Cabriole Leg - Proof of Concept

[mezalick]

Great work.
Something I must try.
Michael

[jseiler]

I think this can be done if the model does not go past the centerline. I suspect the problem is in the translation. Something like other software is translating the xyz 3d model into a cylindrical model and that's more accurate than simply unwrapping. I'm thinking that it would be possible to modify the original 3d model to account for the inaccuracies created by the unwrapping/toolpathing process. I'm not enough a math geek to get that done, but I think I can do it by cutting the model as given rotary, measuring to see where its off, modding the original model to fix the distortions, then cutting again, repeat until its good enough.

If someone has a model of a cab leg to send me, I'll take a shot at it on my indexer. digitalcarvings at gmail.com

regards,
John

[chostet]

Im happy you posted this in reference to the video that was posted earlier. This helps he make a point that I have been trying to make for some time now. In a post a long time ago I asked what sort of toolpath strategy does a company like Enkeboll use for production? ( http://www.vectric.com/forum/viewtopic.php?f=27&t=7864). My point is this: I dont think it is efficient at all to use the raster toolpath strategy that Aspire is based on for production. Or more importantly, the raster toolpath strategy leaves all of the carvings looking a little rounded and bland while an Enkeboll carving has sharp edges comparable to a true hand carved piece. The video posted ( proves my hunch: At the 2:00 minute mark of this video you start to see the bit travel ALONG the shape of the acanthus carving. This motion simulates the stroke a chisel would make leaving the leaf detail much shaper- none of the rastered toolpath marks to sand out (or spend hours routing out). To me most of the Aspire carvings look "mooshy"... rounded over... does anybody have the experience to share what type of toolpath strategy is seen in the video? What is required for Vectric to utilize a similar strategy?

[chostet]

Here is the type of machine I would assume an Enkeboll would have: http://www.thermwood.com/header/brochur ... Gen2sm.pdf

This is a great video for my question.... http://www.thermwood.com/?gclid=CPC8q-O ... 7QodrD4kuQ . This is an 8 head 3-axis Thermwood production carving router that first utilizes a toolpath strategy similar to Aspire and THEN a carver traces a finished carving to store data into the machine for production detailing (see video.. it does a great job of showing this). Are there any other known tricks for this type of carving?

[dealguy11]

On something other than a true 3-d carving like the cabriole leg, it is possible in Aspire to create a profile path that goes around the carved part and cleans up some the edge (just make sure the path is all arcs, not bezier curves, or it will be rough). You can also use fluting toolpaths to carve veins and indentations, but this takes a bit of doing (although it will cut much faster than rastering).

Unfortunately, on the cabriole leg using my method, it wasn't possible to do a clean-up cut because it's cutting at such an angle down the leg. Also, the bit leaves pretty significant tooling marks on the sides, even in the smooth areas, that have to be sanded out.

The Thermwood video clearly used some kind of software that codes for the 4th axis directly, rather than wrapping. It looks like they do use rastering, but with the bit at more of a 90-degree angle to the work throughout, then they use fluting-type cuts to add detail to the leaves, which looks a lot more like what a hand gouge would leave. I agree that this produces a much cleaner look. The software that I've looked at that directly drives the 4th axis is generally more than $10k, which is a bit out of my budget. Someone mentioned other software in an earlier post, which I've spent a bit of time looking at today. It looks like it may possibly have more capability here - has anyone used it for 4-axis machining?

[gravirozo]

steve

im really glad you got trough with this project... theres a user, mr chuck, he also wanted to make a table with similar legs... he has also legacy router...

i started out modeling, and i got some lookalike result... however i was not far satisfied with... i think i sent him out the first results... but my main issue was after modeling sides, how i could really position the scrolls, and other decorations... it doesnt mean lay a flourish on the corner... that relatively simple...
the leg has some ""ridge" and that run from leg bottom up to sides, and back down on other leg...

i did not want answer earlier for this post, i was curious how you can solve it... i made two rail sweep as rails were on bottom and top, the profile of leg was the sweeping profile... and same vector i used for cutting around the model.. rotated the stl model and imported back 45 deg rotated... but did not posted because i did not like the result, also my router has only 4 inch under gantry, so it limit everything..

