Offset has drastically altered cp count

Hi
V5
BlendCrv skyg method for each profile,
A good curvature graph results, and keeping point count to 8 with deg5, This should surface well.

I then offset by 0.5mm result to allow for fabric covering of item, result is a mixed bag giving rippled surface with Loft or Sweep2 or NetworkSrf.

I look at details and they are now deg3 and point counts are 30 32 34 28 31 27

As the whole approach was to avoid what rebuild does, (not quite match curve) and get a good result with deg5, why was there no option to keep degree and point count in offset ?

Now what is best way to get these back to deg5 and point count 8 ?

I need to keep the exact shapes now. surely same shape just offset a bit shouldnt require a massive difference in control points ?

Steve

You cannot, in general, offset accurately without adding complexity. The solution is to rebuild a copy the offset and edit the result to closely match the offset.

-Pascal

Hi,
blast !
means doing all that lovely curve work again.

I will try for changeDegree back to 5, then -_Rebuild with its select a master curve option and select the original 5/8 and get a distribution of matching CP’s.

would that do it ?

what a pain, cant believe 8 cp became 30. when its the same shape, to my mind anyway.

Could offset have added as options a degree choice and a -_Rebuild master method , called keep CP count maybe ?

Steve

I already did request this in another thread: Link

But I think this is a good occasion to repeat this request.
Please make the “Loose” option from “OffsetSrf” available in “Offset”!

To Steve:
Here’s a workaround for Your problem: Extrude the original Curve (and Convert the extrusion to a surface) Then use “OffsetSrf” with “Loose” option to create a Surface with the same CV structure with (roughly the desired offset. Use “DupEdge” to extract Curve: Done!

Still it would be so much handier to do this inside the “Offset” command.

Cheers, Norbert

Hi,
I will give that a try, what a pain though.
I tried my idea of ChangeDegree then -_Rebuild but it lost the tangent CPs at either end of curve. so -_Rebuild master doesnt keep tangent aspect of MASTER. :rage: damn !!!

I cast lines across from my original 5/8 curve and dragged the control points to them then proceeded to nudge them until the curve visually matched the offset one. It does not take 30 control points to recreate the offset 0.5 curve so why offset has done this to me is annoying. 10 mins per curve and 13 more to do. :rage:

apart from my V5 now acting up and not letting me know when I have selected something, :rage: I managed your method though it also loses the tangent control points in that they are angled and need putting back, and then tweaking to match curve. :grin:
I have several curves to do and two halves for each. all my hard work being done all over again. This has to be wrong. :rage:

I did try for joining the two halves of my curves together (egg shaped profiles) but extend surface doesnt work when they are joined, so no shortcut there, that would have retianed the tangent controls if it was possible. Why is that ?

I had also scale 1D’d every profile so I have that to do all over again as well. :rage:
when I use offsetSrf on a model it causes all sorts of issues, naked edges, holes etc and minute twisty aberations, I didnt think it also attacked healthy curves.

from your linked thread… Learn a great ‘hack-nique’ for off-setting clean curves with equally clean new curves, as opposed to the usual complicated mess. These cleaner curves will make a huge difference in creating higher quality organic surfaces.

This needs fixing please McNeel.

Steve

Steve, except for lines and arcs it’s not possible to accurately offset a curve without using a bunch more control points. The offset of, say, an ellipse, is NOT an ellipse, it’s just not mathematically so at all. Ditto for everything more complex than that.

Steve, what would you have McNeel fix, exactly?

-Pascal

I think he wants an offset curve “loose” option. I honestly don’t see the need as there are so many work a rounds with scaling. You could always just extrude the original curve as a surface then use offset surf loose and dup surf edge. But I guess that would require him to really learn how to use Rhino.

Well… Offset Loose will be in V6; but neither Offset Loose nor any of the V5 workarounds solves the problems Steve is encountering, if I understand. - he wants exact accuracy, but loose.

-Pascal

You can (and should) simply use Match to regain the tangency of your rebuilt curve. Takes about 2 seconds per end.

Steve, the problem is that you are requesting something that is not possible.
Having an offset curve with the original control points while maintaining tangency AND a constant offset just doesn’t work in your case.

There is no “fix” that can change this apart from starting with a more complex original curve to begin with.
Just add a few control points to your original curve, than offset (using “Dave hack” or “Loose offset” method and match the tangency (if needed, it should already be okay at this point)

As a rule of thumb, the more control points you have, the closer you should get to the desired offset.
It will always be a trade off between number of control points and offset tolerance.

If You want identical cv structure between original and offset curve, you have to adjust the original accordingly beforehand.

Cheers, Norbert

Hi Guys,
I went with Pascals idea in this other thread to solve the problem.

      Nice...

@Willem - offsetting EditPoints may make a more accurate offset than moving control points. Not as Macro-able though it looks like.

-Pascal

in thread by Dave…

I will accept advice that an offset needs loads more control points, though when I converted the offset to my existing control point structure, and tweaked the points I got it to match the shape visually when zoomed in enough to be happy with fit.
I then also used Daves video and use of MoveUVN and Pascals advice to him of use edit points which makes a far better result.
I then realigned the tangent control points, then I scaled the shape 1d top to bottom as I had done to the previous offset shape. I need a matching control point count for the aerofoil type profiles so cannot just go with what offset did. It seems whatever method, Daves or Rebuild, that the tangent ends need adjustment (match etc as suggested) afterwards. I will go with whatever is said as acceptable, here its a visual match, visual doesnt seem the normal practice and gets shot down but here it is the way to do it.

whatever has prompted the fixes by other posters, Daves video is quite revealing…
Offset to have a loose option, and/or whatever has caused others to comment on its results.
An option for best fit with existing control points and retain tangengy if thats possible, keeping same control point count, …or just one for keep control point, let user realign tangent ends if they get ‘broken’. So if someone draws an egg shape and offsets it, the curves at top and bottom dont go bent, or user gets to realign afterwards.

Steve

Steve, your talking in riddles…

Steps taken:-

use Dave video method combined with Pascal suggestion.

  1. turn on edit points
  2. MoveUVN enter 0.5 and move edit points on N axis outwards one click
  3. realign bent tangents back to x axis
  4. adjust other Control points to get VISUAL match to original offset curve shape.
  5. use Scale1D to shorten curve to height of original before offset made.

I am doing a visual match, Rhino is normally about snaps etc, here its the Mk1 eyeball. For my tolerance of 2mm, and unable to use beyond 0.01 as per Pascal advice, this will suffice.

I am suggesting Offset command is fixed following the same song sheet as other posters. Some option maybe for best fit given an input option of number of control points required, maybe a degree option as well. I shall leave that to the others as to what would help.

I see 1.3k hits on daves mention of the Offset ‘mess’ as he calls it and his fix, thats half the number of hits on the V5 webinar getting started video !..that tells me a few more apart from myself are keen for some solution/additional options.

Hope that clarifies my riddles

Steve