Ok,
Here is a very basic python gcode previewer that accepts gcode from the RepRap host or skeinforge and displays each layer like this:-



I have attached the files, usage is Preview.py gcodefile. I will expand this to show dimensions and stats in the coming days.

Edited 1 time(s). Last edit at 04/20/2008 06:58PM by nophead.

---

[www.hydraraptor.blogspot.com]

This is the type of thing that I was looking for someone with real expertise to do. I hate how the Java 3d stuff clogs up my computer. I haven't taken a look at any of the files yet, but this GUI just slices and prints the Gcode right? I then have to port that over to Chris's control GUI in order to get my machine to print?

---

Jay

Hair combing has been added to Skeinforge. The default for the Comb Hair checkbox is true, and the hair gets combed between the perimeter and the fill. Since there is little room between the perimeter and the fill, it is recommended that when the combing hair, use the fillet.py script to bevel the fill, so that there will be less corner blobbing and more room for the hair to be tucked in. Gcode can be beveled in fillet with the following code:

import fillet
gcodeText = getBevelGcode( gcodeText )

Also diaphragms have been added. They are layers which are solidly filled, even if there is a sparse infill. The default diaphragm period is nine and the thickness two. With diaphragms, we can extrude structures which are lighter than water, but have many watertight slices and won't easily buckle.

For the next few weeks, I'll be porting Skeinforge over to python, during which there'll only be bug fixes.

Nophead,
Thanks for posting all the code and more in future. It'll be neat to actually see the head lifting and nozzle wiping on screen

Greenarrow,
I have found a now squashed negative array index bug, which hopefully was your bug, so please try again with the new version.

Jay,
Indeed the GUI just slices and prints the Gcode, slowly. You can preview gcode in the python script preview.py or in the Art of Illusion script Skeinview.bsh. You then have to port the gcode over to Chris's control GUI in order to get your machine to print.

Enrique

Edited 2 time(s). Last edit at 04/21/2008 01:37AM by Enrique.

Enrique,
You can see the nozzle wiping and head lifting on the video at the top of my blog. I only wipe the nozzle once at the start of the build. The head only lifts 0.4mm so it's hard to see, but you can hear it as my z-axis is noisy.

I think the reprap host turns the extruder off before the end of each fthread and wipes the nozzle before starting the next one. I am not doing any of that yet which is why my prints with PCL are more hairy than Adrians. Hopefully, the piano wire trick will mean none of that is necessary, but it is very easy to add in the meantime.

Not sure what you mean by hair combing and bevelling but I will have a look this evening.

I noiticed that in some cases less hair would be produced by reordering the fill threads, possibly reversing some and then rotating the outline order to start and end near a fill start. The wineglass is a good example. Since the number of threads is typically small, I might have a go at doing a brute force optimisation. It will be O(N!) but N is typically 4 or 5 in the objects I am playing with.

---

[www.hydraraptor.blogspot.com]

Enrique,
I just tried the new skeinforge with a simple cube again and it worked first time! Also, i've finally figured out why I couldn't write to the svn so I've added your scripts now: [reprap.svn.sourceforge.net]
I've re-arranged things a little now, it's in a proper installable python package. Check the readme for install. You now have to use 'import reprap.fillet as fillet' but this can be done from any directory (does not need to be the one the scripts are stored in). I've updated this in the docs.

nophead,
That gcode previewed looks good, I will have to give it a go later. Thats and impressive glass on your blog too!

Hi Nophead

I downloaded python (Windows Version)
I have added it to Path and I tried Enrique's fillet.py and that worked OK so the path setup is OK.

I tried your preview.py and got the following:-

J:\ReprapCode>Preview.py Object-0.gcode
Traceback (most recent call last):
File "c:\python25\preview.py", line 1, in
from layers import *
File "c:\python25\layers.py", line 2, in
import Image, ImageDraw
ImportError: No module named Image

Have I used the wrong syntax or am I missing something that needs installing? For the moment I have put all the *.py files in the Python25 directory to make sure they are locatable.


J:\ReprapCode>

---

Regards

Ian

Hi Ian,
You need to install the Python Image Library, it doesn't come as standard but it is free for non commercial use.

[www.pythonware.com]

Edited 1 time(s). Last edit at 04/21/2008 01:44PM by nophead.

---

[www.hydraraptor.blogspot.com]

Hey guys. I have a windows machine that is pretty clunky and I was hoping if you guys could tell me if python and its image libraries could run on linux?

