Both are things that are supposed to work but aren’t working yet. Whether they will ever work, how accurate they will be and in what timescale is anybody’s guess.
6 Likes
Well, certainly not camera-based alignment.
Two-sided and pass-through operations will necessarily involve camera alignment. I think GF is banking on the head camera (supplementing the lid camera) to save the day.
3 Likes
Sigh…
OK, I get issues around manufacture and shipping - until you do it you don’t know… But why this is still even a question when the product is 18 months late really bothers me. Promised from day 1, but not working and AFAIK should be driven by software not by manufacturing.
7 Likes
I really think people are expecting too much from two sided printing. As you mentioned we can get something approaching precision creating templates using cardboard. I do that a lot, but I’m willing to live with the imprecision in my application. One would have to consider kerf for more precision.
I think it is expecting a lot of the cameras to think we’d be able to turn a piece over then align a design to it, especially if the design isn’t symmetric, and especially if based off current single sided experience
I am expecting it to do what I was told and sold when I paid 2 years ago.
Seriously, I do expect to get better than quarter inch alignment. I expect the much touted trace function to actually cut where the trace is - even at the edges. I expect to be able to cut through a material by flipping it. I expect those things because they were Day 1 promises.
I’m sounding grumpy now and at the moment I really am hoping these are not too many days from being resolved, but I am disturbed that units are in production with these fundamentals not yet working…
9 Likes
Please see my topic here, which was done with camera-based alignment, not by using a jig:
3 Likes
It seems your method would work if the offset in alignment is uniform across the bed. But we’re seeing that it changes considerably depending on location on the bed.
4 Likes
cynd and scott. Yes I can make manual tests and adjust every print, I can build jigs and stuff. But this is only fudging around an issue that ought to be working but isn’t. Neither of these solutions will fix the trace or the flip issues for example. Also doesn’t relate to my other issue which is rotation - an equally vital part of getting proper registered alignment.
5 Likes
You’re preaching to the choir, my friend.
6 Likes
We all sound grumpy right now. 
I am not pleased with the 1/4 inch offsets either and just wrote a note about that in another thread. The tool is predictable for the most part, but I haven’t used it enough and changes are still happening. I knew this when I accepted the unit.
Doll house around one of the things I’ll be doing too when I get a little more time. I don’t see any barriers to using the GF for that other than designing around the physical bed size.
My hopes in order are not having full access to the 22x20 sheet size and the new arbitrary power and speed scales that came with the precision power upgrade, followed by graphics resizing and graphics placement.
All that said, this is a really cool tool
1 Like
The facts that the software is still in beta, that Dan has said that significant improvements to alignment will be forthcoming, and I have a way to get reasonably accurate placement across the whole bed, satisfy me.
2 Likes
But doesn’t that speak to a problem already… Everything is always forthcoming …
1 Like
If you want true accuracy, jigs are the only way. Find a good jig material. I like hardboard. Learn the kerf. Design around that.
Optical placement for accuracy is a gimmick IMO and shouldnt be trusted. Im hoping they prove me wrong, but I dont think itll be possible to have 100% accurate placement in the glowforge v1 until they make way better use of the head camera, which they havent made use of at all up until this point, which is curious.
If theres one way to check if your camera guess is correct, it is to send the head cam to the spot where it thinks its supposed to be to verify. They arent doing that. Im not sure why…
5 Likes
@takitus, can’t use a jig for registering against pre-printed materials because the accuracy of the print is not guaranteed.
2 Likes
Can you give me an example?
You can do it as long as you have one straight side. I’ve done pencils in jigs that require pretty close tolerances and regularly think it terms of registering within the width of the kerf. Again, it’s with a jig and a way to register the preprinted material to the jig, which in many cases is just a piece of cardboard with a large rectangle cut out of it.
But your original question is valid: can you register a full bed worth of design by optical alignment alone without a jig and just by camera alone. For example, if @PrintToLaser had used some polyster label material that was printed with the design for the food tokens on one big sheet and laid it out on the bed and put the cut file of multiple circles exactly over each and every preprinted circle patterned out. Can it be done? @cynd11 had a long discussion once about regarding her cat stickers. She does it. She found a way to align. Is it happening the way everyone expects that you just zoom in and eyeball it. Until the lid camera image exactly reproduces without distortion a 12x20 image in the bed, in once sense the answer is no.
Is it possible in the future? It will most likely keep getting more accurate and closer to this scenario. Not having a timeline and having to wait so long does raise doubt for sure.
3 Likes
While you may be right, two sided printing, even with the ability to flip and cut to allow for thicker pieces is what was promised. The only way that works, or oversized through the pass through work is if camera placement becomes very accurate. For a lot of us that feature was a very big selling point.
6 Likes
To accurately align a cut to a printed image this would be nice:
Let the machine calculate the alignment, rotation… no manual zoom-in and arrow-keying…
Maybe one day this will be available in the cloud for a premium…
8 Likes
it won’t be the first thing over-promised.
I don’t see flipping non-symetric pieces and having the GFUI figure that out, but I can see it picking up on symmetric parts that are jigged into position, maybe.
Reading PG labels is still hit or miss on my machine. That should be easy compared to things that designs could throw at that GF that include shades of gray, colors, etc.
2 Likes
@sneggi Thats the ideal.
Actually, ive got phone apps that pretty much do that already… There are lots of apps that scan papers that wil auto-rotate and auto-crop to the photo of a paper on a desk for example, including doing perspective correction.
2 Likes