Michael,

many thanks for all this info, especially the layer equivalent in MOI ie. styles and for the info on both version.

I want to mention again that I'm very impressed by MOI. I consider myself to be very newbie at this time and I can only imagine how productive I'll be with more experience.



Is it the same for sweeps, lofts, etc? I know, I'm a very curious guy and I figure that if some operations are more like math function and others are more of an iterative type, the result of these ops. may differ in relative quality. For example, I notice the 1 rail sweep I used seem to proceed by strait line segment to approximate the curve part of the rail. If this is the case then I would ask is there a way one can "control" the step size or would it preferable to use another operation if possible of course.

I use a join operation on all the surfaces of my shape and MOI reports it's a joined surface and not a solid as I would have expected. Would it be better to use a Boolean union instead?

Another observation, the rails I used are made in 2 parts, a strait line and a some spline I would believe joined together. When I select the original rail it's all selected but when I try to select the edge of the sweep surface, Moi select only a single part at a time. This behavior makes me curious to understand what's going on.

Regards,
Felix

Burr,

thanks for your input though I think I messed up bit. I requested your input because of your knowledge of BobCad/Art. What I was trying to say about "drawing" the medallion is yes I could use Z Surf especially if it's "organic" and it could be done in MOI if the geometry is relatively simple as your example, maybe using the method I describe earlier and what I didn't really mention is that it could be done in BobArt as well.

I think I answered my own question in a way and I'm sorry I bothered you with my self questioning.

Regards,
Felix

"A toolpath would choke on a multi-million polygonal count, where it would be a breeze on the same NURBS surface, to acheive the same "curvature resolution", at say, .0001."

An older machine with a hardware controller would choke on that, but it also would not be using splines either. With a modern PC based CNC controller, the file size is irrelevant as long as the PC can send the data fast enough to the machine.

Of course using smaller linear moves increases the file size of the toolpath. However, just because the toolpath file size is large does not mean it cannot be machined sucessfully. Perhaps your system uses arcs and straight line segments? Using arcs would account for a smaller toolpath file size.

Can you post some of the G-code?

I'm speaking about "Generating" the "toolpath", not the gcode that comes from it or the actual cut at the machine. The program that he is using tops out at around 100 MB... (Trying to bring in a 100 mb + STL usually doesnt work, let alone, toolpath it..)

Yes, the software he is using has a "fit arcs" selection, which will turn a 30,000 lines of code file into a 6,000 line file, but again, this is not really the area I was refering to...

Hi Felix,

> Is it the same for sweeps, lofts, etc? I know, I'm a
> very curious guy and I figure that if some operations
> are more like math function and others are more of an
> iterative type, the result of these ops. may differ in relative
> quality.

Loft is something that's calculated directly, but Sweep does go through an iterative fitting process, in a way the sweep is kind of like an automated loft that cooks up a bunch of profiles starting with a small number and measuring how accurate the result is and adding more profiles until the generated shape hugs the rail curves to a small enough distance.

But the error of such iterative functions is checked, in MoI the refinement will continue until the result is accurate to within 0.001 units.


> For example, I notice the 1 rail sweep I used seem to
> proceed by strait line segment to approximate the curve
> part of the rail.

Well these iterative methods actually produce curve output, the curve just gets more control points in it as it gets more refined.

The only way you should be seeing straight line segments in your sweep is if your rail curve is actually made up of straight line segments instead of being a curved shape itself.

Can you please post an attachment with the 3DM model file of your sweep so I can take a look at it and see what you are referring to?


> <...> or would it preferable to use another operation if possible of course.

Well, which tool is best really depends on the situation a lot - it would help a lot if you would post the model file to give some more specific context to what you are describing.


> I use a join operation on all the surfaces of my shape and
> MOI reports it's a joined surface and not a solid as I would
> have expected. Would it be better to use a Boolean union instead?

Well, that again depends on what particular problem you're running into - do you possibly have a surface missing leaving an open hole in the model? If so then for a situation like that you need to model a surface that fills in that hole and then join that in as well in order to make a fully closed skin. If you could post the 3DM model file that has your non-solid joined piece in it, that would help me to give you some more specific advice.

