Pass Though issues

Ok. I went for broke tonight and tried the pass thru feature tonight. After a lot of try and fail, I got the laser up and running.

What I got the GF for is my hobby. I built warship models with high detail, 3D parts, and they are remote controlled. The structures are made from laser cut gator board which I have been sending out to be cut costing a lot. I bought the GF to save me the money and cost of travel.

The issue that I am asking for help with is the photo of the material and alignment of the parts. There is a curve effect coming from the camera that is making it very hard for the parts to align on the bed I dont know if it is because with the part being 48 inches long, its going through both slots, and causing it to be above the bed.

On my second cut, I took a chance and aligned it the best I could with GF camera image. I was close, but my parts can not be this far off.

Does anyone have tricks? I really hope the software works better than what I am doing now.

Thanks
Eric

18 Likes

Ran into the same thing myself yesterday. Unfortunately until they get the alignment issue resolved, I think large jobs are going to be an exercise in frustration.

The GF crew might have seen this request from my previous post, but I’ll repeat it here just in case:

A) Post a “known issues” section that prominently displays itself in the app. It’s no fun running into a bug or a limitation, but it’s even worse discovering that it’s something that was known beforehand and could have been avoided/worked around. Digging through hundreds of forum threads and taking notes isn’t a great solution.
B) Post a detailed changelog so we can know when these issues get resolved.

9 Likes

There’s a tutorial for using fiducials for alignment: Passthrough Alignment Tutorial with Manual Indexing

@palmercr conjectures in that thread that you could do something similar with a sawtooth cut into the jig and then progressive sawtooths into the material to match up every 10". That is, you cut the next sawtooth into the left edge of the continuous material as you cut out that section of the material. Slide it forward 10" and place it on the jig sawtooth and cut the next frame along with the next sawtooth, etc.

10 Likes

The bending of the material should be counteracted by placing in- and outfeed tables. You could laser some yourself. Or way the material down in the machine using something heavy but sacrificing cut space.

2 Likes

Having thought about it some more: a sawtooth would give precise location in the Y direction but be kerf dependent in X. Dowel pins are going to be kerf dependent in X and Y.

A pair of tapered dowel pins with straight shanks that go into a linear bearings (with preload) in the bed would be precise in X and Y.

1 Like

Some of this might be solved by one of the indexing methods. Some is definitely caused by flex if the material is not supported in and out. As it lifts the material from the honeycomb, it alters the optical and laser focus to hit a different spot than if the material is flat. Material being flat and correct thickness are two of the first places to look for any alignment issue if the laser is not hitting the right spot or cutting through. (Even with Proofgrade, material being flat is the first place to look for troubleshooting cut/engrave issues.)

2 Likes

Although I have not been able to experiment with mine yet :neutral_face:, some people have talked about checking alignment by doing an initial pass at 1% power, just enough to leave a mark on the masking.

2 Likes

I love what you did with the marker on the outside of the cardboard. Did the board raise up a little causing any issues? Also, I feel like I need some bridges to keep my parts connected so it will not have issues going through the other opening.

Thinking that that rubber flap is going to be an issue.

Eric… Nice model mate. I purchased my Pro simply because of the pass through feature and optical alignment. I got mine to build large RC planes with. Unfortunately the features that I wanted the GF for are not working and based on what I’m seeing may never work correctly.

After spending 4K on this machine. We should not be having to do work arounds with indexing etc. The machine was supposed to do all of that automatically!. I cannot even place smaller parts in the top area because the camera cannot even show the material and seems to be short by about 1.5" in the upper scan.

Personally I am very disappointed at this time and have not seen much if any improvement in performance and especially optical alignment.

3 Likes

What exactly are you seeing that would lead you to that conclusion? That seems like a pretty fair jump unless you are privy to something that the rest of us aren’t.

Don’t get me wrong - I sympathize with the passthrough software not being enabled. I have a Pro as well. And Dan has stated as much that he also gets frustrated with it not being enabled yet.

Drawing the conclusion that it will never be functioning after only having your machine for a few weeks seems to be quite a stretch though.

