I have a file with two groups of text, and I need to run a contour operation around the text.
Using 2 toolpaths (one for each group of text), I see pathing like the following. Not perfect, but most importantly there is only ONE long travel.
If I apply one toolpath to both groups of text (‘Letters test’), I get this:
There are 15 long travels.
File:
Test.c2d (1.3 MB)
It looks like it machines the objects in the order they were created. However, it machines all the outside profiles first, then all of the inside. This simulation showed all the outside profiles cut, darwyn first, designs second, 2023 last, then the centers of the 0 & g… I paused it.
V-Carve & Adv V-Carve also follow some sequence. clear out the centers (If selected) then It seems to do a bunch of picking out tiny details first, then carves the majority of the profiles. Then back to pick out all the corners. It also looks like there is some rudimentary optimization in that it appears to the next closest shape or cut that is closest to where it left off on the previous cut.
Create an identical toolpath for each group of letters. Of course then you will have to change bits more often.
I’ve tried watching my machine do its thing but have learned to just let it do the job I hired it for. I’ve seen videos where guys reduce retract height to a bare minimum in order to cut time down. Unless I am in a very tight time window, I just don’t worry about it.
Still, it is an interesting puzzle.
The point here is not “If I fiddle with it I can do better than the computer”.
The point is that the computer should be able to do WAAAAAY better.
I know the Travelling Salesman Problem is NP hard, but we don’t need and optimal solution. We just need a “pretty good” solution.
It seems the current algorithm is dependant on both winding direction and Y axis value, and that’s a silly algorithm. I mean, look at this:
Exact same file, I just moved the second block so the whole thing is vertical instead of horizontal.
Even the most basic ‘greedy’ method would WAY out-perform this, and it wouldn’t have such wildly divergent results just because one scenario is 90 deg from another one.
This topic was automatically closed after 30 days. New replies are no longer allowed.