Boolean union (and the booleans in general) are more oriented towards intersecting objects with one another and removing some material from them. So for example if you have 2 cubes that are partially pushed through each other you would want to use boolean union which will intersect them and remove the area that is on the interior.

If you have a set of surfaces that all touch each other at common edges, where no material needs to be removed from anything then that's when Join is the right tool and not boolean union, although boolean union will often work the same for that, it will just try to do a lot of additional work to intersect things as well though.


> Another observation, the rails I used are made in 2 parts,
> a strait line and a some spline I would believe joined together.
> When I select the original rail it's all selected but when I try
> to select the edge of the sweep surface, Moi select only a single
> part at a time. This behavior makes me curious to understand
> what's going on.

Could you please post the model file so I could load it up and take a look at it?

If you have a rail curve that has sharp corners in it and is also not a planar shape, then that can cause the sweep to be made in individual sections along each smooth piece in between the sharp corners and you can end up with several disjoint objects from the sweep in that kind of a case.

If you have sharp corners in your rail and all the rail segments are on a single shared plane, then it will do corner mitering to make it all one piece.

But if you can post the model file that would help me to see what you're running into - it's really pretty hard for me to guess just by a written description because there are quite a lot of different things that can cause different results to be generated...

- Michael

Hi Michael,

at the time I wrote message 4123.19 I had only the no save version but I downloaded the 30 days version. Since I had to start from scratch I tried a few different things. Instead of creating the faces 1 by 1 and joining them into a close volume, I used booleans ops, a 2 rail sweep and I used a fillet instead of the 1 rail sweep.

I haven't finish what I wanted to do but I uploaded the new file as requested. This new version doesn't present the same qualities as the jpg I shown previously. Basically, we can't compare the two, this one seems much better and MOI reports it's a solid. Also, I deleted most if not all the construction items I used but if you try to select edges around the fillet or the sweep you will see that only a single part of a edge can be selected at a time and this may add a bit of complexity if further operation are required, unless of course if there is way to rejoin those parts into a single element.

I understand that it must have been pretty difficult to figure out what I meant from just a few line of text and a jpeg. I apologize, from now on, I'll supply a file.

Out of curiosity again, I can foresee the need of say warping a rather simple shape like the (flat) medallion in my file to fit on the curve part of the post and I think it will most likely be most of the time on cylindrical shape like in this case, do you think it's possible to write a script to do just that? Before you even think of it, I'm thinking of writing it myself of course but I would just like to know if you think it's possible with the currently available MOI JavaScript classes, methods, etc?

Regards,
Felix

Hi Felix,

> but if you try to select edges around the fillet or the
> sweep you will see that only a single part of a edge
> can be selected at a time

I tried selecting some various edges in the model and I couldn't see anything unusual...

Maybe for this case I might need both the model and also a screenshot that shows the particular edges that you are worried about.

But it is normal to have edges that surround each surface - that's part of how NURBS surfaces work, each surface has a trimming boundary and that trimming boundary is made up of edges.

I can't really see anything out of the ordinary or that would cause any concern with the structure of your model there. But also I'm not sure exactly which pieces of the model you're referring to by the sweep and the fillet - since I didn't watch you actually construct the model I may need a bit more description yet to follow along with your question...

One thing that will generally happen is that if you have curves made up of various sub segments (where they can be separated into individual pieces using the Edit > Separate command), that will tend to generate one surface for each of those segments. That's totally normal for shapes that have a sharp corner in them, but if you want pieces that are smooth to one another to generate a single surface instead of multiple surfaces then you want to make your curve structure have single large segments in it instead of having the smooth pieces made up of multiple small segment fragments. One easy way to tweak curves to remove segmentation like that is to use the Rebuild command: http://moi3d.com/2.0/docs/moi_command_reference10.htm#rebuild but really a better way than that is to just draw larger segments for smooth pieces more from the beginning rather than drawing a smooth curve portion in little bits and pieces.


> do you think it's possible to write a script to do just that?

No, that's way out of the scope of what's possible to do just with scripting.

