t matrix:12x3
I used stlwrite('test.stl',t,p),but I received:
Attempted to access facecolor(:,1); index out of bounds because size(facecolor)=[0,0].
Error in stlwrite (line 133)
c0 = bitor(bitshift(bitand(2^6-1, facecolor(:,1)),10),c0);
I was confused on how to use your command.Thank you
Following Will's comment (29 Mar 2015), I too am still getting an apparently incorrect message if I write out in ASCII mode: I consistently get "Wrote 4 facets". When I write out in binary mode, I get correct values: e.g. 149194 facets or 205532 facets, verified by checking the number of faces I provided as input. Besides this minor glitch, the ASCII file seems to be okay so far.
Thanks for the contribution.
Hi Benjamin, I suspect your issue is simply a matter of which reader you are using. If your software doesn't interpret the colour format chosen (see the help contents) then it will look like nothing has changed.
See for example the comments to this file below around September 2013 - others have found some software (Meshlab and Paraview) doesn't interpret this version of colour, while other software (VisCam) does.
The example you give using FaceColor does not seem to work. Adding face colors produces the same output as without colors.
Great work, helped a lot
Hi Tyron, sorry for the slow reply. I think that most solid meshing will be done downstream in a mesher - I would expect that Abacus itself has a "tetra-mesh" function that can fill a closed STL volume with solid elements.
Hi Sven, Thank you this works a treat!
I've combined several patches to form an enclosed 3D object and exported that to *.stl using your function. How could I "solidify" this such that it could be meshed in Abaqus for example?
Simple and nice.
Could u teach me how to use it？I am so stupid.
@nithish: Please formulate your question in a way that describes clearly what problem you have.
Most important steps:
1. Get your data into two variables:
"vertices" (an nx3 array of xyz coordinates. If you have 10 coordinates, it will be a 10x3 array)
"faces" (a px3 array of integer numbers (1, 2, 3, 4, etc) for your faces. If points 5, 6, and 3 form your first triangle, then the first row of this variable will be [5 6 3; ...])
2. Run the command:
If you can't get these two steps to work, then stlwrite will not work for you either, so it will not be helpful for you to try and work out why you can't write an stl file.
Here's how I would prepare a cylinder for stlwrite (or the patch() command). Note that I'm using the surfToMesh() and triangulateFaces() functions from David Legland's excellent geom3d file exchange entry:
[X,Y,Z] = cylinder(1,10);
[vertices, faces4] = surfToMesh(X,Y,Z,'xPeriodic', true);
faces3 = triangulateFaces(faces4);
Thank you for the code. But What i dont dont understand is how can i pass My data which is stored in array into stl ?
Hello and first, thank you for your contribution !
I have a little problem with my export. A lot of points in my figure have the same x and y coordinates (but differents z), because of the cylinder-like shape. So the export using the coordinates for arguments doesn't give me an appropriate STL file. I'm currently searching a way to create a triangulated patch of my geometry as you mentionned it in the program, but I really don't know how. Probably you know how to ..? Thanks a lot again for your time and your job.
I stored my text data into an array. But I dont understand how to give my array as input. Can please someone help me
Thx for the answer.
I have tried to display the data with patch respectively the surf command of matlab but sadly I don't get any result.
But would it be possible otherwise to create a stlfile by using the coordinates of the triangles computed in the connectivitylist of the command convexhull after a triangulation?
The help of read_vtk says:
% [vertex,face] = read_vtk(filename, verbose);
% 'vertex' is a 'nb.vert x 3' array specifying the position of the vertices.
% 'face' is a 'nb.face x 3' array specifying the connectivity of the mesh.
The help of stlwrite says:
% stlwrite(FILE, FACES, VERTICES) takes faces and vertices separately
So why not just the following?
[vertex,faces] = read_vtk('file.vtk');
This is direct. I don't understand what you're trying to do with MX, MY, MZ.
Here's a hint: if you cannot send your faces/vertices directly to MATLAB's patch() command (or your X,Y,Z data directly to MATLAB's surf() command) and get what you want on the screen, then using stlwrite() won't give what you want either.
[vertex,faces] = read_vtk('file.vtk');
MX(1:length(vertex), 1)=vertex(1, 1:length(vertex));
MY(1:length(vertex), 1)=vertex(2, 1:length(vertex));
MZ(1:length(vertex), 1)=vertex(3, 1:length(vertex));
Basically I am sending the vertexes information of my objects, which is a simple cylinder. But the resulting object as stl-file isn't a closed volume anymore.
This is the result
@Will: Ah, yes, that print error was annoying me too :)
I had it fixed on my machine, but the fex entry has been updated now as well.
@Matthias: You need to provide some more details such as the command(s) you run when calling stlwrite, the variables you are sending to stlwrite, and what goes wrong (an error message?)
