Feature Request: Numerical positioning

Is it ironic that not having limit switches because they can fail and damage the laser, causes the head to crash anyway?


How much of an issue would the encumbrance of patents and intellectual property be for some of these features. Is it like, "hey, there are patents here we might run up against and we just aren’t going to pay royalties but design a system as much as possible that is not encumbered by patents. I have no idea, but I expect due diligence has got to be a factor in this with a small company and the necessity of due diligence. Maybe it isn’t an issue at all

1 Like

Certainly a valid concern for any development project.

The bulk of these feature requests have either been something people have been able to do in other programs for a very long time (numerical positioning, for example), or unpatatently obvious.

Anybody see any outliers?

HA, I like your sense of humor.

This thread got muddled with the mention of limit switches.

I think numerical positioning in the setup stage within the GF app is a good idea though. It’s just one more little bit of flexibility and convenience particularly if you need to position copies ran in multiple jobs because the complex file causes issues with the GF app.

If the GF app would at least allow exporting an SVG file of a setup, then positions and scaling could be preserved on the user’s computer for future use.


A lot has been said in this thread about how it has been discussed before, and ‘decided’. For reference, here is a thread from 4 months ago with a few of the choice quotes that relate to the OP’s requested feature:

Absolute 0,0 reference point on the BED:

Numerical positioning within the GUI:

@dan has said no to the ability to use a reliable 0,0 reference point on the bed.
@dan has said that numerical positioning/scaling/rotating on the bed are definitely being considered.

A side note, directed to the forum elders:

Thank you for taking the time to have previously discussed, at great length, most of the topics inquired about by newer users. Your thoughtful prior discussions have added great value to this forum.

However, when you shout down a new post by claiming “we’ve been there, discussed that, you’re beating a dead horse”, it feels to me to be largely dismissive of people. This often leads to heated tempers, as has happened in this very thread, and isn’t really adding anything of value to the discussion.

A kind link to where it has been previously discussed would be nice, and would helpfully direct new visitors to valuable content they may not have known about. Otherwise, if you are not interested in discussing it again, perhaps you could just move on to another thread that holds your interest.

Personally, I see the repeated posts on topics as a valuable indicator to the community (and more importantly, to Glowforge) of the intense interest in that topic and/or feature and/or shortcoming and/or awesomeness.

So, maybe give the equine abuse rhetoric a bit of a break. :slight_smile:

(bonus points to the first person that quotes a thread that discussed this type of civility before)


I think the 0,0 thing is because of the way the bed was designed etc, and 0,0 would require a change to hardware from dans stated understanding. He’s seemed to be open to more modular implementations of this concept though.

The great news though, is that if the Glowforge supported numerical input, it would be a lot easier for us to implement a 0,0 setup ourselves, so I think that would be the biggest win here.

Still holding out hope for this!



The caveat is that I don’t know how repeatable the homing is over the long haul. Does it shift as the lens / G logo gets dirty, and back again when you clean it? Will improvements or changes to the cloud software change it? Will the upcoming changes to correct for the camera lens distortion change it? Probably yes to all…

User tolerance for those variances would vary based on use case. So, everyone’s mileage would vary.


I wrote a post a while ago about what I saw were going to be problems with optical alignment. The fact that the camera is attached to the lid and the bed isn’t locked in place means that all the homing has a lot of margin of error.

This can be fixed by affixing the bed to the case and allowing a homing process using the head camera in relation to some sort of fiducial on the bed.

1 Like

My thoughts on what Glowforge is doing.

This was my favorite “fix” that was proposed.

1 Like

does the bed matter though? shouldnt the camera be looking at something on the actual head to place it in the center 0,0 location (or whatever machine location the camera is located at)

I wonder how much of an issue the plastic body flex is causing. I was thinking about it this morning and the fact that the camera is mounted on a flexible piece of plastic that isn’t really attached to the actual machine portion of the machine must be causing some hard to solve issues.