Warping of surfaces is actually a very involved process, to do it requires a major reconstruction of all the surfaces and edges in the solid. For example if your base surface that you want to warp is a plane, that surface only has 4 control points at its corners - a surface with 4 corner points cannot be reshaped just as-is to have a cylindrical shape, it has to have a lot of additional control points added into it. So it's a quite tricky process to do this to all the surfaces of a shape and also to all of its edge curves.

Rhino added some tools for this in the Rhino v4 version, so if you need to do this kind of warping right now you'll need to use Rhino for that part of your modeling process.

It is an area that I plan on working on adding to MoI in the future though.

- Michael

Michael,

sorry I couldn't be more useful with the file I provided, I know there is nothing wrong with it, at least that's what the Rhino analyze tools say.

But I still find this edge thing a bit funny, so I've made a simple test in the file I've attach. When I did a union on the 2 gray boxes on the left I get the red box where the edges at the intersection of the 2 gray box just disappeared as I would expect. I then copied the red box and made a strait line and a curve that I joined and used this curve to create the green and orange shape using a diff. They're normal I suppose but as you can see I would have expected the black and the red segment to select as one edge just as the black edge on the red box. I think I understand where the blue edge on the green shape comes from but I would have thought I would see no edge there.

I'm not complaining about this, it is just an observation of something I wasn't expecting. I think that a break like this can cause a retract in the g-code and on simple objects like these it doesn't matter but on more complex object with a lot of similar breaks, all these retract can increase machine time significantly. If there is a way to rejoin them, I'd like to know about it.

Thanks Michael and though it my not sound like it, I'm very impressed with MOI and it's quite possible that I follow your advice to the letter and get both MOI and Rhino in the end.

Regards,
Felix

Hi Felix,

> When I did a union on the 2 gray boxes on the left I
> get the red box where the edges at the intersection of
> the 2 gray box just disappeared as I would expect.

That's due to some special handling of plane surfaces - during a Boolean Union the boolean code looks for areas where there are 2 pieces of planes that would end up adjacent to each other and switches them with one larger plane. It can also combine surfaces together if each adjacent surface is a trimmed sub portion from the same original larger surface.

But in other cases it's normal for the very same edge structure in the original objects to come through into the boolean result.


> I then copied the red box and made a strait line and a
> curve that I joined and used this curve to create the green
> and orange shape using a diff.

So when you joined the straight line and the curve together, you have a multi-segment curve. When you use that curve as a cutting object, it will become extruded into a surface and that extruded surface is what will actually be used in the boolean.

When you extrude a joined curve, it is normal for the extrusion to have one surface in it for each segment in your original object.

So for example, extruding a rectangle made up of 4 lines joined together will result in 4 surfaces for the side walls of the extruded result, like this (I turned end caps off for this example):





Here's another example where the curve is made up of 8 joined segments like this:



Extruding this 8-segment curve will result in 8 surfaces, one for each of those segments, like this:




So in your case your curve was made up of 2 segments with the segments something like this:


So with 2 segments in your curve, it then follows that the extrusion will have 2 surfaces in it just matching the same structure as your curve, making a result like this:


So that's where that edge that you're mentioning comes from, it's the edge between the 2 extruded surfaces, and there are 2 surfaces because you had 2 segments in your curve...

Having the extruded object structure be the same as the original curve structure can be useful in many circumstances because it means you can work on the surfaces in the same way as the original curve, like for example with this one you can very easily remove some portions of the object that correspond to your original segments:






Does that help to explain where that edge is coming from?


If you don't want that edge it is possible to get rid of it by making a different kind of curve - instead of a curve made up of 2 sub segments you will instead want to make a curve that is made up of just a single segment. The extruded result of a single segment curve will be just a single surface instead of being in multiple pieces.

You can fuse the 2 segments in your curve together into a single segment by turning on control points and deleting the control point where they join at, or you can use the Rebuild command which will reconstruct a curve and it will make longer single segments out of areas in the curve where there are currently multiple segments that are smooth where they touch each other. More info on the Rebuild command here:
http://moi3d.com/2.0/docs/moi_command_reference10.htm#rebuild

So in your particular example it looks like the 2 segments were tangent at their join point, so you can use the Rebuild command on that curve to make it into a single segment curve instead of your previous 2-segment curve, and if you then use your new single segment curve as the cutting object in the boolean it will produce the result that you wanted where there is no edge at that one spot.


