Super manual engraving

I wanted to see if it would be faster to make my own engraving paths. Basically I made a series of concentric or spiral shapes and scored them to see if it would work as an engrave. The end result is that it does work, with mixed results.

All settings are 500/15 on Baltic Birch. This is not a great setting for anything, it was just reasonably fast and showed what I was looking for.

First: This is a traditional engrave. Nice and crisp. no hot spots, but slow. Engrave time: 7:00

Next up, I took 300 .2mm lines, converted stroke to path, and did a path->intersection with some text outlines in Inkscape. This yielded a ton of skinny “rectangles”-ish. They aren’t really rectangles, but they are skinny horizontal shapes. Pretty tough burn, and some obvious steps on the sides of things, but a bit faster, at 6:15

After that, I took a spiral shape and did a similar slice. This was even faster, at 5:06.

Some additional detail on the spiral engraves:

And here are all three engraves stacked up.


But really it all started with seeing if a set of concentric circles would work as an engrave. On left, a traditional engrave. at top, nested concentric circles, and bottom right is a spiral. Speed was really interesting and predictable here.

Traditional engrave: 1:46
Concentric rings: 0:27
Spiral: 0:13

Slightly more detail:

Closeup on the spiral:

Spirals are very very fast compared to traditional engrave, but I think the main reason for that is because it is one unbroken path with no sharp corners. As with all scores, you see hot spots at the starts and stops of paths, which is why you get that unsightly burn in the concentric circles. This could maybe be improved by rotating the circles, and breaking the paths at different spots, but I didn’t get to try that yet.

What do I take away here? The spiral engrave is visually very similar to the traditional engrave style, and only takes about a seventh of the time. If we can find a way to manually offset paths that results in a single unbroken path with no or few corners, we can likely see faster engraves for primitive shapes, but this may not have much application outside of simple jobs. These results are definitely subjective and skewed, because I was operating at a .2mm pitch which I think comes out to about 64 LPI, but even linearly scaling for LPI, it’s still much faster.

If Glowforge solves the hotspot problem on scores, then we’re starting to see real potential to use this, especially when engraving text. Concentric rendering techniques were a bit under a third the time as a traditional engrave, at least for circles.



This is great! Very cool concept.


Instead of spirals you should try a tesseract next. I expect it would have an impact on the time.


A tesseract? As in a 4-dimensional cube? Not sure that’s something that even Glowforge can do…

Very interesting and cool! Thanks for posting your findings. I especially love the look of the spiral engrave in particular. I imagine you could effectively mask the start and end burn points by doing a final score around the whole thing. As far as total project time, how much additional file prep time does it take? I see the big advantage on saving tube hours, but was wondering if longer file prep negates the lower engrave time.


You’re dead on about file manipulation time. My laptop struggled with the sheer volume of nodes in Inkscape, it got bogged down trying to do the path clipping.

That being said, a good portion of what took me so long was the learning process. Now that I have a handle on how to do it, it’d go much more quickly. YMMV depending on your individual setup and skill level… I’m an enthusiastic novice at best :slight_smile:

As for the scoring to mask the end points, I didn’t try it, but my thinking was that it’d have the same problem, the issue is score starts and stops, so any score used to conceal it would have the same issue.

I don’t think we’ll see that go away until the next time Glowforge updates their algorithms.

1 Like

Hit it with a Hilbert curve. :wink:


Not a bad idea, though filling an area with a vector pattern is not trivial with Inkscape (though it can be done – see Halftones in inkscape ), and getting said pattern to be a continuous path is not something I’ve figured out how to do yet [nor seriously tried]. I hope that someone reading this is more clever than I (safe bet!) and has a sneaky idea because of what I posted.

I did wonder what it would be like to fill a vector with a tight vornoi pattern. I’ll queue that up.

1 Like

Vornoi fill converted to vector…


Takes 1:22 to score… Run it twice already, it absolutely torches the wood at the normal 100/15. I ran it down to 500/9, still torched it. Will try something ridiculously low, like 500/5.

That turned out OK, but still really dark. Anyway setting aside, it’s a valid concept.

Speed is 500, bottom right is 15, bottom left is 9, and top one is 5.

Raw detail on the 5 power.

After cleanup with denatured alcohol and a light sanding:

Again, not sure how useful any of this is, but it’s another technique to keep in mind.

To get this done was a bit of a pain. You make your shape (vector) then use the Extensions->generate from path->Vornoi pattern action. You can change the parameters to suit your needs. I chose very small cells because of the scale I am working on and the fact that I am trying to approximate an engrave.

When that’s done, I then had to export the pattern to a PNG, and re-import it into Inkscape as a raster object. Then I choose Path->Trace bitmap to get it back into a vector form. Once I’m there it’s just a matter of sending it to the GF.

There may be a more direct path to get this done, but if there is I haven’t found it. Maybe some Inkscape ninja can tell me how dopey I am being?


Nice work! I was thinking about this a few days ago myself but no time to experiment. I’m hoping the GF toolpathing will evolve to create continuous engrave paths, it should dramatically improve engrave times. I was thinking about writing out some paths that do just this, but the software I would use to do it would only be able to write linear line segments, not curves, so the data could get quite heavy. Not sure how many vertices the GF can handle.

If you’re going to do this, it would probably make sense to have a two-step process where most of the manual engrave happens with massive defocus for big fast spot size and then a quick cleanup score around the edges.

1 Like

Good ideas. I only have so much bandwidth to try this for now. Anyone else wants to tear it up, please post results here :slight_smile:

The forth dimensional vectors should cut very quickly in three dimensional space. You do have to be careful though as the vertices will likely concentrate a huge amount of power on the substrate. As long as you follow normal safety precautions for bending space/time in the vicinity of a class four laser it should be ok.

The original post was a joke. :wink:


I remember seeing Carl Sagan explain Tesseract’s on the original version of Cosmos.

I was very young at the time but he made dimensions (1, 2, 3 and 4th) all so easy to understand… perhaps too easy to understand because it was one of 3 mind = blown moments that defined my childhood



Because it was brought up in a Discord conversation - and also because

was solved, so I bet the spiral would work even better :slight_smile:


Yeah, I’ve never come back to this. I should maybe?