Changing an low poly model into something that can be cnc milled from plywood. i think I've built it all wrong!

Hi all

I am in the process of learning rhino for the purpose of creating some low poly plywood sculptures and have a few questions for more experienced modellers.

I want to mill the the faces from 16mm - 19mm ply and mitre join them at the edges. I have created the model from surfaces got it to a point that i was happy with and now i am going back and trying to ad volume/extrude the surfaces to create the thickness of the plywood and the angles for the mitre joins.

  1. have i gone about modelling this all wrong?? what is the best way to create a model like this?? should i start from scratch??

I am using the small objects mm’s template but the measurement tools are giving me a measurement of 83.35 with i took as 83.35mm. which is much to small i want a finished height of around 1200mm when i try to scale it up to what i think is this size the object is to large for the gridded artwork area.

2. Is the measurements infact in cm not mm and where can i find out what units I’m actually working in??



LowPolySculpture.3dm (133.7 KB)

First, looking at your model, you’ll never be able to make that as it stands in plywood, as some of your faces are not flat. Remember, only a triangle defines a plane, if you have surfaces that have 4 linear edges, you cannot assume they are planar. So you will at least need to split the quad surfaces into triangles and have an additional miter edge at that spot. Probably just delete them and replace them with two triangular planar surfaces.

Second, the question is do the surfaces represent the inner or outer surfaces of the object? You can try OffsetSrf the joined polysurface to one side or the other with the thickness you desire - do not use the solid option, it will not help much in this case - that will give inner and outer “skin” surfaces. Put each on a separate layer and then explode them into individual surfaces.

Edit: actually trying this, I see that OffsetSrf does not give a reliable result in this case… so, we need to backtrack.
Fix your surfaces as in the first paragraph, then join into one polysurface. Then use the command Mesh with the option “Simple planes” to create a mesh object identical to your polysurface. Then use OffsetMesh (not OffsetSrf) to offset the mesh to the inside or the outside. Then take the resulting offset mesh and run MeshToNurb on it to turn it back into a polysurface. Then you can delete the two meshes.

Then, I think the easiest way to create the miters will be to use DupBorder on each of the “matching” inner/outer surfaces and Loft the two borders together to form the edges. That should give you the correct miter angles (hopefully). However, given the arrangement of the faces, the miters between two panels could very well be warped surfaces (requiring 5 axis milling).

As far as units/scale go, the file units can be found in Settings>Units, the file is currently in mm. Your scale is arbitrary in this case, if you want it to be about 1200mm high, then you are going to need to scale the structure around 14x. Do this FIRST before you do anything else - always make sure you are working at 1:1 scale in the correct dimensions/units.

HTH, --Mitch

Thanks very much for taking the time to review Mitch, I appreciate it. Will definately try what you have suggested tomorrow just to see what happens and understand more about what’s actually happening with the model. I think I may also start from scratch again to give me more practice modelling, and to get set up working 1 is 1

Re: surfaces I thought this might be the case with them not being planar but kinda eyeballed them and like you said assumed. Re: Inner or outer surface not as important as it just a proof of concept project at the moment so have no limitations I am workIng with in terms of size. but will likely need to model and build similar things in the future if successful so i suppose whichever way is easiest.

The shape is somewhat unimportant, I am just aiming for an abstract low poly asymmetrical shape that is fairly easy to mill and more complex and interesting than a pyramid​:joy:.(pretty sure my friends cnc in 5 axis?? From memory). I don’t want to buy or download a model or pay someone do it for me I Like to learn things for myself, but am a bit out of my depth​:grimacing::grimacing: and none of the online material and tutorials seemed to cover my situation.

My workflow was, I modelled a surface I liked and got to a result that I liked then started thinking about how to manufacture wich i am learning is clearly not the best way to work​:joy::joy:

What would be the best way to approach this project if you were to start from scratch?

Thanks again for taking the time to reply and review my file.


Well, in my mind there would be two approaches - each with its own set of advantages/disadvantages.

  1. model the surface as a free-form NURBS surface, then try to “panelize” it into flat panels for manufacturing.
  2. model the surface as a defined series of flat panels from the outset

#1 above gives you a lot of freedom to model the surface the way you like, but the panelizing process may be somewhat difficult. There are some programming/modeling tools available to help you like paneling tools or even (what I would suggest) Grasshopper, but that takes it to another level above basic Rhino modeling.