1 Like

My proposal is to use the camera that is on the head of the laser for accurate repeatable homing instead of the camera on the lid. That way the laser could finish its job, send the head back to the resting spot, and take a photo at that point to assure that it looks the same as when it started. If it does it knows it’s in exactly the same spot, for every job, and hasn’t lost positional accuracy during a job. It could even potentially do this between sub jobs if necessary.

1 Like

The bed matters because if someone sets up stops on the bed to make their own 0,0 origin for placement of materials, but the bed is able to slide around a tiny amount, that means the 0,0 point of the bed can not be perfectly calibrated to the 0,0 of the laser head every time.

On the other hand users that know enough to see or need the benefits of a repeatable 0,0 will also see that it’s quite easy to alter the shape of the depressions in the case in order to make the feet on the bed index into them in exactly the same location every time.

This would be great, except that the camera can be off by 1/4" - Even on the center of the bed. Until they have better positioning, its a moot point

At first I thought I wanted numerical positioning, but then I realized that I really didn’t - I want alignment positioning to what’s on the bed, and typing in numbers is just how I’m used to doing this on other machines. If I could drag drawings around on the bed picture and have then reliably snap to the edges and holes in the material, it would be trivial to (for example) position a drawing against the bottom left corner of the material, for repeatable prints that would be correct to the material no matter how it’s positioned on the bed, or snapped against an existing hole so I could reuse material efficiently. And I’d like that a lot more than the ability to type in coordinates. It’s harder to implement, of course.

Of course right now the optical positioning is (for me) quit inaccurate, so this is pretty theoretical. But I can dream!

1 Like

Makes sense, i just misunderstood what was being discussed.

At the very least they could just add registration marks using the head camera; that is a feature already available on other machines

1 Like

You were totally right about the body having flex. It has an enormous amount of flex. They even went as far as to make sure I had a second piece of paperboard under the right side of my laser at MakerFaire when I had put a piece under the left. It was really thin stuff too. This was to avoid case flexion.

And yeah, registration marks under its resting point would be ideal. I tried to keep it less restrictive in description as I’m sure sometimes material would cover that area and they would need to account for that as well.

Can’t wait to get my prod unit so I can start building out some of this stuff


I was more thinking you can send your image to the glowforge with the registration marks on the fixture and it can read those with the head and know where to cut the rest. the lid camera can tell the head camera roughly where the marks are and then the head camera does it with precision.

You can find videos from trotec about using their camera with registration marks on printed items. So same idea but put them on your fixture instead

1 Like

I wasn’t clear, sorry. I was referring to initial placement of the file after rendering. It’s subsequent operations after the bed has been re-imaged that the lens distortion is introduced.

In those jobs where I had a dozen sheets of tokens to engrave on both sides the work is accomplished exactly where I initially placed the image across 2,000 pieces. After I opened the lid and flipped the tokens, the new camera image shows the design being displaced by about 1/4". That concerned me at first, but as long as the sheet hadn’t been moved, the second side engraved perfectly registered even though the camera image was off by 1/4" at the sides.

I have had one instance in 6 months where the engrave was off by 1/4, which I reported. I don’t know why, my guess is the described ‘drift’ from running a lot of jobs between power cycling the machine. It hasn’t happened since.
The initial placement dragging the file and nudging with the keyboard has been very accurate for me.

1 Like

I am somewhat confused as many folk are talking about an optical device (instead of mechanical) that would make a good limit switch and yet @dan is saying that he has the camera as the ultimate “limit switch” that is on at every mm in the range.

If there were a grid at every cm and the laser would aim to put a dot at every intersection you would have a very good idea of where and how much it was off across the entire bed, and from that you would have a clear adjustment that could be set for each individual machine.

By running such a test when the machine is finally setup any vagaries caused in packing,shipping, magnetic fields nearby, or any other remotely possible influence could be adjusted for.

In addition how hard would it be to simply report a x, y, & z number it is Using? Even if the working area was 4x4 inches it could still report the working area and the placement at that moment in the working area. However the centering is arranged those numbers would still be used during the cut and that “g-code” could be saved and sent to the active computer before deleating from the cloud?