> I think I understand where the blue edge on the green
> shape comes from but I would have thought I would see
> no edge there.

But I'm not sure why you would expect this - doesn't it make sense for a curve made up of 2 segments to produce a surface made up of 2 faces when it is extruded? It just makes a result that is the same as your curve structure...


> I think that a break like this can cause a retract in the
> g-code and on simple objects like these it doesn't matter
> but on more complex object with a lot of similar breaks, all
> these retract can increase machine time significantly.

I can't remember ever hearing that having a model made up of several joined pieces has any negative effect on g-code production...

I mean it's really a fundamental thing that there will be many different edges in a model even where pieces are smooth with each other, if there was a problem with that it would basically be impossible to model anything with fillets in it - every CAD program that produces fillets will have edges around the fillets where they join together with the other pieces.

The only way to not have any edges in a model is to have the whole model made up of just one single surface and that's not very practical...

Could you maybe describe a bit more about why you're particularly worried about that?

But if you are really worried about it you can remove such edges by making your cutting curves be only single segment curves instead of multi-segment curves, and the Rebuild command can help with that.

But I think it's likely to be something that you don't really need to worry about so much though.

- Michael

Hi Felix, also having that edge there actually increases model accuracy because with that edge that you don't like in place it means that this area of the model here:



is able to be represented with a simple analytic plane surface rather than just a generic NURBS curved surface as it would have to be if that whole cutout piece was one larger surface.

With it segmented as it is currently, many CAD/CAM programs will be able to recognize that face is a plane and be able to do special case processing on it when doing various kinds of operations involving it like intersection calculations and projections.

When programs get to use special case processing for intersections it helps get both lighter and more precisely accurate results...

So your result there is good - there isn't any need to try and remove that edge - doing extra work to remove it is more likely to degrade your model a little rather than improve it.

Now, one reason why it's good to have it in your particular case here is that it maps nicely to a large planar region of your model - having a large planar region be constructed as a plane that is bounded by edges is good.

But that doesn't mean that having more edges is always good - there are other cases where it can be good to work on removing edges - like if 2 coplanar pieces are going to end up adjacent to one another, it's better in that case to have one large plane instead of 2 adjacent fragments - that's why boolean union merges those kinds of coplanar fragments together.

It's also not so good to have things split up into a lot of little tiny slivery fragments - say for example your cutting piece was a more arbitrary wiggly thing like a sine wave rather than having one piece that was a line, then for something like that it would be likely be better for it to be one long surface without any internal edges in it.

So maybe the part that's a bit confusing is that there isn't one single answer as to what is best - if you have plane that's next to a curved piece like in your last example, it's good for that case to have those as separate faces so the plane can be a simple analytic piece. But then in other cases it can be good to reduce internal edges though too.

It kind of depends on the particular situation, and it can even depend on what you specifically plan to do with the model....

In your particular case that you posted, I can't see any reason for concern about it though, having that segmented with the plane as its own separate face is most likely the best structure for that particular object.

Hope this helps explain things in some more detail! :)

- Michael

Michael,

> Does that help to explain where that edge is coming from?

Of course I do understand and I know why I expected a different behavior. I've been using Skecthup for a while and when I joined (actually weld) segments together they become a continuous curve as in MOI but this even after extrusion (push/pull in SU) or many other operation.

>But I'm not sure why you would expect this - doesn't it make sense for a curve made up of 2 segments to produce a >surface made up of 2 faces when it is extruded? It just makes a result that is the same as your curve structure...

I figure that since MOI uses the structure of the 2 segments curve I've drawn it makes perfect sense that this blue edge is where it is, that I understand. I also understand that if I want a continuous single curve even after subsequent operation, I just need to build it continuous to begin with, I've tried this and it works perfectly.

>Could you maybe describe a bit more about why you're particularly worried about that?

