What is the realistic length a material can be for the pro version

specs

#1

Okay so I wanted to do look into banners for the pro version of the laser. I was hopeful that the infinity statement was realistic, but after looking at the numbers, it looks like the length I would get is about 23 to 25 feet. The numbers are based on a 50% printed area with a run out of 1 degree. The internal plate looks loose in the videos, (slide in) so I don’t see a way to align long material. Anyone know how to deal with the problem? have you run into it before. Has anyone heard if the camera system finds the edges of the material and orients itself to long items. I am picturing those rolls of paper used for plotters here… Any insights?

Okay I just saw another post that using something like a continuous punch card for a musical instrument. So I guess the long sheets might be more common than I thought. Now I am thinking about player pianos and the long sheets.


#2

we haven’t seen the passthrough in action yet. I am under the impression that the camera system will see the edges of your material and the last portion of lasered design, re-register/correct for misalignment, and continue the cut/burn.


#3

Yeah, I didn’t quite understand the 23-25 ft issue with the 1 degree and loose plate. But everything I have seen posted suggests what @jbv said. The registration is done optically independent of the slight horizontal position differences.


#4

Thanks, but I was hoping the guys at glow forge would answer. I understand the re-alignment with the software, but if it doesn’t know where the edges are in a material then it will eventually track off the 20 inch material. The numbers are based on a set of parallel lines. If we use an image 10 inches by 30 feet (think MC Escher here) then if the material (paper roll) goes in off by one degree then the image if it is centered in the glow forge will track to the edge of the paper in about 24 feet. The image will be aligned correctly relative to itself, but not relative to the material being feed into it. A 250 foot roll of paper has even less alignment window.
If the machine has a reference edge for the material being input then you are more likely to achieve the infinity length given a hard “fence” to reference. The tray with the honeycomb looks like its just slides into the machine so theres no reference on it as far as I can see. The way they are aligning the machine for a zero reference is a “secret sauce” sort of item, but given it has a laser and cameras built in it not hard to image how they are doing it.
Dan and his team are stepping into a 20 to 25 billion dollar market with this little Gem and I am not sure he has seen all the potential beyond the 14 billion dollar craft market. But once he goes live watch out he will have himself and “E” ticket ride.
Regards,
David


#5

The Glowforge team does not follow the forum posts closely. Dan the CEO skims the forum most nights but he might not notice this unless you reference him in the question. I would guess that infinity in the context of what you are asking may not be practical for the Glowforge. But just so you know… There have been zero official answers to any of our S/W specific questions. The company close holds that info for a bunch of reasons.

The GF does not align to the honeycomb tray. There are no guides internally as far as we know. Alignment of the image to the material as you pass it through is theoretically possible for straight edge material but registration over extreme lengths would not only require an edge detection but a morphing of the imagery since paper rolls can enter the unit a little left, a little right, a little crinkled. I would guess there will be some human adjustment involved. Also the GF has to deal with irregularly shaped material. What if I wanted to engrave a very, very large boomerang?


#6

You deserve more than a like here, @rpegg. I’m still amazed by your grasp of all the technical details and ability to communicate them. Thanks.


#7

In another post, I remember way back, It was confirmed that there were hardware edges, rail, or bed walls on the left and right sides. This does in a way, help this issue. It was referenced or mentioned in a thread when I was asking about template file vs snap image to software guides/grids. Whether or not those rails end up in the finished product is up in the air, but it makes logical sense.


#8

Hi Rpegg,
interesting, but boomerangs are easy it gets placed into a fixture, I asked Dan about this sometime ago not the boomerang something similar. The image would follow the shape and the camera follows the profile. Of course the boomerang must move through the glow forge. I asked the question because the advertisements and subsequent videos, interviews all say “as long as you want” realistically not possible for the reasons you state, the problem with bringing human into a machine environment is “human error” I like Sawas post about hardware edges, but those may be the case sides either way I can hope one of the glow forge guys trys to print a banner for someones birthday or a holiday. If they run into the problem on Laser Thursdays maybe they get a fix or have a solution they can pass on to owners.
can’t wait for the forging to begin!


