How to increase cutting precision with Shaper Origin
Вставка
- Опубліковано 10 лют 2025
- I sell selected items on Etsy: creatoriq.cc/3...
This is a video explaining how to prevent deviations in cutting precision, when working with your Shaper Origin. The video shows how rotating your Shaper Origin can lead to significant deviations in cutting quality, which can ruin your work pieces. The video also shows how to prevent that.
More about the power tools and materials I use: bit.ly/401tKyv
Welcome to my woodworking channel. I am passionate about taking the most wonderful woods mother nature created and turning them into beautiful objects of day to day use. This is not a DIY channel about teaching the woodworking craft, showing you new and nifty techniques about how to build wooden objects or anything alike. It’s about celebrating nature’s beauty and providing inspiration for projects of your own.
Thanks for your support!
This behavior is readily explainable if you understand how the dominos are used by the cameras and machine vision algorithm. Each domino has a dot pattern that allows it to be uniquely identified and a respective location recorded. When you're scanning, what Origin is doing is creating a mesh map of the dominoes, each one's position being recorded relative to the dominoes immediately adjacent to it. Each of these relative coordinates has a finite level of accuracy and a respective positional tolerance. The more dominoes you have in camera view at once, the more accurately the position of shaper can be calculated. Where the "scan down one side and then back up the other" runs into trouble is at the turnaround point. There are a lot fewer dominoes in view, so the position and angle of the origin as it starts to acquire new dominoes up the second side isn't as accurate. As you scan up that second side, the angle is already off starting at the turning point and it just continues along with the angular deviation increasing as you go.
If you want to do something like this and get an accurate cut, don't scan down one side and back up the other, start at one end with origin facing the dominoes in a vertical orientation and try to get as many dominoes in view as possible and make sure that the you always have dominoes from both sides in view at the same time while scanning, this way you get the dominoes on each side fully linked to one another as far as relative coordinates are concerned. If you can't span the center and pick up dominoes from both sides, then you need to add another strip. When you get to the end, you can turn origin 180 degrees and scan back towards the starting point to capture any that were not visible at the starting point.
If you scan this way, you get a properly linked field of domino locations and it won't matter what orientation you use while cutting.
The way this was scanned in the video resulted in a long string of dominoes being referenced to each other, followed by a pivot that had fewer reference points and introduced angular error, then a long string of dominoes that were well referenced to each other but not directly to the ones on the first side along the full length, so the resulting location mesh was basically a long U shape instead of a fully linked rectangular field.
Bottom line, this isn't really a flaw with Shaper or how its tracking, its a limitation of machine vision when using relative location indicators. You see the same thing when doing 3D scanning when you don't have enough target markers in view, however that software tends to be a lot more rigid about reporting a "loss of tracking". There's nothing Shaper can really do in software to fully overcome the issue of reduced precision when a scan field is created the way it was done in this video. Any machine vision system would suffer the same tracking problems with additive error in this case. The only thing they could do is change the software so it refuses to link new domino coordinates unless it has more active dominoes in the field of view. If they do that, people will be complaining that they are being forced to use more tape. Maybe that's a better situation than allowing cases like this where the additive error becomes really problematic, but I'm not sure everyone would agree?
Thanks for the detailed explanation - I hope it helps a few users to prevent the problem
Thank you for bringing this to my attention. It explains some not quite perfect cuts I have had in the past.
Most welcome, Frederick. Sorry to hear about your not quite perfect cuts. Hopefully you can reduce these now...
Thanks!
Most welcome, and thank you for you “Super Thanks” 🙏🏼
Thank you for pointing this out. I had this happen and I just thought it was something I did.
It made me think that the tool wasnt everything I thought it was.
Most welcome - happy if this was helpful...
I just completed a relatively large calendar grid and noticed these variations at certain points in my grid. After watching your demo, I can say I did exactly what you did in your first pass. Now I know
Glad if this was helpful Jered! Hopefully you can prevent these variations in the future...
Interesting observation but I notice you setup the experiment with two sets of tape, one above the cut and one below. I wonder if Shaper would say this is insufficient. If I setup the same cut I would have used tape right through the cutting zone and simply cut through the tape. I think this would provide Origin with a larger and consistent scanning field but of course I'd have to give it a try. I like your channel and thanks for sharing your experience.
Thanks for your kind words, I am very glad to hear that my channel is helpful to you! As for your question: I am not sure I understand. You mean like putting tape on the area I cut on?