I have no idea what their plan is for the passthrough software, as far as how it will work from a designing perspective. The current methodology is that designs need to be broken up into bed-sized chunks, one chunk processed and then onto the next - which is essentially software indexing. Even if the camera alignment is flawless, I have a feeling anyone with existing designs not conforming to that design methodology (which would be no one) will be upset because they need to configure their files to work using that method.

As it is, yes, you have to index manually. Though, I can repeatedly hit within a kerf with a simple piece of tape attached to the crumb tray. Not saying that’s the way it oughta be - but it’s not horrendously difficult either. It really simply involves sliding the material through to match a mark.

Maybe they’ll work some awesome software magic and designs won’t have to be broken up. Out of curiosity, how does one design for other laser passthroughs?

1 Like

Yes it’s supposed to do it automatically. You agreed to accept beta software that noted this feature was not working. If it was critical then deferring shipment and waiting for it to be delivered & moved from beta would have been more productive. Accepting it with known limitations and then pronouncing it unacceptable because of those known deficiencies is nonsensical - you accepted it in its beta state, ergo it was acceptable. It may not be acceptable for your ultimate use case, but that was why you were told up front what was not working.

I use a $9000 ShopBot that requires manual indexing. (As does the Redsail laser.) I don’t expect anything differently because they don’t advertise ever intending to deliver auto-alignment.

Nor do I expect my Pro to arrive (by end of year) with auto-alignment. I accepted the beta limitations. If it never comes and they declare it no longer beta, then I’ll be unhappy & rightfully so.

4 Likes

JB… My point is what I’m not seeing. After two years plus of development the optics should have been complete by now. Lets face it the optics is the heart of this machine. And yes its supposed to be Beta. Well, after this amount of time it should be production and fully working. Also, the snake oil salesmanship was to be able to flip a part over and cut the other side. Again one must do a work around for this.

All I see from the GF peeps is its in the hopper. It seems to me that the hopper is also broken cause not a lot comes out of it. The machine is very pretty and it cuts well. I would not care if it was a tin box and rusty for as long that it worked to specs. For that kind of money I feel kinda ripped off at this stage.

The reason I suspect that we will never get good optics is that by this stage of the game that should have been the first thing that actually worked besides the cut side of things. No matter how many times I rescan material or move things around I cannot see the top part of the bed which is ridiculous to limit placement space and so forth.

2 Likes

Parallax is the issue that GF must solve.

Simplified, the alignment problem will only be solved if:

  1. they can move the camera over the bed
  2. install multiple cameras
  3. and / or calibrate EACH one of their single, fixed position camera lens (for positional aberration) to then be able to calculate parallax for each lens and at EACH point on the bed where the laser is cutting, and for EACH material thickness.

That is a lot of combinations / calculations.

It is not a trivial problem to solve, but the mathematics are well documented. .

If you don’t know about parallax distortion, Google it. A photo demonstrating this concept is attached. Pretend each eye is the laser. Will the laser hit the same place? Thats the problem.

There are some good suggestions using jigs to overcome the current limitation.

Parallax

2 Likes

@eric2 Do you have a website that shows your work!!! Cool!!!

BTW: There are multiple cameras. The head camera which is to be used for aligning the pass thru and the original double sided cuts has not yet been enabled for those purposes. Still TBD. Yes, the positioning accuracy of the lid camera needs to improve but was never intended for those high accuracy purposes.

2 Likes

One source of frustration is that the reason that we were given for multiple delays is that @dan wanted to make the :glowforge: as good as it could be, rather than delivering a half-baked product.
But now we have to accept beta software, just to get it two-years-plus late, and for something that they admit is “flaky”. And it does not yet do much of what was promised. Which is exactly what they said the wait was to avoid. Not. Cool.
I still want it though, as it is miles better than what I currently have, which is nothing.:neutral_face:

2 Likes

I cannot speak to the pass through alignment, but not having used other equipment with similar issues; meaning making parts larger than the machine bed I can offer you some work around suggestions.

For example, I just made a 15’ kayak and we know that plywood only comes in 8’ lengths. I had to join the pieces together somehow. There are many options here, scarf joint, puzzle joint, stepped scarf, butt joint, etc. Your models seem similar in concept.