#9

The following is just me throwing out thoughts not a disagreement: When I said very large boomerang I was just using that as an example of a curved or irregular surface. Think 15 foot boomerang shaped object. The pass through might be able to handle it but the S/W will have no way to determine whether the rest of the object is straight, curved or jagged. Alignment to the object becomes impossible. There has been a lot of discussion about automatic alignment to materials but as far as I can determine all of the company responses were about materials that could be imaged as a whole, like the Macbook or materials with deterministic shapes.

This response from dan was not about the pass through but the far easier condition of an object wholly within the Basic.

All of the academics aside… I can easily see why someone would want to etch a pattern on an 8 ft board and the registration issues you have discussed are easily handled. The effect is also something that is difficult to do in other ways. But trying to make a 25 ft banner in greyscale is not something that I personally would consider visually striking or cost effective, and would take a long long time.


#10

like this one?


#11

Ah very cool, but watch out when it comes back.:slight_smile:


#12

Hi Rpegg,
Actually Dan and I discussed fixtures and fiducial with respect to an items location, the pass through just opens the front of the machine for access. I am glad they are working on the problem. It may not be what everyone will tackle but if the machine is the only option you have to make something… then you use what you have. This machine will have the ability to make long materials and if you place the alignment marks there’s no reason to keep you from stacking 20 inch wide strips to make a larger item, like a wall sized stencil for some interesting mural work, most of the vision systems I work with are contrast based recognition. There is no way to know what code Dan and team are using for the vision, but capturing the edge of the material is easy using a back light. We use this approach to define certain shape recognition inspections as the silhouette is high contrast. We also use a laser to define a cross section, but that uses a sensor designed to read the laser. The potential for the machine is really high, can’t wait for the machine to arrive


#13

yeah, ouch


#14

What they already have possible… no clue.

What they currently plan to do… also no clue.

What they COULD do… That we can answer without being on the team. Doesn’t mean they ever WILL, but it does cover what is possible at all.

There are two cameras, a macro and a micro. The macro is the lid-snapshot, and the micro is mounted on the cutting head to get high resolution.

If you are cutting a 5,000 foot long banner, with detail from one end to the other over every solid inch, then flawless alignment is trivial.

Worst case scenario, each manual feed advances the material only 6 to 8 inches. This leaves you with 4-6 inches of previously cut material still exposed to the macro camera. Snapshot is taken when you hit resume, compared against the previously cut components, then the micro lens runs across the end of prior cut lines to dial in precise positioning and alignment. At most (with full details on every inch of this banner), you would need 1 inch thick of micro-scan to obtain flawless re-alignment for the newest cut portion.

Now, if there are gaps in your material longer than 10 inches, things get harder. The real ideal would be that you are making your banner less than 20 inches tall (but feeding material that is 20 inches wide, or at least wider than the banner you plan to make), and so the forge is cutting the material along the edges. This ensures that your cut field is always in laser range (there are material “bumpers” in play), and it gives you straight lines to use as your alignment guide for the full length.

But, if you have a material you do not plan to cut the sides of at all (it also happens to fit through the passthrough slot), and you have gaps between cuts of more than 10 inches… I still see a way for the Glowforge to obtain alignment from one end to the other. But it pushes WAY beyond the GlowForge staff’s current plan for the software.

For this scenario, you would pass your material through two times. First to scan it in as raw material. This makes the canvas on which you lay out your banner. Then you get the full inkscape/adobe/whatever software capabilities for alignment. Then once you have encoded your cut file on the scanned material image, you could load that combined file to the glowforge, and it can now reference the bare material scan for aligning each individual cut.

So… your idea is possible. Infinite banner lengths will work. It is just a matter of convincing the software guys that this is a good use of their time.


#15

Not sure I followed all that, but we do know that the software will scan the previous work and align the image to itself, but how much overlap is needed well that we don’t know.
The alignment of the image they have said they are doing that in several interviews.
Aligning a long piece of material really requires a reference even if the reference is hard edges on the machine or if its software detection. I just hope that they give a long banner a go at some point so we can feel like it’s possible after all they say “as long as you want”


#16

If you are expecting to do a lot of very long things like this, it seems that infeed and outfeed flatforms may be a very useful addition, to help keep the material flat and level when its center of gravity is outside the Glowforge. And if you build those, you should be able to attach a guide rail down one side of it to help maintain a consistent feed line.
I plan on building a set for mine, but will need to wait until my new toy - umm tool - arrives, because they would need to be very accurately aligned to be useful. And removable/foldable too, now that I think of it, since we will need to open the front door to slide the bed out when extra depth is needed for something.