#2 may be limiting in that you are probably going to define the number of panels and their arrangement beforehand, then do some editing to refine the shape. But it is probably easier to accomplish to start out with.

Rhino provides a few interesting tools to do #2 - what I might do is draw a series of connected planar triangular surfaces in Top view that define the plan view of the structure, and join them all together into one. Then, with the command SolidPtOn, turn on the solid points and you will be able to pull the vertices around in both plan and vertically to push the panelized shape into 3D. You can always subdivide faces and create new ones if necessary as you go along. When done, you would need to follow the original procedure to create thickness and miters. You could also do essentially the same thing using a low poly triangular mesh object instead of connected planar NURBS faces.


you have the loose concept already… now might be the time to start thinking of joinery and assembly… how you’re going to clamp it, splines?, metal tabs? nodes? etcetc.

you say your friend has a 5axis machine which theoretically, should be able to carve any random miter you throw at it but doing it that way (carving) may be entirely inefficient / impractical…

have a look at various chamfer/miter bits… you may find you’re better off if limiting joinery to available bits… say, 7º, 10º, 15º, 22.5º, 30º, and 45º…

once you go though the actual construction process and come up with a system that’s going to work for you, you can then go on to refine the design within the limits of what’s able to be built or- how you’re going to build this specific object.

to skip this phase in the process is going to lead to disaster down the line… or, down the line, you’ll be forced to come back to this phase and do everything over again :wink:

so yeah, try to think more about joinery/assembly for a while… or- design any random looking thing then have the 5axis machine carve it in its entirety without any need for pesky joinery… or outsource it to a 3D printer. (but in these two hypotheticals, no need to worry about planar plywood)

Thanks Jeff,

Yeah will definitely head into the shop this weekend and have a chat about what is achievable, as i hadn’t even thought about the joinery (slap some glue on the mitre put a block on the back and shoot some brads into it!!:fearful:). the panels are going to be printed onto with a flatbed printer thus the need for plywood and because flatbed printing to timber looks good! your right maybe it could just be cut not cnc’d. i think the more time i spend working on the model is helping me understand rhino and modelling in generals i think refining the model to make it as simple an easy to manufacture is probably time well spent.

thanks again for your input, there is a great active community here, i am a print based designer by trade and reminds me of the early design forums, you get nothing like this kind of input in any of the adobe forums anymore.



Thanks Mitch

Started from scratch again used the method you suggested for defining the shape, perfect match more efficient than my bezier drawing technique :joy:

missed the bit about not using offsetsrf :anguished: so used it went through and lofted all the shapes. i did notice they weren’t lining up but thought maybe i could extent the planet meet fill the gap and because it is all good practice making me more familiar in rhino. Have attached the result.

Will have another crack at it tomorrow night using the mesh technique. Though i could possibly use offsetsrf and do it to the inside then trim with the overlaps somehow. Anyway i think i will try both tomorrow night and let you know how i progress, the practice is making me feel less like I’m bumbling my way around a maze:grin:

Like i said to jeff, thanks for your input, there is a great active community here, i am a print based designer by trade and the community reminds me of the early design forums, you get nothing like this kind of input in any of the adobe forums anymore. i wonder if the big boys like solidworks and autodesk have such active communities?



@jeff_hammond I was hoping you were going to jump in here… thanks.

1 Like

hi Guys

So i went back to the start and re modelled it. Good practice. Have come across a few problems maybe you can assist with.

  1. when offset mesh then mesh to nurb is used it changes 1 surface to 2 surfaces in this process. so i though no worries still join into polysurface then loft the 2 edges. However this does not work and gives weird results… not a closed surface. (green layer)

  2. it looks to me like the mitre joins are not the same angle all the way along. Is there a way to control this or adjust after the loft process.

  3. Am i using the right tool for the job??? am i just trying to do something maybe rhino is not best suited for? or is there a plugin that will take a surface and add thickness and deal with the joins automatically??

Thanks again for all your feedback.