The glowforge will cut accurately within the 12" x 20" boundaries. It might cost you a bit in yield but arrange your work so smaller parts are never outside the boundaries. I see in the photos above that you have mismatches in small parts. Avoid this with care layout work.

I would recommend you create puzzle joints on long parts, then tape and feather the joints or glue a backer piece on the inside to strengthen the joint. I am sure you do this already and know how to do it better than I would.

Sorry if you were just venting as I understand the frustration. But the steps I mentioned will get you 90% to where you need to be without much fuss.

My understanding is that GF is committed to essentially all of these. There are already 2 cameras (at least) in the unit. The main camera is on the lid. There is another on the head that will ultimately be instrumental in a lot of the high accuracy work. That is your item 1 and 2.

For item 3, one of the recent updates indicated that they are taking very many calibration measurements, etc, from each unit before it goes out so that they can apply those corrections when the software reaches that point.

2 Likes

I for one do understand the Parallax view issue having worked in the military flight simulator business and aerospace on helmet/weapons tracking etc. Regards scarf joints. Yes, I had already planned on that method even though it will make wing ribs somewhat weaker and will have to be reenforced across the joint.

My concern is not that one cannot find a work around since you can for most everything. My concern is that I do not see much of an improvement on the optics side in the last nine months or so. Example… If you go back and look at the video by Adam Savage from tested who was given a prototype GF before anyone else. You will see in that video that the optical alignment is totally off from the cut placement file. This is still the same alignment issues that are still with us today.

Another issue with using fixed cameras is that each electronic device will have a certain amount of drift based on ambient and or working temperatures which also adds to difficulty in alignment. It makes you wonder if it may be worthwhile placing a fixed bed referencing graphic/scale of some type that the cameras/SW can reference to?

So again… My concerns is that not much that I can see is being done on the actual development and improvement of the SW. I read a lot that lots of work on the manufacturing test side is in progress. Regards agreeing to a Beta unit at this time. ARE YOU KIDDING ME!!! After two years plus wait do you think that most everyone would agree to anything that GF says just so they can finally get what they paid for. By having to agree to a Beta unit after so long a wait is a way that the GF folks can manipulate us once again. Its called good marketing and promises. Look back in history of all the vids that have been posted. Not much difference today that was being shown in maker vids two years ago.

Its seems that a lot of effort is put fourth in to making proof grade products available. I for one will never purchase any PG products due to the prices being outrageous for serious parts making. At this time I would rather see the effort going into making the product actually work the way we where originally sold on.

I find it unfortunate that many people see Dan has a Messiah or similar and hang onto every word that he says and go along with whatever he tells you all. In reality we have been duped from day one with his never ending promises of delivery that never came about until two years plus later. So do I believe that the items in the hopper will get fixed and or implemented in a timely manner… DO NOT THINK SO.

I’m sure most of you reading this will think that I’m a Pr&*^&K and that is fine. I purchased this product based on good faith to what was promised up front. That is not what I received. But I will say that the physical hardware is very nicely designed and looks to be quite robust so that is something to cheer about. However, I would have rather have had a metal box that was functional with all the features working than a fancy cabinet with flaky functionality.

Just my view…

2 Likes

You did not “have to agree” to Beta. You were given the option to wait. Some in fact are waiting. Would it suck? Sure. It’s another 6 month wait on top of the past 2 years but having made that decision and then bitching about the software isn’t fair. You most definitely could have waited until all you dreamed of came true You chose not to.

I don’t have a kool-aid mentality around Dan or his performance as a creator of something new. I would really really prefer this was all complete now and would have preferred it was ready 2 years ago. Ain’t gonna change history though - it wasn’t, it isn’t and I’m not gonna make it happen. I either have to believe that they’re going to get where I want them to go or I have to move on. So yes, I believe, And yes I understand the marketing and the business decisions and strategy going on here. But at the end of the day, it really doesn’t matter what you want or what you think - what matters is what is and if reality doesn’t match your assessment of the way it should be, too bad because reality trumps wishes 100% of the time.

You want to complain about PG material not automagically cutting, go ahead. It’s fair. They said it worked when you clicked Yes on the shipping email. But you want to complain that something isn’t there that they told you didn’t exist yet and you could click Wait until it was available, that’s just churlish.

2 Likes