---

Jay

Yes Python runs on Linux and is generally preinstalled. I don't know about PIL, you might have to build it from source. I will investigate.

---

[www.hydraraptor.blogspot.com]

J Sweezy Sincerely,
Python is installed on almost every linux distribution by default. I had to install the 'Python Image Library'. In Fedora linux I ran 'sudo yum install python-imaging-tk' to install the package.

stefan

OK, thanks Nophead and greenarrow. This is great! I don't think you understand how happy I am for all of the work that you guys have done on this tangent! My goal for this project originally was to have my Asus EEE be the computer that would operate my Reprap, but with all of the problems in the Host, it was looking like I would have to use a beefier computer to get the job done. These python scripts are pretty simple and they don't require loads of memory to operate. I swear this is a blessing.

Now this just leaves me with the debugging of my machine so that I can scramble all of my stuff together to make it print by the weekend. The Washington DC Area RUG is going to RobotFest!!! on the 26th, and Dorkbot on the 29th where we will be sharing the RepRap project and my *hopefully* finished Darwin. Anyways, thanks for all of your guys' effort in making this new slice and dice GUI/Viewver, even if I didn't understand half of what was going on for most of the time.

So again, Thank you Enrique, Noph, and Greenarrow

---

Jay

Hi Nophead

Many thanks. Again......

Preview working fine

---

Regards

Ian

Hey guys,

Is it possible for one of you to compile all of this stuff into one file? I want to be able to email it to a bunch of my friends because I think that it is some pretty neat work. If it will take too much time don't worry about it, but otherwise, i would be extremely thankful.

---

Jay

If you want to send one file it's better to zip them up. I prefer to keep it in seperate files because Python makes each one a module with its own namespace. That helps being able to reuse the code. For example, I posted just the modules required for the preview, but they are also part of a bigger program used to control my machine.

BTW, I have improved the previewer so I hope to post a better version this evening.

---

[www.hydraraptor.blogspot.com]

Oh, ok. Thats a good idea. Thanks Noph

---

Jay

I found a squashed bug where 'Comb Hair' would turn the extruder on and off repeatedly. So attached is the updated Skeinforge, along with the in progress python port. Also in the attached reprap zip file are all the python postings I've seen to date.

With the attached files, to turn an STL file into filled, filleted gcode, first import the file using the STL import plugin in the import submenu of the file menu of Art of Illusion. Then from the Scripts submenu in the Tools menu, choose Skeinforge and select the imported STL shape. In the Fill radio button group, choose Slice and click the 'Export Selected' checkbox. Set the parameters and click OK. Then type 'python fill.py' in a shell in the folder which fill is in and when the dialog pops up, set the parameters and click 'Save Preferences'. The parameters from fill will override any parameters of the same name from Slice. Then type 'python fillet.py' in a shell and when the dialog pops up, change the parameters if you wish but the default bevel is fine. Then click 'Fillet', choose the file which you generated in Skeinforge and the filled & filleted file will be saved with the suffix '_fillet'.

Documentation is now produced by the most excellent pydoc, which I wish I knew about years ago! It is described at:
[docs.python.org]

In the next week hopefully the slice code will also be ported into python, which is several times zippier than beanshell. Soon our gigahertz computers will keep up with our repraps

To avoid file capitalization problems, all my python scripts have lower case names. To avoid class & variable masking problems, from now on my class names will be upper case. The vec3 class was lower case, so the vec3 file now has the Vec3 class and the old vec3 class is deprecated and will be removed within a few weeks.

Cheers,
Enrique

I found a squashed bug where fill would draw duplicate paths. So attached is the updated fill.py, along with the rest of the python port. Also in the attached reprap zip file are all the python postings I've seen to date.

With the attached files, to turn an STL file into filled, filleted gcode, first bring the Export GNU Triangulated Surface script into Art of Illusion, by dropping it into the folder ArtOfIllusion/Scripts/Tools/. It is also available at:
[members.axion.net]

Then import the file using the STL import plugin in the import submenu of the file menu of Art of Illusion. Then from the Scripts submenu in the Tools menu, choose Export GNU Triangulated Surface and select the imported STL shape. Then type 'python slice.py' in a shell in the folder which slice & fillet are in and when the dialog pops up, set the parameters and click 'Save Preferences'. Then type 'python fill.py' in a shell in the folder which fill is in and when the dialog pops up, set the parameters and click 'Save Preferences'. Then type 'python fillet.py' in a shell and when the dialog pops up, change the parameters if you wish but the default bevel is fine. Then click 'Fillet', choose the file which you exported in Export GNU Triangulated Surface and the sliced & diced file will be saved with the suffix '_fillet'.