TimberSculpture2V3.3dm (329.4 KB)

Why are you using meshes instead of using only surfaces?

When you say “loft” do you mean the “Loft” command or something else?

How are you creating the offset surfaces? Mitre angles should be constant along each edge.

Hi David.

Offset srf command. Doesn’t work gives funky result. Mitch suggested change to mesh’s offset mesh then mesh to nurb.

Yeah the loft command on the green surface (one edge is a polysurface) gives a wierd not closed result.

Is there a better way to go about this operation??



rhino is the right tool for this.

you can’t just offset the initial polysurface then expect all the intersections to arrive at a result of easily machinable miters… because what you’re expecting to see happen isn’t possible with the input geometry… you’ll have to figure out a system (maybe not the right word?) to make this possible…

once the logic(?) has been figured out, you then work outwards from there… adjoining planes will be positioned more strictly based on maths as opposed to just looking cool.

but without going down that road which would be required to get machinable / well fitting miters at all intersections, there are maybe some other methods you could try… with the tradeoff being more finicky assemble , or, require neat thinking on some sort of brackets etc.

using your model, two approaches might be:

something more node based… all of the panels are cut square (no miters)… gaps are left in between panels but this eliminates most or all of the complications involved with intersecting all of these surfaces… but increases complications with what to do at the corners of the triangles.

a sort of similar approach but without the gaps would be to overcut all of the miters… cut everything at 45º then the long points will meet tightly without the worry of collision from adjacent panels.

(view from underneath)

with the right brackets, the miters could probably be left as is… or you could fill them with a polyester resin etc…

from the top, (with a little luck :wink: ), the outer surface will look just like the model’s outer skin.

either of these two would be child’s play for a cnc machine to cut.

there’s another approach in which half of the necessary miters could be cut at a standard tool angle… say 20º… the other half of the miters would need to be cut according to the angles between the individual panels… numbers like 27º and 33.4º (ie- numbers which don’t have corresponding cnc bits)… it’s doable but you’d be better off with maybe a router and horizontal panel saw… track saw could do it too with enough patience… doing it like this would be much more work than having a cnc router cut the panels… the backsides of the miters wouldn’t match up perfectly (because one half will have longer miters than the other) but the entirety of the joint would match perfectly resulting in a strong glue joint…

this is the way you’d have to make the cuts with your current approach/shape and desire to be left with well seamed miter joints for glue & brads… it’s a lot more shop work (annoying shop work-- plotting points, a whole bunch of blade angle changes/tunings/etc) but not impossible by any means.

I’m not sure if this would work without trying it, assuming the parts are planar and the ply is 19mm, ie easy to handle on a saw, perhaps a method using cnc to get the accuracy and table saw for the random mitres might work.

Taking the outer surface of the model and joining into a polysurface;
Offset inwards with solid=No
Explode the resulting inner offset polysurface;
Offset each panel back outward with solid=yes;
'DupBorder each poly where it meets the outer skin;
Delete these solids used to create the DupBorders;
Unroll the poly Explode=No, along with these polylines;
Mirror (if necessary) the unrolled surface so you are looking at the underside of the finished part;
Explode the polysurface and group the labels and polylines ready for nesting;
Nest the parts for cnc and put the polylines on a separate layer to become scribe lines on the cnc;
Run each part through the saw using the scribes to create the angles for the random mitres;
Don’t forget to number each edge…

Thanks Jeff.

I had thought of the 1st option but not the second, i wasn’t sure about the joints in the first option and thought they might interrupt the image to much, but thinking about it i could play up the angles and work around the joints. I am printing some artwork to the skin using the same method as the attached examples, so i need to dry fit everything then send to printer print then reassemble when it comes reassemble. So this might be the best option.

i think i may try and model a version limiting myself to maybe a few mitre angles, and more simple again to start with, as you said then it could possibly be cut on a table saw, just as more modelling practice any advice for doing this?

i didn’t even think about there being different mitre lengths :rolling_eyes: as you can probably guess I’m definitely not a carpenter or an experienced 3d modeller.

the other idea i had was use much thinner ply and design a frame for it to fix to as it would not be self supporting. I thought this was probably quite a bit ahead of my modelling ability and would probably be pushing the friendship with my chippy (carpenter) mate.