earlier i made a sample a clawfoot, and greg, from australia made testcut... but that also made with indexing...
this is posted here...
http://www.vectric.com/forum/viewtopic.php?f=29&t=6946&

another example a finial, which one works both way, with rotary method, and with indexing too... on this finial i calculated 1/8 of the round model to toolpath...

http://www.vectric.com/forum/viewtopic.php?f=38&t=8701&

and these samples for how i think to solve the toolpathing... for your leg i think as greg milled the clawfoot, that could be an answer...
the termwood videos showing they using a method, where tool moving out of the centerline of rotary... as the leaves bottom milled i think that is the """pencil milling""
the toolpathing very important because it can reduce the sanding almost to nothing... in toolrooms molds can be milled almost to polished surface... so those methods possible working for wood too..

about enkeboll looking... in local shop, i got a scroll as they wanted slightly different, and size different... as i earlier stated it just dont worth to mill to ready... it is too much time...

i hope you get forward more... aspire can make very easily flourishes scrolls... some help in the program would be great to ""assembly"" modeled sides...

keep up to posting... interesting to see different ways..

best regard
viktor

edit : modeling took about 3 hour... or so...

[metalworkz]

Hello Steve,
Great work with the Cabriole leg and an excellent explanation for the method you used. Thanks for sharing it with us all.
I wonder if the model could be made in sections with matching centers in the ends of the material and then each section could be wrapped and cut on the centers corresponding to the section(similar to a cam)? I have not done any wrapped toolpaths yet and was just wondering if that may be possible. Thanks for posting the pictures and information!

[esteeme1]

Here is the type of machine I would assume an Enkeboll would have: http://www.thermwood.com/header/brochur ... Gen2sm.pdf

This is a great video for my question.... http://www.thermwood.com/?gclid=CPC8q-O ... 7QodrD4kuQ . This is an 8 head 3-axis Thermwood production carving router that first utilizes a toolpath strategy similar to Aspire and THEN a carver traces a finished carving to store data into the machine for production detailing (see video.. it does a great job of showing this). Are there any other known tricks for this type of carving?

In Aspire 3 you can create a vector to highlight desired areas. At the bottom of the create V cut tool path options you can select the Project tool path on model (not sure of wording). The tool path will follow the contours of the model with no problem. It is definitely cheaper than spending $10K. I have used this option with text and vectors and it worked well with both. I think is was originally set up for text but as I said it works with vectors to.

[dealguy11]

The challenge here in creating a cleanup path is a couple of things:

1. You really only can clean up the leaf carving from the side view, and you need to use something small like a 1/16" ball nose to do it - a v-carve bit will cut off too much material. Creating the cleanup vector from the side view is a non-obvious task (to me anyway). I created the leaf blank by importing a component and laying it on its side - it is merged with the side component and "sticks up" through it. I need a vector that describes the interface or joint between the leaf carving blank and the side, and there is no obvious way (that I see - I'm way open to suggestions) to draw it.

2. You can't clean up the leaf from the front view, because the angles are not right. The bit comes down from the top, and it needs to angle in to reach the area that needs to be cleaned up.

HOWEVER, I do have some ideas that I'm going to try to improve the carving and provide more options for customizing the leg (inspired by Viktor's comments):

1. The easiest improvement to reduce sanding is to change the raster direction. My indexer is parallel to X, and I rastered the side views in X because that's the way they did it in the Thermwood video. This leaves major machining marks on the curved sides - even if you set the stepover small, you lose this on the sides as the bit plunges down and up. Rastering in the Y direction solves this problem. I discovered this on another job and just forgot it for this one.
2. If I can find a good STL editing program for not much cost (free is good), I will export the side view to STL, then, in the STL editing program, mirror the side and make a full 3D version. This will be rotated 45 degrees around the X axis. Then I'll re-import this into 4 separate Aspire files - 1 for each of the 4 sides of the leg. Each of those files should be flatter in the Y direction, and should toolpath much better. In addition, it should be possible to generate good cleanup toolpaths around the boundary of the leaf. Finally, you can add details such as beads or v-carved grooves, and they will not be distorted as much by the angle of cut. Downside is that now there are 4 files to cause registration problems, not just 2. I suppose this could just be modeled from scratch in another 3d CAD program, but Aspire does such a nice job making the shape with the 2-rail sweep, and it handles models from Vectorart 3d.

[woodmouse]

Very impressive.