The GNU Triangulated Surface format is supported by Mesh Viewer, and it is described at:
[gts.sourceforge.net]

The python slice and dice tool chain is about four times as fast as the beanshell script. Now that all the slice and dice code has been ported into python, Skeinforge.bsh is discontinued.

Cheers,
Enrique

I found a bug which slowed the slice and dice toolchain down, squashing the bug doubled the program speed. So python is actually about ten times faster then beanshell for this application, rather than four times as fast which I though before. Doing minor optimizations doubled the speed again, if psyco is installed the program is twice as fast again. Attached is a zip file with the new code and many of the python postings I've seen to date.

With the attached files, to turn an STL file into filled, filleted gcode, first bring the Export GNU Triangulated Surface script into Art of Illusion, by dropping it into the folder ArtOfIllusion/Scripts/Tools/. It is also available at:
[members.axion.net]

Then import the file using the STL import plugin in the import submenu of the file menu of Art of Illusion. Then from the Scripts submenu in the Tools menu, choose Export GNU Triangulated Surface and select the imported STL shape. Then type 'python slice.py' in a shell in the folder which slice & fillet are in and when the dialog pops up, set the parameters and click 'Save Preferences'. Then type 'python fill.py' in a shell in the folder which fill is in and when the dialog pops up, set the parameters and click 'Save Preferences'. Then type 'python comb.py' in a shell in the folder which fill is in and when the dialog pops up, change the parameters if you wish but the default 'Comb Hair' is fine. Then type 'python fillet.py' in a shell and when the dialog pops up, change the parameters if you wish but the default bevel is fine. Then click 'Fillet', choose the file which you exported in Export GNU Triangulated Surface and the filled & filleted file will be saved with the suffix '_fillet'.

The python slice and dice tool chain now makes sharp corners and the comb hair module combs more hair. The script vectorwrite.py, derived from Nophead's preview.py has been added which makes scalable vector graphics files, an example of which is at:
[members.axion.net]

Unfortunately, for reasons unknown to me, although when I view the file on my hard drive I see the graphics, when viewing the file online I only see the text. So if you only see text, try downloading the file and view it with mozilla or another svg capable browser.

The python slice and dice tool chain can now make multilayer shells. The shell thickness can alternate, so the fill can be interleaved with the shell and should have less horizontal delamination, as occurred when Nophead made an HDPE opto bracket:
[hydraraptor.blogspot.com]

The threads are now reordered so there is less travel and hair. Fillet.py can now fillet to a ratio, rather than just all or nothing. The bugs where the tool chain combed even when comb was not selected and where it did not handle all the files in a directory, are now squashed. The updated zip file is at:
[members.axion.net]

The instructions to use the tool chain are the same as in the last post. The full instructions are in "fillet.html".

This is a really cool piece of software, but I'm having some issues. After generating a (admittedly rather complex) STL file, and using AoI to create the GTS file, I feed it to the slice.py script and, after about two hours of processing, get the following error:

----------
C:\Users\Kyle\Projects\RepRap\skeinforge>python slice.py
File C:/Users/Kyle/Projects/SkyDial/triangle_mesh.gts is being sliced.
Exception in Tkinter callback
Traceback (most recent call last):
File "C:\Python25\lib\lib-tk\Tkinter.py", line 1403, in __call__
return self.func(*args)
File "C:\Users\Kyle\Projects\RepRap\skeinforge\preferences.py", line 268, in e
xecute
self.displayPreferences.execute()
File "slice.py", line 621, in execute
sliceFile( filename )
File "slice.py", line 312, in sliceFile
gcodec.writeFileText( suffixFilename, getSliceGcode( gnuTriangulatedSurfaceT
ext, slicePreferences ) )
File "slice.py", line 276, in getSliceGcode
skein.parseGcode( slicePreferences, gnuTriangulatedSurfaceText )
File "slice.py", line 574, in parseGcode
extruderPaths = self.getExtruderPaths( self.getSlice( layerIndex ) )
File "slice.py", line 540, in getSlice
return self.getLoopsFromMesh( zAround )
File "slice.py", line 514, in getLoopsFromMesh
originalLoops = getLoopsFromUnprovenMesh( self.triangleMesh.edges, self.extr
usionWidth, self.triangleMesh.faces, self.triangleMesh.vertices, self.slicePrefe
rences, z )
File "slice.py", line 151, in getLoopsFromUnprovenMesh
faceTwo = faces[ edge.faceIndexSecond ]
TypeError: list indices must be integers
----------

I'm running Windows Vista with Python 2.5.1. Attached are the STL and GTS files that were used when this error was generated. Am I expecting too much for the software to slice such a complicated object? My next step is to try it in Linux (I dual-boot, but all of the engineering software I use is windows-only), but I wanted to know if you had any insight.