Yes, exactly. I do it when necessary and I've seen several other videos where it's been done as well including on some of the instructional videos by Shaper. I think the only downside is that the wrecked tape won't be as much help in guiding Origin if you need to go back and make continual cuts and adjustments so it may not be the best solution for every situation but it might work for the specific scenario you outlined. @@Woodensoul
I route over ShaperTape all the time, but it only makes sense to put tape on those areas if you need it at one point for orientation. In this case, I didn’t really need it there, because the camera wouldn’t be able to use it for orientation, given the angle
You are so right that happened to me while I did a cut out for hob they should have told us about this problem
Sorry this happened to you. Yes, they should tell us in a much transparent way...
@@Woodensoul exactly I reached out I said is my foul
This is very helpful. I am working on a project that uses a 4x8 foot sheet of plywood and requires cuts near each of the edges (which would require turning the SO, so it doesn't fall off the sheet). I will probably add a 1 foot strip of plywood with Shaper tape along the top edge of the plywood so I can always keep the SO pointing in the same direction.
Perhaps I could instead create two workspace/grids - one with the SO pointing "north" and the design anchored at the southwest edge of the sheet, and a second grid/workspace with the SO pointing "south" with a flipped design, so I can anchor the same corner of the design to the same physical corner of the sheet. I may experiment using your straight line test.
Thank you for pointing this out, and thank you for your great videos!
Thanks for sharing your experience - I know what you are talking about. The idea with the "extension" I have used in the past and it helps (but it is annoying). Your second idea sounds kind of crazy, not sure if I even understood it, but I wish you the best :D
To me, such effect happened when using _the_ _same_ dominos and turning the Origin by 90 degrees. I had to get a bigger sheet of wood and redo vital parts of the "Metric Vertical Workstation" by Friedrich before I could fit the parts together.
I suspect there's an improper deviation in the offset calculated between camera and the center of the milling cutter.
It doesn’t depend on using the same tape. As you see in my demo, it happens also when you use different tape (but also with the same, you are right. It’s the rotation of the device that leads to problems…
I would think if you filled in the gap with tape, even one row, you may not see this problem. When you do the scan you hardly have any of the upper tape visible as it transitions to the the lower tape, seems only a brief moment they are able to reference one another. Or scan down the middle first to get both sides in the camera.
I know what you mean, but that's not the problem. When I first found the problem cutting a c-inlay channel, I had the entire workpiece plastered with tape, and it still happened. Also, when cutting out a circular table, I had the entire workpiece taped, as you can see here (ua-cam.com/video/VnrwmvsKdmI/v-deo.html) and the circle didn't turn out to be perfect. I will try it out again next time I do a c-inlay, but I am very sure that your explanation isn't correct.
I read through the comments. Please allow me to offer some advice that may likely resolve your deviation. 4 things primarily afflict you. I will list them in order of importance. Camera motion blur either in-scan or not, is real and effects movement. Shaper Origin itself is a router and as-such is not equipped with a Phantom 4K, but commodity hardware. It will re-orient itself very well, but fast movements through sparse fields will introduce error (more sparse, the greater amplification of error). Secondly, the sparsity of tape, while you did mention in comments you have addressed in the past, does have an effect. However, just because you have enough tape does not mean you can neglect other points. That being said, sparsity acts as an error amplifier if/when you perform something that generates an error (such as moving too fast, regardless of whether in-cut or simply migrating from one tape field to the next in a rotation). Third, you should scan the way you cut. Shaper HQ recommends a waffle-scan, but I recommend a predictive scan - a scan that predicts the cutting behavior. If you are accustomed to turning at a particular corner because (for example) your dust collection or electrical cable requires you to, then mirror this in your scanning. Last, but not least, you cannot over-scan. Take the notion of predictive scanning and amplify it by scanning every area from every conceivable angle that you may exhibit during operation of the router (notice that I did not say “cut” - motion whether cutting or not is still effecting where the router thinks it is; it does not matter if the spindle is on, the router needs to look through the camera to see where it is, and it needs dominos, and it needs blur-free images, and it needs to have seen what it is looking at, and it needs to have seen it from that angle). If all these things are attended-to, then I see no reason why you cannot replicate the results I get in my shop making custom hand planes demanding tolerances of +/-0.001” or 0.025mm on almost every cut (and some of these hand planes are very large, coming in at 20” or 500mm long). Greetings from San Francisco.