I intended to build a 5 axis CNC router at some point and use the CNCToolkit to generate g-code, a MAX script that works both in 3DS MAX and GMAX. This script though very useful because it's free and can produce g-code for synchronous toolpaths for 3 and various configuration of 4 or 5 axis CNC. It's not a CAM program per say and one of the notes mention to watch for breaks or segmented elements because the script interpret a break as retract operation. But this maybe irrelevant with a true CAM program such BobCad that I intend to use.

As I wrote this above I was interrupted and my wife and I took the time to discuss about all this idea of CNC and 3D modeling highly detailed objects and she convinced me to do something much less demanding for old folks like us and her major point his that we do something that we can do together more like a hobby then a business.

But I like MOI enough that I'll probably buy it anyway instead of Sketchup and use it more with the idea of having fun with it then of working with it.

I want to apologize for the time you spend with my silly questions. I want you to know I've appreciated a lot conversing with you.

Thanks,
Felix

Hey Felix,
Home 5 axis CNC would be pretty rough on the ole relationship!!! :o More like a hair pulling event than enjoyable fun... lol

Though, with MoI and a couple other programs, you could do "ART" together, for fun and hobby...

Take care.

Hi Felix,

> I want to apologize for the time you spend with my silly
> questions. I want you to know I've appreciated a lot
> conversing with you.

Don't worry, it was no problem! :)


> and I know why I expected a different behavior. I've been
> using Skecthup for a while and when I joined (actually weld)
> segments together they become a continuous curve as in
> MOI but this even after extrusion (push/pull in SU) or many
> other operation.

Yes, the way SketchUp works is pretty different - SketchUp doesn't actually have true smooth curves in it at all, what SketchUp calls a curve is actually made up of a bunch of little line segments. A "weld" type operation in a polygon modeler is a bit different than MoI's Join function. MoI's Join is somewhat more like making a group, but kind of a special group where the pieces know that they are connected together either end-to-end for a curve or edge-to-edge for surfaces.


> It's not a CAM program per say and one of the notes mention
> to watch for breaks or segmented elements because the script
> interpret a break as retract operation.

Well, 3DS Max and gMax are polygon based modelers, so this was possibly referring to needing to have polygons welded together so that the triangles that make up a surface share a common vertex point which then can make it easier for program code to find adjacent polygons to traverse them.

The equivalent kind of "traversability" in a NURBS based program is having 2 surfaces be joined together so that they have a shared common edge rather than being 2 completely separate surfaces that just happened to be nearby one another.

So that edge that you were worried about previously was not a problem in this sense because it was a joined edge between 2 surfaces.


The geometry in 3DS Max and Gmax is made up of triangle meshes, so it's actually chock full of edges everywhere, it just won't display them all the time unless you are in wireframe mode. So I doubt that the CNC Toolkit thing is actually sensitive to just any edge or else it wouldn't be able to handle more than just 1 triangle at a time, that warning was most likely about an edge where it could not find another adjacent polygon that was hooked into the same vertices as the current one.


> But this maybe irrelevant with a true CAM program such
> BobCad that I intend to use.

You could actually find a similar issue in there, but just to reiterate - the issue is probably going to be if you have just totally separate unjoined surfaces as compared to surfaces that are joined together and share a common edge (making a connected skin).

- Michael

Hi Mr. Burr,

thanks for your time and yes sometimes, maybe more then I would like, I have crazy ideas. This 5 axis CNC idea is probably one of them especially that I would need to spend countless hours to learn a lot of stuff on own. My wife suggested this simple idea of doing something simpler we could do as a team just as we do already for quite a few things. We are close to retirement time and I think this idea of doing even more stuff together is just great.

But I also like to keep my brain busy sort of speak so I'll probably buy MOI even if only to do just that. I might also build myself a small 3 or 4 axis CNC for the same reason.

Regards,
Felix

Michael,

as I said in the beginning, I'm a newbie at all this NURBS stuff and same goes for 3D design. I'm old enough that when I started drawing stuff it was on paper, my first computer was an Apple II and at that time I was in the minority, of course today it's the other way around.

As I said earlier, I'll probably buy MOI even if it's just for the fun of learning new stuff. Unfortunately for you guys here, I'll probably have few more silly questions coming by this forum. I just hope they'll few and far apart.

