Thanks @palmercr, looking forward to your results.
To me there is a clear distinction between accuracy of the laser in ‘normal’ operation - i.e. from a vector or raster file and conversely, GF’s attempt at optical alignment via the camera.
I expect the former to be flawless since the hardware seems to be quality but any work involving the camera - either scanning images or aligning materials for cutting or engraving - seems still flawed at this point.
All of this comes back to the requirement for an absolute zero / homing point which has been requested by many. Currently we seem to be resigned to makeshift solutions of rigs & jigs & trial & error. If we’re not able to reliably cut or engrave from a known physical (calibrated) 0,0 point on the bed, this will hugely limit the unit’s functionality for us. And it will be a major design flaw.
Yes without limit switches the cameras need to be very accurate. But they also need be accurate for trace mode, double sided cuts and pass through, not to mention dynamic autofocus. So they are fundamental to the machine.
You are not, so you may want to consider a refund to purchase a system that better meets your needs. I’d hate to lose you as a customer, but I want you to get the system that’s the best fit for your requirements. I’d hate more for it to arrive and you to be surprised or disappointed.
I have been following the lid camera alignment threads a bit, although I am apparently not up to date with all of the results. I have seen a lot of hand drawn grids, and tests against that, but has anyone Vector engraved a grid to test the accuracy of that? It would seem that should be accurate enough to then test against.
I was under the impression that the head camera was eventually going to be enabled for precision alignment. Is this still in the hopper? That would seem to be more important than a true 0,0 alignment, but I’m surprised that is not also going to be an option.
If I want to do precision alignment on small pieces, one off, I use the camera. If I want to do precision alignment on large stuff in relation to a full sheet of material, I use a jig. It works for my purposes. Am able to do repeated engraves of pencils after closing the GFUI and removing the jig and then starting up again after doing something else.
I use factory edges for a piece of cardboard, after checking to make sure there is a square corner. One side goes against the plastic right edge of the crumb tray and the other butts against the front drop down door. I’ve also used a squared piece of cardboard and registered it against the edges of the garage door. The crumb tray and front door is easiest.
Then I cut out the areas where the pencils go or the clothespins, or the tongue depressors. I have also kept the same frame and then use a replaceable center rectange that fits with whatever frame jig I use.
That presumes the crumb tray goes back in at exactly the same spot every time. I’m not so sure it goes in so precisely. But no matter what you still have to drop your image somewhere, so it probably doesn’t matter. I haven’t made a jig yet so I can’t say for sure.
Thanks. I was thinking that was the best option but then my second thought was I wonder how it will register on the front door on the Pro units as that will be in the same spot as the pass through.
I did 100 tongue depressor rulers. Not all at once. 12 at a time in between various and sundry projects. Granted, I didn’t take out the crumb tray in between these, that I can recall. I have on some other projects. I have been wanting to do a video of this but when they pushed the bigger bed access, all my previous jigs are small. No big deal, it’s nice to have more room. And with a new GFUI being pushed, glad I didn’t make a video.
I just don’t get why you guys keep being stubborn about this. I consider the Pro to be a ‘prosumer’ product, not a toy. It’s like you decided at the start that camera alignment is the way to go because of the funky trace function. But you haven’t solved accuracy alignment with the camera and now it’s like, it’s a unique feature in lasers so we’re sticking with it, whether it works or not. But it doen not have to be either-or. We could have both. There are other ways to create a reliable homing point.
Many of your backers want this feature, Dan. Try not to be dimissive of it. It’s not an unreasonable feature on a laser machine. It’s one that enhances its functionality significantly.
I’m not - the system doesn’t have a reference corner like you’re asking for. If you wanted to solve the problem of “put this thing in a precise location” or “align things the same every time” or any other particular use case, we’ve done lots of work and have lots more to do.
But if you want that specific implementation of that solution, I can’t help you, and I want you to have a solution that meets your needs.
But help us then - what are those things? There are numerous threads and I imagine thousands of posts about this, most of the speculative, time-wasting kind. Could you please disseminate for us what is the current best practice for achieving accurate repeatability in the absence of a known 0,0 point on the bed?
I am wondering how much the decision to do only camera alignment is bolstered by manufacturing and assembly and calibration. I was watching some videos of other laser brands in operation and looked at the beds of the lasers and the registration setup. Not at all dismissing the value of having dialed in hard margins and a physical 0,0 but wondering how much that would add to the cost of production to get them accurate. I can see making the decision: we are going with camera alignment and we are just going to keep working on the software to keep improving. Perhaps there were trade offs. Certainly having the lid close accurately and reliably to allow the camera to register well is also a crucial component that takes time. Just thinking out loud.
I think part of the argument here arises from different perceptions of how something like this is supposed to work. For those well versed in CNC fabrication and those type of machines, I’m sure the lack of a 0,0 coordinate is frustrating. For me, where my use cases aren’t as fabrication centric, so my tolerances are higher, I like the visual alignment. Easier for me in many ways. I think this is where we can see how the Glowforge when through a completely different development process than other commercial lasers, by people coming in from a different angle, looking at a different market. Not saying those who would like the 0,0 configuration are wrong, or that wouldn’t be helpful at times, but I can see how the development process went off in a different direction.
Limit switches don’t normally need to be super accurately positioned, they just have to repeatable and then they define the origin and everything works from there.
However if the origin was defined by switches and fixed relative to the frame, you would still need to work out where the lid camera is relative to that.
I am surprised they didn’t put targets for the head cam in the corners to reference the lid camera to the frame. But that would eat some build area.
️