Sketch mapping enhancement [merged to assemblyMergeMaster]

[DeepSOIC]

Hi.
I think I might begin working on new ways to map sketches to something.
The planned new mapping modes are:
* mapped but disabled (to enable pocketing with arbitrarily oriented sketches)
* mapped to a planar face (the way it is now)
* mapped to a tangent plane at a vertex (for curved faces)
* mapped to endpoint of edge (the edge is to serve as a normal to sketch plane)
* mapped to midpoint of edge ... or even arbitrarily specified portion of length of the edge

Please let me know what you think! Especially ideas on the user interface of this.

[saso]

What are you smoking in Russia, your mind seems to be working in the N-th dimension?!

Don't take it bad, I am just joking, it sounds very interesting but I do wonder if this concepts are not too complex for a lot of users... Still, not a reason not to add it

[ian.rees]

* mapped but disabled (to enable pocketing with arbitrarily oriented sketches)

Isn't it possible to do this now using boolean operations? I do think it would be nice to have Sketcher's "choose orientation" box support manually entering centre, +x, and +y vectors, rather than the current workflow which requires position and rotation around a vector to get the sketch oriented correctly.

* mapped to a tangent plane at a vertex (for curved faces)
...
* mapped to midpoint of edge ... or even arbitrarily specified portion of length of the edge

For these two, how about this for UI:
* When a part is clicked, add a temporary datum at the last point that was clicked on a part - this could be on an arbitrary face, edge, vertex, etc. Maybe it could graphically look like a regular selected vertex.
* To create sketch mapped to that vertex/edge/face, user then clicks the "Create a new sketch" button as usual.
* If an edge or vertex was clicked, put the origin of the sketch on the temporary datum and make the sketch plane normal to the part's surface at the datum. Orient +x of sketch along an edge if it's there, or along axis of an extruded (not sure what the right word is...) surface. If it's not an extruded surface, +x goes towards +/-X of object space.

-Ian-

[NormandC]

Before going too far on this, you need to review jrheinlander's code in the jriegel/dev-assembly2 branch. That branch brings a lot of profound changes to the Sketcher and the Part Design workbench. jriegel has mentioned that for the Assembly workbench to come to fruition, big changes to the data model need to be made.

You can get an idea of what type of workflow to expect here: PartDesign Bearingholder Tutorial I

But that work is almost 2 year old now...

[ickby]

Before going too far on this, you need to review jrheinlander's code in the jriegel/dev-assembly2 branch. That branch brings a lot of profound changes to the Sketcher and the Part Design workbench.

I was about to write exactly the same, but honestly, the work on that branch is that slow that I'm about to change my mind. This upcoming changes (upcoming for multiple years already) severly limit part design evolution, and as there is no indication that it will not take multiple years again until it gets integrated we should go on developing one of the most used workbenches in freecad.

[cox]

Nice, can't wait to start testing...

I would love to read more about your plans and thoughts during development.

[r-frank]

* mapped to a tangent plane at a vertex (for curved faces)

Hello DeepSOIC.

@Work I am using Solidworks 2014.
Speaking of the mathematical side, there are two possible Solutions for the plan.
If I would define such a plane (sketch) in Solidworks, Solidworks would show me one
of the possible Solutions (in most cases the undesired one ) and then I would
have the possibility to tick a box to switch to the other possible solution (see screenshot).
[The tick-box labeled "Umkehren" does Switch the solution]

Maybe you could use the Planes created with the Work-Feature-Makro?

Roland

[DeepSOIC]

Thanks for input - there's quite a lot

mapped to a tangent plane at a vertex (for curved faces)... .Speaking of the mathematical side, there are two possible Solutions for the plan.

I didn't get what the solutions are from your pic, but I can now imagine a situation where it can happen. I'm not sure now, how to deal with such a case.

Before going too far on this, you need to review jrheinlander's code in the jriegel/dev-assembly2 branch. That branch brings a lot of profound changes to the Sketcher and the Part Design workbench.

Got it, I have to take a look there.

mapped but disabled (to enable pocketing with arbitrarily oriented sketches)Isn't it possible to do this now using boolean operations?

Yes, but you don't get links to external geometry in a completely detached sketch. But actually, I think it's better to simply enable links to everything, as I did for myself. In this case, pocketing is still a convenient way to apply a Boolean op, getting the job done in less clicks, and looking nicer in feature tree.

[r-frank]

Thanks for input - there's quite a lot

mapped to a tangent plane at a vertex (for curved faces)... .Speaking of the mathematical side, there are two possible Solutions for the plan.

I didn't get what the solutions are from your pic, but I can now imagine a situation where it can happen. I'm not sure now, how to deal with such a case.

Have a look at the attachment.
I said:
> Define new plane perpendicular to existing pink plane and tangential to cylindrical face
> there are two Solutions (normal vector of new plane pointing either in positive or negative z-direction)

[jeno]

Hello,

I would vote for mapping a sketch to the center line of a rotation-symmetric objekt. Would be helpful for the groove tool.
Thanks in advance