A spiral? (Just guessing.)
I’ve been thinking about engraving times. Since almost all of my engraves are closed path vector fills, I can combine and group and do efficient engraving. Yesterday I did some engraving of multiple rulers each with a bitmap on the piece, same bitmap. Because I imported one bitmap and copied and pasted it in to get the resolution I needed, that broke up the engraving into discrete sections. Trying to think through this fully, but having vector filled engraves really can speed things up. However, you lose gradients at the moment.
I’m confused. “Closed path vector fill” means doing a vector score around the perimeter of a closed area, right? Or does it mean doing a raster engrave over a closed region defined by a vector curve? Or does it mean a raster engrave with a fixed power/speed, so it’s a region with a uniform level of material removal?
Vector engraving a fill means zigzagging within the boundary, just like an FFF 3D printer draws an outline and then fills it with a zigzag. A raster engrave scans a rectangle turning the beam on and off, like a CRT monitor. Theoretically the results are the same when engraving a monochrome image but vector can be a lot faster than raster if the image is sparse.
So the second one of my guesses, right?
Neither are right because you say raster engrave. A raster engrave is what Glowforge calls engrave where the head scans a rectangle and creates a picture in that rectangle by modulating the beam. You can raster engrave raster images (they are bitmaps resampled to the LPI) and you can also raster engrave vector images where vector fills must be rendered to a bitmap at the correct LPI.
There are two more combinations in the laser world. You can vector engrave a raster image, that is tracing. You can also vector engrave vector artwork. The laser doesn’t scan a rectangle, it follows the vectors. I think GF then only supports unfilled shapes and that is cut and score in their terminology.
Other machines can also handle filled vector shapes and zigzag within the bounds of the shape with the beam on. Yes that would most likely be a uniform removal. Care would need to be taken to prevent over burn at the edges where the head reverses. It may needs some overshoot and then it becomes a bit more like a raster engrave. The difference is it would scan each feature separately rather than doing long horizontal sweeps
I understand there are apparently various terminologies that some laser cutter manufacturers have adopted, which are at odds with the GF terminology, and you are describing those. I was going back to more basic terminology.
I relate the term “raster” to mean a process that performs a task in sweeping parallel motions. A raster printer doesn’t sweep out a rectangle, but stops at the end of the last character on each line, thus sweeping over a non-rectangular region.
“Vector engrave of a raster image” you call tracing. To me, tracing is a process of creating vector shapes at transitions in intensity of a bitmapped image. This produces a bunch of vector outlines, and would not be related to any engraving process, in my mind. If I had to define something called “vector engrave,” it would be a bastardization of “vector fill,” which would be the act of scoring a 2D space filling curve within an arbitrary boundary.
Yes a raster scan can be optimised to have non-rectangular edges when mechanical motion is involved but always covers all the features in the image in one horizontal sweep, although you can have several raster engraves run in a sequence.
Yes tracing a raster image creates a vector image.
Yes you could use any space filling curve but horizontal zigzags are the most efficient for Glowforge. Spirals probably involve the least change in direction.
There’s definitely a case to be made for optimizing sweep paths to eliminate time spent sweeping over empty areas that aren’t being engraved.
It’s hard to imagine doing the zig-zag approach without overshoot though, and once you’re overshooting you’re back to doing ordinary raster engraves.
It would be interesting to prepare a design with two thin vertical lines on the far left and far right sides of the work area. How much time would it take to raster both at once compared to rastering each one separately? If one line takes one minute and doing both at once takes 16 minutes, optimizing the rastering areas would shave 87.5% off the total job time. With these times, time could be saved all the way up to doing 15 lines individually instead of all at once.
I suspect the faster the machine can accelerate, the more helpful optimizing the raster areas would be. Since the Glowforge UI has such accurate time estimates, someone with access to the UI could accurately answer these questions with regard to the Glowforge without needing to waste any material. Any volunteers?
I could check the time estimates with my Trotec, but I don’t know how accurate they are.
Yes the same effect could be achieved by analysing the raster and breaking it up into localised scans to optimise the time. A very long time ago I wrote a graphics compression algorithm that did just that to compress cartoon style animations into a format that could be rendered very quickly with a graphics engine with a blitter. It was worked well compared to video codecs around at the time that expected video to be pictures.
“Blitter”? Typo or a word I don’t know?
Word I don’t know. ;p
(too far out of my wheelhouse to read right now)
Yes they basically perform logical operations on rasters very quickly.
I basically computed the difference between frames and then split that up into small rasters using an algorithm that knew the overhead starting a new raster and the rate at which it would run. So it could decide whether to render two areas of change with two separate rasters or a bigger one that encompassed both.
It was in the early nineties I think so I don’t remember the exact details but it was probably a heuristic rather than a guaranteed optimal result.
I regret that terminology can be ambiguous according to what people are used to. I am literally talking about my design as it appears in Inkscape. My designs rarely contain bitmap images. All the elements are paths in Inkscape. Those paths I want to engrave on the Glowforge are closed paths representing shapes. Their bounding path does not have a color assigned to it. The enclosed shape is filled with a color. I engrave that filled color area. I can choose to score or cut the bounding path vector as a separate operation if I make the path a different color.
It seems that any time I mention the fact that Glowforge can take a vector path and turn it into an engrave operation using a raster motion of the head discussion ensues. I try to relate my experience and understanding of the design to engrave process as the Glowforge does it.
Can’t find it now, but i believe it was @jamesdhatch who was also advocating for breaking the patches of engraving into zones to speed up the processing time. (It is quicker that way, if you’ve got a sheet full of engraving to do.)
I haven’t needed to use it much yet - most of the engraving that I have done is fairly small… a few inches or less, and it doesn’t span the entire 20 inch wide bed. And I tend to cluster all the parts together to save material too. So the time savings turned out to be pretty negligible in those cases.
One trick I do use a lot though is one that @dan mentioned a while back…when I engrave, i try to keep the bulk of the engraving running along the X axis. That has a pretty large effect on the time as well…(see below.)
Same size, same engraving settings on both rectangles, the only thing that changes is the orientation.
So that’s something else to play with if you want to speed up your prints.
You give GREAT examples.
One of the cool things about a blitter is that when programmed correctly it can also do (at least) 90-degree rotations.
Makes sense. Far fewer direction changes in the long horizontal vs the skinny vertical version.
And yes I split any wide area engraves into different colors so I can make the laser do them as individual units to prevent long sweeps of the laser on intervening non-resized sections where it’s just wasting time moving between the engraves with the laser off.
Well thats good to know. makes sense now that i think about it.
Thanks for sharing.
Man I wish I knew I could do this before I started my 3.5 hour print. I imagine doing it this way would shave at least 2 hours off of the time. Though I’m actually using the 3D engrave setting for this job, so that technique probably wouldn’t work. I’m making a spice rack and 3D engraving the slots for the shelves to slide in. I used black as my color so that I would get a nice deep groove. Is there any other way you can think of to break up the job? I’ve got four 1/8th inch grooves spanning about 16 inches, its taking forever.