Thanks again for the input and feedback.



Thanks for the input brian!!
very clever idea i went through and followed your steps and didn’t understand what you intended until the very end when it clicked and i was like ahhhaaaa i get it now.

it seems like it would definitely work however I’m hoping to limit the shop work as I’m imposing on a friend for this as i definitely don’t have the hand skills needed. :flushed: So i want to keep the machining and construction as simple as possible, definitely going to file that technique away though.

I think I’m going to remodel much simpler and limit myself to some simple mitre angles so hopefully it can be made on a table saw or investigate the option of leaven the joint edges exposed.

Its all good modelling practice and i think modelling within manufacturing limitations is probably good practice? i guess any muppet can model an interesting shape on screen :joy:

Any advice on achieving a model with limited mitre angles and complexity greatly recieved!

Thanks again



Don’t compromise your design around standard bits if it’s not what you had in mind for the design, at least until you’ve exhausted other ideas.

To remove the table saw cuts maybe something along these lines would work. The scribe lines, mentioned in the other post, can become secondary cut lines for a cnc rebate, to a depth which leaves say 1mm of material at the base of the rebate. You would then have the accurate mitres where the panels touch and resin fill the cavities and maybe a bit of fibreglass tape etc…

You would nest the panels with enough clearance to do the rebate cuts first, then the final cuts.

for clarity, using standard angles offer a lot more options than might be evident upon first glance… a 20º angle doesn’t have to butt up to another 20… it could be 20º and a 15… or 20 &22.5… or 30 &22.5… or etcetc…
you should be able to cover all the angle differences that are discernible to eye… it’s just that they’ll be precise/machineable instead of a bunch of random in-between types of numbers that, design/aesthetic wise, you won’t be able to really notice anyway.

fwiw, the model in this thread can be modeled with accurate miters without any modifications… it’s a lot of manual work and, for me at least, i find this type of modeling boring… it would be more interesting if i were modeling it based off a procedure as opposed to shaping it then slicing it up in a somewhat messy manner.

TimberSculpture2V3.3dm (134.1 KB)

that’s just a few of the panels for example but i didn’t slice up the intersection where the purple/blue/red/orange surfaces come together… i get a headache just looking at that :wink: … just the long intersections were done in the file.

Brian, your idea shown in the cross section could , with the right material, possibly work without making a cut where it says ‘cutline’ … just flatten the design as you did earlier then cut square grooves (your ‘rebate’ area) on the mill but don’t go all the way through.

then just fold them up. similar to doing miter fold boxes on cnc machines except you’d used an end mill bit as opposed to a V or chamfer bit.

Hmm, I guess you would need a v-bit to get the crisp crease or at least come in after with a vee bit to score at the ‘cutline’. On second thoughts, the shallow angle joins wouldn’t have much room to use this method.
Wonder much material thickness is left on the MDF in the video - trial and error to get a good fold without tearing.

Mixing up the angle bits has promise, just needs a bit more time working out what bits are required.

promising in theory :wink:
getting all the bits isn’t necessarily going to be cheap and it’s doubtful the owner has all the various angles already… and even less likely the machine in question is set up to use them all without manual changes.

i think the fastest way to do this without modifying the design would be to figure out all the angles in a similar manner as that .3dm i uploaded a couple of posts ago… have the cnc cut out all the triangles (just square 2D cuts) since this would be the hardest part about doing it on a table saw…
then take those triangles to a tablesaw to cut the miters… using a setup where the angled blade dies into the fence so you’re only cutting the angle off… i’ll see if i can find a picture showing this setup because i don’t think i’m explaining it too well. :wink:

weird… can’t seem to find a picture online of what i’m wanting to describe… i’ll take a picture of the setup tomorrow if i remember… it’s a sort of risky process but the risk can be minimized by using a dado stack.

like this:

this way, you wouldn’t be removing any of the triangle’s dimensions or modifying the long points in any way… just removing material beneath the edges…

it’s sketchy because that little piece trapped in between the blade and fence can and will shoot back at you once the through cut has been made… to eliminate that from happening, use a dado blade so all of the miter is cleared away by the fat kerf leaving nothing to shoot back