Thanks again,
Felix

Hi,

today I tried something easier, just practice stuff and fun, nothing to fancy. Then I decided to render the mirrors just to see, so I exported them to obj which loads directly in Kerkythea and tried a few thing but without success I'm afraid. I assume it as to do with UV mapping or most likely me, I'm afraid I'm newbie at that as well. I though of asking you guys for suggestions on rendering apps that are as easy to use as MOI and as cheap as possible that would make a good companion for MOI and whatever you can think of that I could read to learn as much as possible on rendering?

Thanks in advance,
Felix

Hi Felix, some links to discussions on renderers here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=4098.3

There are a few different suggestions in those threads that you could try.

Often times renderers are not particularly focused on being easy to use as a big priority though.

- Michael

< Often times renderers are not particularly focused on being easy to use as a big priority though.
Except some like Podium for Sketchup ;) http://www.suplugins.com/

Michael, Pilou,

thanks to both of you. Lots and lots of info to digest but I'm not sure I'll live long enough to just scratch the surface of the rendering world. It's like the United Nation where there are a lot of different languages and cultures but no indications where to find the translators.

I came across this video http://www.vimeo.com/4810249 for example. It's not bad per say, some may even say it's excellent but it's almost useless for total newbies. It's what I would call a recipe type document, do this, do that and by the way you can do this and that as well. You see I was a teacher in another life and I know that people don't learn much from document like this because it looks so very easy for someone who is already an expert using these 3 apps together. But for a newbie, after downloading and installing Blender and Lux he or she basically hits a cement wall 4 foot thick. As you well know Michael, I don't think anyone could say that one can learn to just navigate with some ease through Blender UI in a few minutes, maybe not even in a couple of hours. More or less, the same could be said for Lux, MOI and any other apps for that matter.

It's true that one of if not the most efficient method of learning is by <doing> stuff yourself especially when you have a project that gives you plenty of motivation. But that's forgetting the main ingredient, basic knowledge of the context or some basic theoretical knowledge as some might say. Say your goal in life is to become a fisherman but you're a total newbie at this and you come across a video tut that basically tells you go get a boat and a net, throw the net at sea, drag it for a while and bring it in, all this while the video shows the net coming in filled with fish. Sound and looks easy but unless you are a very very lucky guy I would bet a week pay that you'll have a nice day at sea but few if any fish to show for your work. One would think it's easy to figure out what's missing in my example but it's not, especially for an expert kind of guy.

The basic idea behind all this is that when someone knows how to do something it always looks easy to a newbie and when one tries it on his own it's not long before one realize the learning curve is more or less steep depending on one contextual prior knowledge of the task.

Though some may find this an interesting or possibly a boring topic. It made me realize that I need to be a bit more specific about what I'm looking for, especially that at this time, I'm a newbie at 3D design and rendering. Basically, I'm looking for reference material (books and or papers) that are use currently as teaching material in university level programs on 3D design and or whatever programs where rendering photo realistic scene is a prerequisite or that is part of the main program. I'm sure such material exist and it's plentiful and that's the problem and the reason I ask you all, knowing that some of you still have those books from college or universities as we call them here. I can read in both English and French just in case someone think of a french book.

Thanks,
Felix

Hi Felix ...

The best way to get started in MoI is to do the tutorials ... as many as you can do.

As you said, the way to learn is by doing. Some things can be done by just jumping in ... others need some instruction.

My advice would be to do the MoI tutorials first and then try modeling some simple things on your own. You don't have to learn all a program has to offer ... just the things you need to complete the project you are currently working on ... make the next project a little more complex and learn a couple of more things the program will do.

I say this as an old timer (much older than you) who has become fairly proficient with many programs over the years ... but not a true expert in any. I have managed this a step at a time without ever trying to get a university equivalent in any of them.

I make a living as an architectural designer. MoI or Rhino or any of the others will provide the tools for a wide, wide range of uses ... more than I will ever need. So learn the basics you need and grow from there.

Start with MoI and an inexpensive renderer (Like Sim Labs Composer) and enjoy the process. You could spend a fortune on great software and get overwhelmed with complexity.

Enjoy,
eric