Hi Kyle,

Thanks for trying out the slice and dice chain. Turns out the slice.py part of the script crashed because the stl file has a few holes in it. The script was supposed to be able to use a slow algorithm to span holes, but I never actually tested it with a holed mesh, and it didn't quite work. Thanks to your finding the bug, I've now added a check for holes and the script should now span them. From now on when it runs over a hole you'll get a message like:

"This should never happen, there is a hole in the triangle mesh, each edge should have two faces."
(edge)
"Something will still be printed, but there is no guarantee that it will be the correct shape.
Once the gcode is saved, you should check over the layer with a z of:"
(z)

I'm still testing the new script now which should work, but if it doesn't I'll post another update. The new script is uploaded at:
[members.axion.net]

If you want to seal up the mesh, I've seen on a posting that mesh viewer:
[mview.sourceforge.net]

can determine if a triangle mesh has holes, but I've never tried this.

The script should now be able to handle the complicated truss, although it is obviously taking a long time. I've never sliced such a large object with what I assume is the default extrusion diameter of 0.8 mm. If it takes more than a day to slice, I suggest killing the script and starting again with an extrusion diameter of 2 mm, which can be set by typing 'python slice.py' in a shell and when the dialog pops up, set the extrusion diameter to 2 mm and click 'Save Preferences'. The time to slice and dice is roughly inversely proportional to the cube of the extrusion diameter.

Just in case there's any confusion, you can use slice.py on its own to slice the object, to see if a file can be sliced, but that's all it will do, it won't fill the shape. To get a filled, combed filleted shape, type 'python fillet.py' and it will call the rest of the tool chain including slice.py to generate gcode for an extruder.

Cheers,
Enrique

Edited 1 time(s). Last edit at 05/14/2008 05:31PM by Enrique.

Great, thanks for figuring that out so quickly!

I have a minor question - what is the correct way to generate a gcode file for a mesh that you don't want filleted? I tried running fillet.py and setting it to "no fillet", but it just error'd that since I wasn't asking it to fillet, it had nothing to do, then quit. I re-ran it and left it on the default "bevel" option, but with so many small features I'm not sure, but this may significantly change the composition of the model!

Well, after between 1 and 2 hours of processing the same STL file, the new script threw the following error:

------
C:\Users\Kyle\Projects\RepRap\skeinforge>python fillet.py
File C:/Users/Kyle/Projects/SkyDial/triangle_mesh.gts is being chain filleted.
Exception in Tkinter callback
Traceback (most recent call last):
File "C:\Python25\lib\lib-tk\Tkinter.py", line 1403, in __call__
return self.func(*args)
File "C:\Users\Kyle\Projects\RepRap\skeinforge\preferences.py", line 268, in e
xecute
self.displayPreferences.execute()
File "fillet.py", line 578, in execute
filletChainFile( filename )
File "fillet.py", line 248, in filletChainFile
gcodec.writeFileText( suffixFilename, getFilletChainGcode( gcodeText, fillet
Preferences ) )
File "fillet.py", line 305, in getFilletChainGcode
gcodeText = comb.getCombChainGcode( gcodeText )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\comb.py", line 161, in getCombC
hainGcode
gcodeText = fill.getFillChainGcode( gcodeText )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\fill.py", line 282, in getFillC
hainGcode
gcodeText = slice.getSliceGcode( gcodeText )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 252, in getSlic
eGcode
skein.parseGcode( slicePreferences, gnuTriangulatedSurfaceText )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 513, in parseGc
ode
self.addExtruderPaths( layerIndex )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 400, in addExtr
uderPaths
loops = self.getLoopsFromMesh( z + zAround )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 472, in getLoop
sFromMesh
originalLoops = getLoopsFromCorrectMesh( self.triangleMesh.edges, self.trian
gleMesh.faces, self.triangleMesh.vertices, z )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 123, in getLoop
sFromCorrectMesh
pathIndexes = getPathIndexesAddPath( edges, faces, loops, remainingEdgeTable
, vertices, z )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 215, in getPath
IndexesAddPath
nextEdgeIndexAroundZ = getNextEdgeIndexAroundZ( edges[ remainingEdgeIndexKey
], faces, remainingEdgeTable )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 182, in getNext
EdgeIndexAroundZ
secondFace = faces[ edge.faceIndexSecond ]
TypeError: list indices must be integers
------

Any ideas?