Thanks for the great utility. When I run this, I always get the message "Wrote 1 facet" when it finishes. Perhaps I don't understand the terminology, but I expected the number of facets to be equal to the number of triangles being written, which I think is the second dimension of the facets matrix. (size(facets, 2))
Hallo, I am having trouble to create stlfile, i am extrakting the vertices and faces from a vtk-file using read_vtk but i dont know how to commit the data correctly to the stlwrite function? Any suggestions?
I am gratefull for every help trx
@ye: I need more information to be able to help you there. Perhaps send your data and command you're using to call stlwrite.
@Rebecca: There is no warning like that emitted from stlwrite itself. I believe you may be specifying that you want "triangulation" of XY data when that isn't the actual style of data you are providing.
It sounds like you have full 3D data rather than gridded XY data, so triangulating it based on its XY coordinate will not make sense due to the fact that you have *multiple* points that are above each other on your cylinder at the same XY coordinate. The warning you see is therefore spat out by the "delaunay" function (which does the triangulation).
Maybe you have quadrilaterals instead of triangles so you're trying to split them? If so, describe (or send) the data that you have, and there's probably a nice way to do that easily.
i'm writing a cylinder to an stl file, but stlwrite keeps eliminating repeating data points, which results in a simple circle instead of a 3D image. how can i get around this?
"Warning: Duplicate data points have been detected and removed.
Some point indices will not be referenced by the triangulation. "
I use this to get a stl file, but when i use solidworks to open it, that become a plan not the 3D picture i polted in matlab, i don't know why,if anyone can help me?
Very simple, clean and fast implementation!
Great work! And binary as well, i like it.
As a minor it would be cool to be able to specify the normals as well. (not that i couldn't do this myself, but it would save a minute ;) )
I Might you explain a little detail how to perform the conversion of stl right through points. I have 22050 points (xyz) belonging to the riverbed. My task is to generate an STL file.
In what format should I take the points?
As I run the compile stlwrite ?. They could give me the steps?
Please, could you help me.
Thank you Sven,
Works out great
@Urlea: Yes, you can do this when (or after) you generate the faces/vertices.
If you are using the isosurface() function to generate your faces/vertices from a logical, remember that isosurface() will *always* produce faces pointing from low to high. In that case, you can just invert your logical mask (to have ones outside and zeros inside) when you call isosurface.
To flip faces after you've got faces vertices, just use the following:
fv.faces = fv.faces(:,[2 1 3])
I am trying to Write binary STL from face/vertex data file with the example you give in your code. I then import the file generated in Magics (Materialise) and use an automatic tool to generate a diagnostig of the body. It says that all the faces are inverted (normals pointing inside out). Is there a way to fix this inside the Matlab code ?
Thanks so much!
Excellent job. I use it for a complex geometry which comes out of images. I created the .stl file but many surfaces are open and not connected. I want to have close surfaces with volume.
Thanks for your comments and help.
@Sven: I have done both your tricks. I found the problem is not your code; it is the software I am using to view meshes. I tried Meshlab and ParaView. None of them works for color surface. Probably they work for the other interpretation. VisCam works well for this. Thanks!
@Ben: Here's an example that uses the VisCam colour specification:
tmpvol = true(20,20,20); % Zeros on the inside means
tmpvol(8:12,8:12,5:15) = 0; % isosurface makes triangles
fv = isosurface(tmpvol, 0.5); % point out of object
faceColVal = fv.vertices(fv.faces(:,1),3); % Colour by Z height
cRange = [min(faceColVal) max(faceColVal)];
nCols = 255;
colMap = jet(nCols);
faceColsDbl = interp1(linspace(cRange(1),cRange(2),nCols),colMap, faceColVal);
faceCols8bit = faceColsDbl*255;
I used the free viewer here http://viscam-view.software.informer.com/download/, and the colours show up nicely.
You'll notice two tricks in my code above.
1. Make sure your face normals are pointing outwards. VisCam only displays the outer sides with colour and the inner sides without.
2. Make sure you provide 8-bit scaled colours (ie, between 0 and 255 rather than MATLAB's 0 and 1).
Did that work out for you? I might include that as an example in the help as long as it still works for you. Note that there are two competing "interpretations" of colour in an STL file. This is just one of them so I hope your software can interpret it well.
Can anyone give an example of writing a stl file with per face color? I tried several times but did not make it.
Very nice code and amazing fast! Thanks for sharing!
@Sven: Thanks. It worked like a charm! It didn't work when I tried to connect it with using triangulation in stlwrite but when I used used mesh2tri to generate the faces and vertices and then used those in stlwrite, it worked! Thanks a lot again for such a quick reply!
The issue is that your input is "gridded data" not "triangulated data" because each of x, y, z is a 20-by-20 matrix, rather than a set of vertices and triangles.
