The information is gone. There is no way to automatically reconstruct a curve from a sequence of straight lines. A human can do this because he can see what was intended.
A better approach is to create a slicer that can work with .STEP files directly. The trouble with that is the mathematics gets hard. If you remember basic algebra, it is easy to find where two lines intersect if you have the equations for each of them. There is no need to guess, you get an exact solution in one shot. It is easy enough that they can even teach this to 13-year-old kids. But just try and find the intersection of a line and a curve. Except for special cases, you are forced to solve it numerically which is very slow and a slicer needs to do this many millions of times. Slow, but in theory, it could be done. So the practical solution, what we all use is to create STL files that have a specified maximum deviation from the ideal surface. You can set this to 1/10,000 inch if you like and then any error is undetectable. Slicing a file like that would be slow but slicing a STEP file would be way-slower. The OTHER thing we can do to make better 3D prints and I suspect if you found a good design, this was done, is to design around the capabilities of the printer. Use simple ideas like filets and wall thicknesses that are multiples of the nozzle diameter. It is not hard and if you are designing from scratch you can make a wall 1.60 mm thick rather than 1.5 mm and it will come out better if using a 0.4 mm nozzle. There are probably a dozen more tricks like that. On Wed, Nov 11, 2020 at 5:41 PM Gregg Eshelman via Emc-users < [email protected]> wrote: > Since it looks like there's little or no progress on an update to the STL > format, there is work going on for 3D printer G-code post processing to > refine and smooth curves or even convert curves defined by a series of > straights to real arcs for printers that have hardware and firmware capable > of doing real arcs. > Supposedly units are arbitrary in STL. There's nothing in the file > format (unless with a proprietary command that only specific printers > understand) for indicating what unit or scale to use. Assume 1 unit = 1mm > and design accordingly. In the 3D software I use, for STL export it makes a > meter, centimeter, or millimeter all equal 1mm. I use the different model > and world unit settings to best fit whatever I'm doing on the screen. > Design tiny stuff in Meters in a Centimeter world and big stuff in > Centimeters in a Meter world. Working in US units produces STLs that aren't > the expected size. > It's all the fault of 3D Systems. When they created STL they were only > thinking of millimeters because their early stereo lithography printers > that scanned a LASER across the top of a tank of resin only made rather > small parts. One of the first products prototyped with those was the roller > clamp used on IV fluid lines and that was a pretty large 3D print in its > day. > A little bit of foresight and they would have made STL able to support > many types of units. Just a simple UNITS= line at the start and have the > printer software able to read that and adapt the movements. > On Wednesday, November 11, 2020, 09:47:14 AM MST, Chris Albertson < > [email protected]> wrote: > Remember the workflow. You start with a 3D model where the resolution is > literally "perfect" as the curves are defined by mathematical functions. > Then you convert the 3D model to an STL and an STL can contain only > straight lines. No circles, no curves. When you make the STL you get to > specify a maximum error but small errors mean big files. > > So when you look for a gear or sprocket ALWAYS look to see if you can get > the 3D CAD file and not just the STL. The best places are the big > resellers like Boston Gear, McMaster Carr, SPD,.. they all have "perfect" > 3D CAD models, usually.STEP file type. > > If you do download an STL then you are accepting some other random person's > idea of what is a good error tolerance without even knowing what he > specified. > > *So make the STL file yourself. * I usually go overboard with resolution in > the STL and specify something like 0.001 mm or less allowed error. > > The next step is manufacturing where you get to specify how it is sliced > and printed or how the file is cut on a mill. > _______________________________________________ > Emc-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
