Im quite new to using grasshopper and I have somewhat of a similar request. I have a surface (Layer v1) I was forming as a roof for my project however it has become far too complex. I tried converting it to a mesh (Layer mesh) and remeshing it but now I want it to convert it back to a surface while also smoothing it.
in my grasshopper file my first cluster is where I tried to turn it into a surface, however it doesnt look very good at all. So I wanted to try after to try and rebuild my surface from a set of rebuilt curves in the second cluster. However, I do not know how to continue. The rebuilt curves are on the layer (Rebuilt Smooth).
I would greatly appreciate any help in Reconstructing my roof surface. Whether that is:
-Somehow making my original surface less complex, without ruining its shape
-Turning my Mesh into a replica surface and not an ugly surface
-Or Rebuilding my surface from my mesh curves
Roof1.gh (15.4 KB)
I cant seem to upload my rhino file as its too big but Ive added it to a dropbox
I assume thatâs why you ask if thereâs a way to make the original surface less complex - however a relevant question would be why is the surface like this in the first place? Canât we avoid it being made this way?
What do you mean by âit doesnât look very goodâ ? Are you referring to its creased appearance as opposed to a smooth one?
You can turn your 4-point surfaces into quad meshes, join them, and convert to a SubD object so it looks âsmoothâ:
Obviously the SubD isnât a âsurfaceâ, but maybe it doesnât matter?
Another fair question would be if youâre looking for a single, untrimmed surface - then it would matter I guess
Hi, thanks for answering! Oh sorry this is my first time posting and I just came across this thread while looking for answers.
When I was saying it doesnt look very good I was referring to the surface that can be baked from the top cluster on grasshopper, which isnt yet baked in the rhino file, as it just doesnt achieve the result Iâm looking for.
Ah yes that definitely looks pretty good, although ideally as you have mentioned I would like to end up with a single untrimmed surface. Would you have any idea how to do that? I dont know if you had a look at the resulting surface from the original grasshopper file I had, but as you might have seen that one doesnât end up looking very good at all.
Also were you able to have a look at the curves as well? On my grasshopper file I also have a cluster where I try to rebuild the curves from my mesh, which you can see in my picture as well. The curves looks a lot smoother, so I would love to be able to make a surface that follows these curves somehow.
This is going to be tough or not really doable given the shape youâre dealing with - to get a single untrimmed surface, such surface would most likely need to exceed the boundary of your trimmed region - plus it seems youâre trimming it for a reason, so youâd end up trimming it again for the surface to be within your roof trim boundary. I did see what youâre doing with the rebuilt curves, perhaps to loft them or to make some grid via curve divisions - nonetheless your curves have been obtained from sectioning a trimmed shape - hence the transition from curve to curve (some short, some long, some split into 2 curves from plane-surface intersections) will not give you the desired result.
Iâd say youâre better off with the SubD method.
Once you get the SubD, you can convert it into a brep, which gives you a polysurface - itâs not the same as a âtrimmed surfaceâ just like your original open surface wasnât, but it would have a smooth appearance.
You can feed these curves (and boundary points) into a âpatchâ component, which will give you a single untrimmed surface approximating your rebuilt curves (as mentioned before this surface will exceed your curves and youâll probably want to trim the surface again based on your roof boundary) - nonetheless the patch will remain as an approximation only, and not an exact transition from curve to curve - again, up to you whether it should matter or not - you can play with adjustments in the patch component, divide your curves by same length, pull these points onto the patch surface, and get a âsenseâ of the âdeviationâ from the rebuilt curves to the patch surface by measuring the distance(s) between curve division points and pulled points - just keep in mind patch adjustments can turn out to be slow: Roof1.gh (20.5 KB)
In this particular instance as you can see the average distance from curve division points (every 100 units) to the pulled point is 0.04 - perhaps not bad if â0â isnât the goal.
Thank you for taking a look at this! It definitely looks like it follows the form quite well in the preview, however I must say once I bake it it shows some areas that could perhaps be a bit smoother. Do you Have any suggestions perhaps from here how the form could be made more smooth?
Perhaps it has to do with the mesh edges that I extracted, which donât seem to be very smooth. It seems when I tried to convert it to a simpler mesh the geometry became less smooth. The edges from the SubD solution seem to be a lot smoother than the mesh I used. Perhaps there is a way of making this trimmed patch look somewhat smoother?
Thanks a lot for joining the discussion The image u have added seems to look quite good!
I must say that I cant seem to be able to run this gh file though. Are you using any plugins?
Also from what I can see I am assuming you are using the original surface, am I correct? It seems to be simplified quite successfully! Thank your for that. My secondary goal other than simplifying the surface was to smoothen the geometry a bit, perhaps get rid of some of the edges that are draping too much. Something that I have attempted to do using a few methods and what some of the other people have touched on as well. Would you have any advice on this?
I keep it short, because its, in my opinion, the only correct answer to the topic of mesh-to-surface conversions. Although professional reverse engineering software has made great progress and occasionally a workaround in GH is feasible, a perfect conversion is hardly doable at this point. And this is because a surface model is more than a mesh. There is a lot of guessing involved in this process. You either build and stick to surfaces, but this requires a decent understanding of this kind of modeling. Or you choose to model it in a different form (mesh, sub-d, clay-scan etc) and then you remodel it manually based on this template. And of course only if you require a surface model at all. I hope this helps you somehow, at least this post should prevent you from wasting your time too much
Yes thank you, I think Iâve got somewhat the result I had in mind, using your prior help. Would you have any idea about the reply @Quan_Li gave me and why it might not be working, since I have not heard back from him. Are there any plugins he has used that I need to get? Thanks for all your help so far Rene.
Starting from an extrcted RenderMesh of surface V1, then using the Tolerance based Surface Approximation to Mesh vi RhinoReverse RROnePatch commad gives the results in the attached files (Tolerances used 100 and 1000 âunitsâ).