Thank you very much for sharing your advice, Devin. These are very detailed instructions, and very clearly laid out. I might try this out to see if it works. But even if it works, I have to say that this is extremely cumbersome, it makes taping, scanning and cutting much more labour intensive, and it's not documented that way, so asking customers to figure that out for themselves is not ok. Putting tape where you need it, scanning it in whatever way you'd like, and turning your SO the way you want (as long as there is tape) should be a given. Everything that deviates from this isn't very user friendly...
@@Woodensoul I understand the burden and the astonishment that this burden has been placed on the end-user. The company has gone to good lengths to explain scanning, which is where the term “waffle scan” comes into the discussion. Last year they produced some videos that went into more depth than I am sharing. They go into details on how the dominos are mathematically linked and even show images conveying the interconnection modeling that the system performs. It was very enlightening, and it has even been said that Gen2 is less prone to motion blur due to a better camera, lights, and newer generation system-on-chip. Gen1 hardware was many years old (even if purchased last year) and as you yourself said, was pushing the envelope in such a way that there was zero competition. I do believe the technology will catch up with zero innovation from Shaper HQ simply by way of refreshing components from one generation to the next. The issue of course is keeping things compactly within a router (which gets easier as components shrink while also gaining performance). All the while, keeping it affordable (by not putting NASA level components or high-end Zeiss lenses and sundry things into the machine). I sympathize that lack of competition impedes development velocity, but I also believe they recognize that you have to innovate or die because competition is often born via demand without supply (if they stop delivering what people want, a competitor will surface to create supply to satisfy that want). As for Trace, I signed up as an early bird. I should be getting it in October. It will solve a need I have and also make some things much easier. I plan on enjoying it mostly with Shaper Origin, but it’s hard to see a down-side to it in-general. I just hope that Trace does not take time away from Origin development.
I am curious: What need does Trace solve for you? I have a hard time understanding that. I would be happy to buy it, but I don't understand what I would use it for...
@@Woodensoul Tool foam is the first thing I need. The Bridge City Toolworks CT-6 hand drill came in a very nice box but in the late 90’s they used foam that is all but rotted out 25 years later and needs replacing. Newer foams are more stable. I need to trace the original foam to mirror the cuts in new modern foam. This will help me complete the restoration and improvement of this tool to hopefully pass down to my child. Basically any time I want to copy someone else’s work (for whatever reason; in this case because the antique is no longer made), Trace promises to make that a simple proposition. I am sure many uses will come forward once I solve that need, but I really needed it for that. After that I plan to use it to make shadow foam organizers for my multiple tool boxes.
@@devinteske Interesting. I would solve that need by taking a photo and tracing the shape on a vector tool. But then again, that would be a one off case, so if I would have to trace hundreds of those foam shapes, I would certainly buy Tracer.
Great video as usual my friend! This May explain why I was getting inconsistent cuts when attempting to cut my circles for my drum project.
Thank you, Henry. Yes, true, that could explain these problems, I didn't even think about it...
I called shaper support recently and described the not-perfect-corner that you showed. They told me to always use the auto mode where the machine will go around corners on it's own but that I should not be moving the shaper while it was automatically moving around a corner. I tried that and it then produced the perfect corner I was looking for. Just a minor quirk I guess. Maybe this small stuff bothers us because in many other ways it is such an amazing machine.
I would be very astonished if that would do the trick here. But many people left proposals on how to prevent this, so I might need to run another test soon...
@@Woodensoulcorrect. His comment has absolutely nothing to do with this issue.
Good info Javier, thanks.
Thank you Chris, I am pumped if this was helpful
Have you submitted a ticket? I noticed accuracy issues since the OS Jenner was released, the last issue I had they acknowledged as a bug in the OS. Great video, thankyou for highlighting the issue for us
Hey man… thanks, happy if it’s useful. I have not filed a ticket, assuming they are aware already. But now that you say it, I will file a ticket and see what happens
@@Woodensoul Did you get any response after filing the ticket?
@@dworkin7110 I never filed that ticket - shame on me :D
@@Woodensoul Well... I ordered a Shaper Origin today, a couple of hours ago partly after (and because of) watching your videos.. I even ordered that set of clearing bits from Amazon UK (thanks for that). Heads may roll sir, heads may roll 😁
@@dworkin7110 Hahaha... you will love it, just make sure to pay attention to the details mentioned in this video, then you will be all set :D
Thank you, very informative, as always!
