Thanks for watching!

All good!

Hi there. Good question. The sweeps are tools used to trim the surfaces back an equal amount from each edge, before building a blend in the holes left after trimming. You can try and add a fillet, but as the tangent angle difference in between the surfaces that are being filleted/blended runs down to zero where the creases run out, it means the fillet will also run down to nothing as well. Basically the fillet chord varies based on the angle between the surfaces. The result using swept pipes results in a visually better result.

All good. I’m pretty sure a squircle has non linear sides. If you get a straight edge and check an iPhone, Apple TV etc, the sides are linear.

@@AndrewJacksonDesignStudio you’re right, it’s not a mathematically perfect fit. But I think the idea is you’d use each quadrant for each corner, offset them, and connect with lines. That technically creates a microscopic curvature discontinuity where you join them, but it’s so tiny that it doesn’t feel like cheating to ignore it. If you can stomach that, the squircle (“super ellipse”) has otherwise smooth curvature change: it’s not piecewise and has no linear sections. It’s also simple to calculate.

Out of interest, do you think the mathematical “truth” is important? Or are you just looking for a practical “continuous curvature” radius shape?

Any curve defined in a cad system is a mathematical truth. Therefore it’s more about what the designers intent was. All radii are continuous in curvature as they are arcs. Im interested in investigating this as are others as there are different ideas around the ‘mystery’ of how apple may define their corners. The placement of the arc and blends in the grasshopper definition gives a result that is quite close to my reference. I suspect a squircle segment will be a bit too pointy, like a conic with a higher rho value

@@aaaidan nah nah through the squircle away. It is too pinched. It is more organic and doesn’t gel well with straight lines

You can get a g3 blend in between a line and arc, with the arc being tangent to the sides. That is one way how I make a lozenge form. I have two angles coincindent to the arc centre, one pushes the line start point back and the other controls the arc size. This can be done with none of the blend CVs extending outside of the line, so no hump. But when used on a 90deg corner, it definitely looks forced compared to having the arc off set or set back.

@@AndrewJacksonDesignStudio Oh for sure it’s doable. I’ve made some pretty ugly G3 splines trying to get smooth curves to fit the wrong constraints. I think the reason it seems forced in this case that we expect the curvature to peak (or in this case plateau) once in the middle of the corner and run downhill both ways from there with no local maxima. It’s possible to blend a radius into a tangent line with a g3 curve, but I don’t think it’s possible to do so with the maximum curvature being at the endpoint.

Sounds good Tony. Looking forward to seeing what you think. I'm going to check the effect of the three piece corner on the pillowed surface/pinched corner. Might have to decouple the corner and pinch surfaces.

@@AndrewJacksonDesignStudio so here are some thoughts: 1) How about other angles? We have covered 90 but need to cover other common angles like 60, 45. 120 and also apply this for the perfect “pill profile”. 2) we can’t be too sure if this is indeed what Apple is doing. Visually it is a very close match. I had some success replicating this rounded corner approximately with a single spline. This would be similar to achieving a g3 continuity between the clothoid segment and the circle. All I remember is I did not use style spline in solidworks to achieve this. Instead used regular spline and manipulated the control polygons. 3) 3 pieces seem very complex. Looking forwarding to see the challenges in making surfaces out of them. Imagine having to construct this for multiple guide curves and what not. Tedious. 4) a protocol for offsetting: I think you should try out various ways to offset this 3 piece corner (both inward and outward). What way of offsetting looks visually pleasing and what way of offsetting strikes a balance between looks and speed.

Definitely a black hole for time!

@@AndrewJacksonDesignStudio Someone has to do it 😀

Cheers!

Thanks for watching. This is all guess work :) but educated guesses!

All good

I've tried several ways to achieve the corner, like you have. One piece, three piece (similar to this vid, except assuming the arc was tangent to the sides) etc. It's all good learning though, that we can apply to other work

@@CraigMcG oh the time I have spent on this. Pre 2020 not a lot of this was discussed in product design forums

Not ott in my opinion. G3 fillets is a staple of apple design.

Not sure about G3 being OTT, the roll off the planar faces on the top and bottom is super smooth and consistent. I think the bottom panel is a pressing judging by the thinness around the rear where you can see an edge. I've not taken it apart yet :)...

Yes, that's my understanding as well. There is an added benefit of having higher order primary geometry which is it trickles down to the secondary surfaces. Like the main outside corner and the effect this has on pinching on the top and bottom case corners.

