osresearch / plotter-vision Goto Github PK

When exporting SVG, the first line reads

<?xm version='1.0'?>

but should be

<?xml version='1.0'?>

See https://www.w3.org/TR/xml/#NT-XMLDecl

Plotting with axidraw reveals that there are quite a few duplicate vectors left.

Off-screen segments are included, which makes for messy cleanup before plotting.

With the Cohen-Sutherland algorithm it is much cleaner.

As demonstrated in @level2three's tweet, curved surfaces end up with very dense numbers of line segments.

stl.min_length = 10; reproject=1

seems to produce reasonable results.

Only binary STL files are currently supported, so the output from OpenSCAD is not always usable. It should be fairly easy to add a parser:

solid OpenSCAD_Model
  facet normal -0.0727553 -0.948536 -0.308198
    outer loop
      vertex -3.89628 -11.9915 -2.40803
      vertex -2.92811 -10.8633 -6.10875
      vertex -3.2793 -11.9442 -2.69932
    endloop
  endfacet

The cube test in #8 shows a rendering error in some orientations.

When loading a file it could add to the current STL, rather than replacing it.

Something is messing up coplanar surfaces and adding occasional line segments.

One idea , how about placing the red/blue colours on two seperate layers. If you are going to plot them, this is going to have to be done one way or the other.

Same test as #8 - certain viewing angles make edges invisible that should be present.

When using the scroll wheel to zoom, firefox scrolls up and down a little bit due to canvas window size issues.

It would be useful to allow precise positioning of object / camera for fine grained controls

Duplicate the STL, create a second camera slightly to the side, etc.

I was wondering if you were interested in sample file that were on the largish side, from various apps.

I am a Mathematica user and was experimenting with STL export of various 3D plots. I will upload as a make them. I will also post tips for reducing / simplifying polygons in Mathematica.

Thanks for the ability to easily experiment with SVG for 3D!

Seashell - you can visibly see it redraw from red to white. SVG export looks correct
seashell.stl.zip

The SVG orientation (0,0 in bottom left) does not match the screen (0,0 in upper left).

Currently the exported SVG has a width/weight, that is fixed and much larger than the actual drawn surface.
It would be good to have a view-port that shows the part of the canvas that is really used.

For example the bunny in https://twitter.com/pilhuhn/status/1345759428239364098
has <svg width='1480px' height='885px' ...>
but the actual path data is:

minX  : 351, maxX = 1051
minY  : 11, maxY = 760

which translates to Height: 700.0, Width : 749.0

The n^2 vector sorter could be applied prior to download so that the SVG doesn't generate double strikes, as seen in Heiko's #plottertwitter tweet

The close_enough() check on the normals should be tunable by the user to handle curving surfaces.
https://github.com/osresearch/plotter-vision/blob/master/triangle.js#L126

When STL is loaded, it would be helpful to always have the same initial size for any STL.
For example, ln does this with a unit cube and fits the object inside the unit cube: https://github.com/fogleman/ln/blob/master/ln/mesh.go , https://github.com/fogleman/ln/blob/12e6c6e7445909f817f1b8d5581be4255ffa1437/examples/slicer.go#L18

This way, no matter what STL you load, your zoom level and camera position can remain the same.

The SVG does not have any bounding box, so the object is not located where is should be on the page.

Hi! I was wondering what is the license for this code?

I am planning to build out a collection of tools on Plotterfiles.com and am looking at various tools with permissive license in order to be able to modify the code and implement new features without affecting the original repo.