Most welcome, and thank you for the positive encouragement 🙏🏼
I noticed this error too, when making mft tables. I could be wrong but in some of the early days of the demos you can see scanning and cutting in multiple directions. This really needs to be addressed. It makes working on the edges of large pieces really difficult. I’m still not even sure whether this means you need to always scan in the direction you intend to cut or that all dominos in a workstation need to be scanned in the same direction. Which is it? I love my origin. But I’ve never used a tool more prone to supposed user error. The tool was really marketed as a way to work at any scale and not being able to accurately work on the edges of large workspaces is a real big problem.
ua-cam.com/video/D-rax7CWI-Y/v-deo.html
Cutting and scanning in multiple directions and angles 🤷🏾♂️
My opinion: The scanning direction is not that important, but the cutting direction is. One should cut with the SO facing in the same/very similar direction. Now if you read some comments, there are all kinds of theories shared, so I don't claim to have the perfect solution. And yes, agree, especially working on edges becomes really cumbersome...
There is a large gap between the sets of rows of dominos. I wonder if there is a slight blip in the scan when the camera traverses this gap. I wonder what happens if a few dominos are placed in this path to create more precise path for the scan to follow when it is turned 180 degrees during the scan to traverse one set of rows to the opposing set of rows? It also might make sense to do it at both ends too because a small miss in the angle between the two sets of rows gets magnified greatly given the length of the rows. Is it really a bug, or is the amount of positional data Origin is taking in during the spin from one set of dominos to the other is at outside edge of what it can accept, but then this minimal criteria is all it has to compute an angle that in this case extends a long distance, magnifying the error greatly.
I know what you mean and it might be that adding tape could reduce the offset a little, but this problem is something I have confirmed over and over again. The best experiment you can run is when cutting out a round table out of a square piece of wood, where you naturally have to face the SO always towards the center of the circle. You always get an offset where you plunge in, even if you plaster the board with tape…
Thanks I’ll remember this 😃
👊🏼
I ran into this problem when cutting out a hole for the dust underneath my table saw. I'm still happy with the SO but I think as well that they should fix some major bugs before developing gimmicks like Trace. Well, without competition, there's no pressure to do so...
Yeah man - competition keeps people on their toes, no question about that...
Wow they really should fix this!
Agree, not sure why this is still happening
Keep the router pointed toward the tape you scanned first.
Yes, that’s the point of the entire video 🤣
Thanks, I thought the deviation I experienced was my mistake.🙏
Most welcome - we need to stop blaming ourselves and start blaming the companies who make suboptimal products 😆
To sum it up: While milling, hold Origin in much the same way you scanned the dominoes. Thank you for explaining this so well and I'm curious if Shaper will notice the problem and what they do. Many greetings from Germany…
Good summary :D Most welcome, and thanks for watching. Greetings from Switzerland...
Great video. Useful information. Mahalo for sharing! : )
A'ole pilikia my friend - happy if this was useful to you :D
I have the same issues when I cut slots for cabinet sliding doors.
😩
wow.. good info here.. Im a 180 degree scanner no more! btw.. did you pick up Trace?
Just to be clear: You can be a 180 degree scanner, but not a “180 degree turner while working with the SO” 😆. Regarding Trace: Depends what you mean by “pick up”. If you mean that I am aware about it: yes. Id you mean that I have invested in the kickstarter or that I am interested: No 😆
@@Woodensoul I feel like the trace is a cash grab for hardware. A modern apple or android phone should be able to do this at full scale with shapertape. Even if you had to pay for the feature/app.... There should be no need to rely on an aluminum rectangle when their $25 tape should get the job done with the same accuracy as scanning for a workspace.
I feel similar about that auto pass nonsense that I need to pay for unfortunately. Grifting for sure
@@FPVtrix Oh right! I didn't even think about it that way, but the aluminum hardware is a) unnecessary and b) limiting (as it limits the area you can use for sketches to a pretty small rectangle). But se that aside, I rarely have even the need to move a sketch to SVG, and if I do I typically just trace a picture of it on the computer. So clearly I am not the target user for this product...
@@FPVtrix have a lot of experience with all kind of tape quality, professional high-end painter for over 30y. I can tell you that the quality of the tape is below zero. Buy the cheapest tape you can find and sell it for over over over priced money. I would never buy this kind of tape for my profession.