This can be handled by stlwrite (see the help file for the "triangulation" parameter), but you'll need to understand that the default "triangulation" of gridded data is a delaunayTri of the X-Y 2D data. In this sense you have repeated X-Y pairs (one at the top of the sphere, one at the bottom). The warning you're getting comes from "delaunay", not from "stlwrite".
You can overcome this by specifying how you want your gridded data triangulated:
Note that you'll need "mesh2tri" from http://www.mathworks.com/matlabcentral/fileexchange/28327 for this to work. I will ask Kevin for permission to include it in stlwrite (it's a small but very handy function) for a next release.
Hi! I am facing the same issues as Alexandra. I was trying to gererate surface mesh for a sphere and then converting into .stl format. When I tried to view it in Meshlab, one half of the sphere was gone. I did get a warining about duplicate data points.
Warning: Duplicate data points have been detected and removed.
Some point indices will not be referenced by the triangulation.
The commands I used are below:
Any help will be highly appreciated. Thanks.
@Alexandra, I'll bet that probably there are some duplicated/disjoint vertices in MATLAB prior to using stlwrite. Can you send me an example of the data that is coming through open?
I have the problem that I get open surfaces after using STLwrite. Has anybody an idea what to do about this?
does anyone know how to show the different colors per face I used? I've tried meshlab and 3d max
Thanks for your help Sven, turns out the problem was in my import settings in Gambit. It needed to be a little different because of the more complex geometries in those files. Again, wonderful program!!
Ha, that last message was from me - logged on from another computer at work.
@Nick and Adam, I'm not sure what issues may be at play with your files. Feel free to email a .mat file with your points/faces data *and* the 2 or 3 lines you use to create a working(!) figure from this data in MATLAB (note that I'll only have time to look at it if the figure creates cleanly).
Great program. It worked fine for 2 of my files with relatively simple geometries, but one other didn't output a readable mesh (I tried to open it in both Solidworks and Gambit), the other didn't connect the mesh into triangles, there were just some points in Gambit. Both mesh just fine using the trimesh function. Any ideas?
@Sven you've been very clear, thanks. i ask an other question, to better understand.
my set of p x-y-z describes a 3D surface that i can see properly in matlab (using plot3 or trisurf), as you say.
to write the stl, I can use directly the function "stlwrite"?
Does this make a triangulation of my 3D coordinates?
In that case, which options I have to give to the function to make it works properly? because I have tried many ways, but when i give 'triangulation' it wants other options and values, and I don't know what to put into.
@Nick: If you create a Delaunay triangulation from a set of p X-Y-Z points:
dt = DelaunayTri(x,y,z)
The result (dt.Triangulation) will not be a "triangular surface mesh". You'll notice that the result is an n-by-4 matrix. This means that each row in the triangulation represents a tetrahedron - a small *solid* with 4 triangular faces - not a surface triangle.
STL format describes surface triangles rather than tetra solids. Perhaps you want the convex hull of your points? (ie, only the *outermost faces* of all of your tetra solids) Note this will only work if your simple solid body is in fact convex.
Here's my advice: plot to the screen before trying to write to an STL file. If you can get a 3d surface plot (using patch, trisurf, etc) working in MATLAB first, *then* you will be able to write to STL without problems.
hi, I have a set of coordinates of a simple solid body, and I have to create finite volumes to work on. So I made a 3D Delaunay triangulation, and I have now a n-by-4 matrix with the points of resultant tetrahedron. Now, I need to create a .stl file to export this work. how do I do this? Is it possible? Because in the option There is only the voice for a triangulation in 2D. Thanks a lot.
@ava: glad to hear that the issue (nodes indices) was resolved
@Sven:I am using STL viewer and also MiniMagics 2.o to see the result file but all the faces are disconnected.
@ava: This is the first I've encountered such an issue. Does the same thing happen when you run the example in the stlwrite help? If not, try using the splitFV FEX entry to see if the original faces/vertices are indeed connected. If the stlwrite example itself comes out unconnected, what software do you use to read the .stl?
Faces are not connected in the result stl. ,Does anyone have any idea?
Works nicely. For my old Matlab version, I had to remove one line to make it work:
Binary output is now very fast! :)
The latest update has made significant speed improvements with vectorised face normal calculation and combining adjacent fwrite calls. Hope it's useful.
Works well, though very slow (in binary mode at least).
Great! That's what I looked for...
Nice companion to cad2matdemo.m. Very useful.
Corrected face count printed output bug. Added example with coloured faces.
Embedded version of Kevin's mesh2tri for convenience.
Added faceColor implementation
Huge speed improvements courtesy of Oliver Woodford. Reduction in write-time is proportional to file size.
Allowed for more general input types, and added capability for more features to be written to STL file.
Vectorised normal direction calculations and streamlined fprint/fwrite calls. 75% faster writing speeds.
Download apps, toolboxes, and other File Exchange content using Add-On Explorer in MATLAB.