#17

Most of what I said boils down to what you said, but with the additional note that the scan of previous work to align the image to itself is also what you would use to align along the longer piece. So long as you have image the entire length of that piece. And then a nifty trick to work around the cases where you do not have image across the whole extent.


#18

We decided early on to focus on software that was common to both platforms first and dive deep on pro-only software later, so we haven’t pushed the limits of the passthrough yet.


#19

I think, as long as the software gives the user the tools necessary to align the cut coordinates to the material being cut, and as long as the material has a straight edge, you should be able to cut stuff as long as you want.

Say the roll of paper was 16" wide. It’s not hard to imagine that you could accidentally start it at a pretty crooked angle. I haven’t drawn it up or anything, but I could believe 5° of misalignment could happen.

(Preface: I’m going to use the word “should” a few times in my conjecture below. By “should” I mean “I believe it is reasonable to expect the machine and software to behave this way”. I don’t mean “this is an announced feature”.)

So, to take this one step at a time…
IF the software accurately measures the misalignment, it SHOULD be able to rotate the cut coordinates to match the misalignment. In other words, if the 16" roll of paper starts off with a misalignment of… say… 4.31°, the software will hopefully be made to accurately measure this misalignment and compensate for it (by rotating the cut by 4.31° as well). If this is done, cut #1 will hopefully be parallel with the edge of the paper, regardless of how crooked it was placed in the machine. As I understand it, rotation is something that is already part of gcode controllers. If the machine used gcode, this rotation could be done with one block: “G68 X0 Y0 R4.31”. It seems reasonable to me to expect the Glowforge to match this functionality.

As long as the first cut is parallel with the edge of the paper, it shouldn’t matter how crooked the paper is when you start the second cut either. We should be able to assume that the edge of the previous, truncated cut and the edge of the paper are 90° from each-other, but if there’s a discrepancy for some reason (slight stretch or a wrinkle or something), the software should give the user a choice of which line they’d like to base the second cut off of. In this case, the user would choose the edge of the paper because maintaining that parallel relationship is more important than having the seams of the cut line up perfectly. (again, hopefully they’re very close to 90° from each-other anyway)

So, as long as

  1. the first cut is accurately aligned with the edge of the paper,
  2. alignment of subsequent cuts can be prioritized to the edge of the paper as well, and
  3. the edge of the paper is straight,
    it seems like you should be able to go as long as you want.

If my assumptions aren’t true, a slight workaround would be to start the cut in the middle of your banner. That way the stacked-error of the initial misalignment would be cut in half.

… Forget anything practical, let’s imagine I want to cut a perforation through the middle of a very long piece of masking tape. I don’t know, let’s pretend I end up ripping it in half a lot and I just want to make it easier on myself. Impractical but illustrative, I believe. Let’s say the roll of masking tape I start with is 2" wide. If I stick the first 12" section of masking tape to the bed of the Glowforge, I would like to make sure it cuts it exactly 1" from the left edge of the tape (0.996" if we’re getting really technical). It’s going to be hard to align the tape; it’s sticky, I’m worried about messing up the adhesive, it’s slightly stretchy, etc. It would be a lot easier to put the tape down in the machine one time and have the software correct for any misalignments. If every cut is based on the previous cut, it won’t be long before the perforation is 0.5" from the edge and after twice as far the perforation will run off the edge entirely. This is the problem you’re anticipating, right @dpruitt ?


#20

Yes sort of… I used a cad tool and made a rectangle 10 inches by 30 feet and another to represent the 20 inch wide cut area of the glow forge. It basically boils down to two rectangles sitting on top of each other. If they are not aligned then at some point the edges are going to intercept. Thats the dilemma right now the machine will be able to see previous points from a cut those are on the image but unless that image is referenced to the edge of the material then it will converge with the materials edge at some point down the materials length. If software adjusts then it need to be able to edge detect at the 20 inch mark which is another problem, but I don’t know the space provided for machine components and advertised cutting area.
I did see Dans post here they are not pushing the pro version only capabilities, but maybe this post will get them thinking about it…