@@TheOneLin3 I know that preserving the Domino scale is paramount so even when I don't use the tape and opt for printing the dominoes on paper, my cut accuracy is only as good as the printed image. I like it cause the process saves me from buying the tape for every job. I even got an 11x17 printer so I can make sheets big enough for small work spaces like the shaper plate.
"Doctor, it hurts when I do that." - - "Don't do that!" .. case closed, no bug fixes required nor entitled.
as an engineer I see 2 possible reasons:
- Domino tapes are too far apart
- Surface unevenness
People think its the distance of the tapes. I might need to repeat the experiment one day.
This is not about orientation, it's about the tape layout. The big tape gap in the middle of the table means origin will have trouble precisely connecting the two "islands" of tape that do have good overlap. Adding some more tape would make sure they are connected better and that an accurate workspace is created.
Now I do think they can do a better job with education and then warning the user of the unreliable scan, but it's physically not something that can be fixed on their end. The tool is being used incorrectly here, it's kinda like putting a saw blade backwards
Yeah, blame the user, my favourite excuse for bad product design. So first of all, I don't think your explanation is correct. When I first encountered this problem, I was cutting out a round table. As you can see in this video: ua-cam.com/video/VnrwmvsKdmI/v-deo.html. You see that there was a good carpet of regular tape, yet the shape didn't turn out to be a perfect circle. Also, when I encountered the problem for my first C-Inlay, I had put tape throughout the cut zone. Still I will make sure to try out your recommendation next time, but I am highly doubtful that you are correct, especially given how many people commented on this video that they have encountered the same. Second, your claim "it's physically not something that can be fixed" seems also very fishy, at least I will question your expertise in this field and your assumption that this is impossible to fix. Why would it? "That just like your opinion, man", would the Dude say...
@@Woodensoul A big part of my comment was about how Shaper can do better with informing the user, but ultimately the machine can only work at the highest level of accuracy when set up correctly and they can't do that for us.
This island problem is just one way a scan can lead to bad results, there's a "tape best practices" video that I thought was very informative.
This may end up being an issue on their end, but I think that kind of claim requires a little more testing.
@@Woodensoul And about the physically impossible comment. If two pieces of tape are not close enough to be viewed in the same camera frame, their relative position is impossible to compute. A better phrasing would be that given the way this machine works there's no way to make some tape layouts work out.
@@gharel396 Agree, the minimum they would need to do is to inform the user that the tape zone is incomplete or of low accuracy. But based on my experience, I am not sold your hypothesis is correct. I will test it further though, because it would be nice if your theory is correct...
@@gharel396 On Minute 1:29 you can see that the scan did cover both left and right hand side tape in the same view, so your theory would be that if it's in the "outer parts" of the camera view, then these are of lower accuracy or quality, which is something I haven't heard of, and something that would be weird. Why should that be of lower quality?
I am new to Shaper, but I have seen many videos of large cuts that are 100% perfect. I think you should try to rather scan the way that you intent to make your cut. Scanning one way and cutting a different (turning 180 deg) will cause inaccuracy.
Welcome to the Shaper family. Yes, that’s exactly the point of this video. But you will see that cutting in scanning direction isn’t always as trivial as you might think, because the size of your work piece sometimes limits what you can do
Good info. So never turn the Shaper 180 degrees and use other dominos...
Yeah... if you can prevent it
Why cant the shaper just follow a line you just draw on the workpiece...?
You mean that you draw it on the wood, and it auto-magically follow that line?
You had too big of space between the rows. If the cut field had tape 2-3" appart, probably wouldn't have had any issues.
That's what many said. I will repeat the experiment one day and see if that is true...
First, I like Festool and Shaper. But, are you surprised? Shaper is owned and controlled by Festool, a very, very haughty German multinational corporation. Overselling a product? Why that's not a problem or mistake, that's a primary marketing method!!!! Additionally, you have to admit that the Shaper Origin is nearly a man-toy. (I, like you, own and like my Shaper Origin so please don't be offended. ) But don't all toy manufacturers deceive their customers in their advertising and marketing? That toy I begged my parents for never did what it did on TV.
I don't disagree with anything you said. But just because it's normal that companies overpromise and underdeliver, that doesn't mean that customers should not be allowed to point out these discrepancies. For instance, I also expect politicians to lie all the time, but just because I expect that, it doesn't mean I should not call them out if they get caught lying.