Hello, how to create a symmetric surface at symmetric curves? Try to create any curves (for symmetry we use Mirror), and now we create a surface on a grid of curves or on rails - the final surface is asymmetrical.
It is possible to cut of course a surface and to reflect a half for symmetry, but why so?
What do you think of it?
Post an example of symmetric curves and and resulting asymmetric surface, and describe what commands and options you used to create the surface from the curvesā¦
No. 
Compare a red grid (I have reflected the left part) to the blocked grid (initial right geometry)
Untitled.3dm (203.9 KB)
And the grid has to coincideā¦
did russia dry out on vodka? or why are you trying to do this 
network surface can create imprecise surfaces you should slowly know thatā¦ really, skip that vodka i am telling you, its nooot good for your brainā¦ 
use sweep2 with those curves. if you have the option sweep2 or network, then always choose sweep. anyway why not using EdgeSr or Loft? next to Patch, NetworkSrf is the 2nd on the list to avoid if you can.
If you want a precise solution use the Loft command.
The absolute tolerance in you file is 0.01, which also is the tolerance for NetworkSrf when I start it in your file. Is that what you used?
The maximum deviation of the edges of your initial surface and mirrored surface is 0.000908579. This is less than 10% of the absolute tolerance. The maximum deviation I found between the initial surface and the mirror surface is 0.00199357 which is less than 20% of the absolute tolerance.
The results appear reasonable given the realities of finite precision digital calculations and how NetworkSrf works. I agree with RichardZ and jim that EdgeSrf, Loft or Sweep2 would be preferable for this geometry.
I am a nondrinker 
However, if to create other more difficult grid of curves - the asymmetrical surface anyway turns out.
I thought over itā¦
Time initial curves identical - total surfaces have to be identical too - it seems logically. Probably I am mistaken!
Do you mean the tolerance which I specify in the NetworkSrf window?
However because of it periodically there are problems with join - I want to understand the decision. Two surfaces though are symmetric, but nevertheless there are small deviations. 
Yes
Surfaces should join if the deviation between adjacent edges is less than the absolute tolerance specified in DocumentProperties. There are multiple reasons why adjacent edges may not identical. Those reasons are usually different than the differences seen here.
NetworkSrf rebuilds the input curves as degree 3 curves with the knots and parameterization arranged such that an interpolating Gordon Surface can be created. Presumably the use of finite precision digital math results in the resulting curves and surfaces not being exactly symmetric.
There 2 numerical parameters for control of accuracy - about what there are a speech? Or both parameters have to coincide with document accuracy for symmetry of result - it possibly?
Then I have been sure that a problem in asymmetry, but perhaps I was only mistaken with an accuracy in the NetworkSrf window.
if you use the option simple sweep which in your example is applicable you will see that the split surface will be 100 % symmetric. i have of course no idea what kind of other ācomplexā surfaces you are trying to create. maybe you should post a ārealā example then.
dont use network if you are asking for a 100 % accuracy. you can drink not drink believe not believe but you will not find peace with this method if infinite accuracy is what you need.
oh and by the way i am going to drink one of those now non alcaholic of course, i tried the ginger already it also contains cucumber juice its tasty. just now i will try the rose lemonade - ŠŠ°ŃŠµ Š·Š“Š¾ŃŠ¾Š²ŃŠµ
@RichardZ, @davidcockey, I so understand, at symmetric curves the symmetric surface, but with small deviations from 100% anyway turns out. Recently I experimentedā¦ Sometimes all tools give similar result - there are no 100% of symmetry. And thereās nothing to be done.
I have assumed that the asymmetry somehow get rid (without Split and Mirror). Somebody can precisely tell - similar small deviations from symmetry can be Join cause of error?
For example, I have a difficult roof of the building which is symmetric on axis Y (the surface is executed without Split and Mirror, and therefore there are deviations). Further I do additional geometry which adjoins a roof (it is only a half on axis Y). I carry out Join - there are no open edges, reflect geometry on an axis Y, Join - there are open edges. If to cut all geometry the line on Y and to execute Mirror on axis Y - open edges are absent.
What can the problem consist in?
Or for normal symmetry it is necessary just to cut and reflect?
How do you know there are āopen edgesā? Are you using ShowEdges with āNaked edgesā checked? ShowEdges is the best way to check for unjoined edges.
Sometimes there can be apparent gaps between joined surfaces which are caused by differences in the display meshes.
Can you post an example of surfaces which do not join?
For check of edges I use ShowEdges
@davidcockey, @RichardZ, Hurrah!!! 
Though any surface canāt be symmetric for 100% is not a problem at all!
In my case the problem was only that I have specified insufficient accuracy in the NetworkSrf window.
If I specify the accuracy which is specified in setting up the document in the NetworkSrf window, then everything is excellent, accession without mistakes.
But it is enough to specify weak accuracy in the NetworkSrf window (accuracy which doesnāt correspond to the document) - on the one hand accession works, but from other party (after Mirror) Join doesnāt work at all.