Its a style spline. It is in the drop down menu below the old spline (which is default in the menu). Select style spline, make sure it is set to 'bezier' in the side panel. You can make a spline with three points to begin with. Roughly arrange the middle CV to approximate the curve you want. Then select the style spline and increase the degree to 7 in the side panel. This will give you 8 CVs. Then make the relevant control polygon segments co-linear to the two linear elements. The file is in the description as well. Cheers, Andrew.

kzfaq.info/get/bejne/gLyhnsqX2cesfp8.htmlsi=98EtTDOj9CJkVjlG

Hi Tony, yeah that wobble... this is like the back of the old Imac with the full pillow with no planar surf. If you struggle to make it in CAD, look closer at the actual product and the solution is right there in front of you. Neither of us have a way to confirm how the corners are defined, it is just guess work. No way to measure at these scales and trace to reverse engineer. The only tinsy bit of information I have to go on is in one of my interviews with the fruit company, I was shown three nurbs corners in Alias and had to identify which was single span etc. From memory, one was SS G2, one was SS G3 and one was multispan G3. Probably means nothing! I bet they use an arc mid section blended into the linear elements :)

@@AndrewJacksonDesignStudio hah! It’s just that I can somehow easily spot the non linear curvature in a g3 curve that’s used to transition between two 90 lines. The curve sort of has a pinched look if you know what i mean. Rather than a “circular” look

@@AndrewJacksonDesignStudio one way to reverse engineer would be to take one of their marketing pdfs. Or even their official dimensional drawings in the Apple accessory design guidelines. Open the vector drawing in illustrator and isolate the path that forms the rounded edges. Export as a dxf or something that would open in solidworks and then try to fit a curve on it

That is assuming the physical corner has linear curvature. I'm sure we've had this conversation before!

Both the luthiers that I've been talking with are machining their necks and were struggling with this area. Mainly with controlling the fullness or concavity of the blend into the volute edge. Fullness is ok, sand it back. Both had concavity, so best to fix in the model as auto 'surface fill'er looks bad on an oiled neck.

No sorry.

Hi Eduardo, finally got around to looking at this!

Thanks!

Yeah, that's why I ended up kneeling to get down into the trench. Still, back feels better than a day in front of the computer!

All good! Hope it works out for your design.

Maybe, however I think it'd need substantial changes to the definition to work on a surface of revolution or compound curvature. At the moment it is set up to work on an extruded surface

Hi Matt, I mainly use Step 214 to export to Rhino. There are some options in the SW export settings which you may want to change, depending on whether you want SW to export curves as well. Hope that helps and thanks for watching! Cheers, Andrew

Hi there. I re-positioned and rotated the scan data to where I wanted it in Rhino. Much easier than in SW. Off the top of my head, what you mention should work though.

All good!

Correct, it is watertight surface body that can be solidified.

All good!

That is strange. I use this technique frequently to cap off corners and other areas where the primary surfaces leave a 2/3 sided hole. Here's a few videos where I have used this technique, except with a G0 boundary (same as the G1 centre line boundary the RH plane you mention in the nose cone). I guess the difference is in these videos, I do not use a ribbon to explicity define a draft angle on the G0 boundary. kzfaq.info/get/bejne/o6liqMuBvrnHiWw.htmlsi=hbTqTQocL5-ASEKW kzfaq.info/get/bejne/r9Wep7KAnLCXeaM.htmlsi=NLhpusdbq2Lr5zBq Ruled surface can be used to make a ribbon, but sometimes the result can be a little average. I've had some success using an extruded 3D sketch (with draft) to make a ribbon surface, or if you don't mind a little variation in the draft, a loft, using two profile curves made in a 3D sketch, tangent to the relevant geometry, then a single edge/curve as the guide. If you are using a boundary surface to make the final patch surface, try playing around with the tangent influence a bit, as some times having 100% in the first direction and 0% in the second direction works, or vice versa where that creates ripples near the boundary.

Thanks!

Terminology or accent? If accent, welcome to robotic New Zealand.

Very much agree, that an interesting topic of SW surface modeling. When to switch to solid ? mid part, copy surfaces to make temp solid and check volume for example or are there other ways to handle surface/solid rebuild stability issues ? Andrew, your thoughts would be highly welcome !

I only use 'insert part' for copying geometry into other parts. You can copy surfaces, solids, ref geometry etc. I guess with AAX/published geom it gets added into the master part file as a feature, so you dictate at which point geometry is captured and published? The problem with insert part is it just copies over whatever is left at the bottom of the master part feature tree, so if a surface body is referenced, then someone adds a fillet to the surface body, that fillet geometry gets sucked into the child parts, whether it is needed or not. Ways around this is to make a copy of the geometry that is to be referenced, further up the tree, then make sure it does not get cleaned up with a delete/keep body feature at the end of the model. This might sound weird, but I have worked on a master where it effectively needed a branch/fork with some deviations in the geometry that was being referenced. Be way easier if you can pick the geometry to be referenced at discrete points in the tree!

@@AndrewJacksonDesignStudio what about Save Bodies? then it saves only the part that you want. Ive done this with different configurations too. I also like Insert part when I want to import more info.

The downside to save bodies is it does not take any reference data with it which is why I use insert part. I guess you could create planar surfaces to act as planes, edges as axes, etc, but then you'd have to be on top of documentation in a multi user environment so others understood what was what.

All good! I needed a push to get back on the bike, so to speak :)

Good to hear! I've used this technique on a rowing hull, worked well.

Hi there, thanks for the suggestion. I can definitely see the value in a video that covers some of that. Might be some time away as my interests currently lie elsewhere.

Just something I've been dabbling with. Need to figure out how to make the camera move faster on straighter sections, then slow in corners.