Oops, sorry about the fillet or quit bug. Indeed, after you pointed out the now squashed fillet bug, I looked at comb and realized there was also a comb or quit bug. The debugged code is at:
[members.axion.net]

The default filleting ratio at 0.7 is quite small and is just enough to reduce blobbing in corners and spare the extruder sharp turns. For example an extrusion diameter of 0.8 mm gives an extrusion width of about 0.75 mm, which gives a half extrusion width of 0.375 mm, which gives a filleting radius of 0.26 mm = 0.7 * 0.375 mm. I believe it could be set to round corners more, but since I haven't heard of anyone extruding a filleted model yet, the default is small.

When you say your model has small features, do you mean that you're using a big nozzle? Because the struts seem to have a width of roughly 5 mm.

No, I'm using the default nozzle size (0.8mm, I think it was). We'll see how it goes - I'm dicing right now.

Ok, after something between 1 and 6 hours, the following error showed up:

----------
C:\Users\Kyle\Projects\RepRap\skeinforge>python fillet.py
The file fillet.csv does not exist, an empty string will be returned.
File C:/Users/Kyle/Projects/SkyDial/triangle_mesh.gts is being chain filleted.
The file slice.csv does not exist, an empty string will be returned.
Exception in Tkinter callback
Traceback (most recent call last):
File "C:\Python25\lib\lib-tk\Tkinter.py", line 1403, in __call__
return self.func(*args)
File "C:\Users\Kyle\Projects\RepRap\skeinforge\preferences.py", line 268, in e
xecute
self.displayPreferences.execute()
File "fillet.py", line 575, in execute
filletChainFile( filename )
File "fillet.py", line 245, in filletChainFile
gcodec.writeFileText( suffixFilename, getFilletChainGcode( gcodeText, fillet
Preferences ) )
File "fillet.py", line 302, in getFilletChainGcode
gcodeText = comb.getCombChainGcode( gcodeText )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\comb.py", line 158, in getCombC
hainGcode
gcodeText = fill.getFillChainGcode( gcodeText )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\fill.py", line 282, in getFillC
hainGcode
gcodeText = slice.getSliceGcode( gcodeText )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 252, in getSlic
eGcode
skein.parseGcode( slicePreferences, gnuTriangulatedSurfaceText )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 513, in parseGc
ode
self.addExtruderPaths( layerIndex )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 400, in addExtr
uderPaths
loops = self.getLoopsFromMesh( z + zAround )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 472, in getLoop
sFromMesh
originalLoops = getLoopsFromCorrectMesh( self.triangleMesh.edges, self.trian
gleMesh.faces, self.triangleMesh.vertices, z )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 123, in getLoop
sFromCorrectMesh
pathIndexes = getPathIndexesAddPath( edges, faces, loops, remainingEdgeTable
, vertices, z )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 215, in getPath
IndexesAddPath
nextEdgeIndexAroundZ = getNextEdgeIndexAroundZ( edges[ remainingEdgeIndexKey
], faces, remainingEdgeTable )
File "C:\Users\Kyle\Projects\RepRap\skeinforge\slice.py", line 182, in getNext
EdgeIndexAroundZ
secondFace = faces[ edge.faceIndexSecond ]
TypeError: list indices must be integers
----------

I just can't seem to get away from this. I also tried opening the mesh in mesh viewer, but couldn't get that to work (I didn't spend very long at it). I may need to try that again.

Hi Kyle,

Thanks for finding the second part to the holed mesh bug. The script now skips over null faces at the only other place where they can come up. The debugged code is at:
[members.axion.net]

Good luck,
Enrique

Cool, you're doing great work with this. I'll give the new script a try when I get home.

This seems to be working! The script has run for two hours so far and found six "this should never happen" occurrences. The latest has a Z-level of 18 (mm, I assume) and the model is about 36mm tall, so I anticipate it finishing in about two more hours.

Ok, it sliced successfully (I assume). There were a total of 6 incidents in the mesh, but the software seems to have navigated them all successfully - at least, as far as I can tell. I have no easy way of checking. Let me know if you'd like me to send you the gcode or svg files just for verification. I'm not actually going to use them for anything, I was just testing, so it would only be if you wanted to check the script.

Total slice time was 6 hours.

Awesome job!