The panels seem identical and overlap on each other in a consistent way.
I would also like to use a grid that is not necessarily square like in the example above, but more like this one:
Is it possible in Evolute Tools to make a mesh with such shapes?
Would there be a way to approximate all mesh faces into strictly identical ones?
Would there also be a way to then set an overlap between those identical faces, avoiding collision issues (for instance panels intersecting each other)?
The panel cluster function seems to be a good direction but I only have the lite version of ET so far, and I would like to make sure if the pro version is worth buying to solve my problem.
I hope the question is clear, thank you anyway for your kind help.
Charles
Hi @Charligoux_Gatal,
let me answer in several steps:
While EvoluteTools is certainly going to be helpful to achieve such a panelization, there is no simple way to get there. One could decompose the tasks into the following steps:
Approximate the given (freeform) surface using a quad mesh having faces which are as equal and quadrilateral as possible (see green quad mesh below). Either use the subdivision and mesh optimization approach of EvoluteTools LITE / PRO to achieve that, or the parametrization approach of EvoluteTools T.MAP.
Convert quad mesh to the mesh you want to achieve. This can be done using etMeshPatternMapper which is contained in EvoluteTools PRO, or using scripting. Maybe it can be decomposed into a sequence of subdivision steps, I’m not sure yet.
It is generally not possible to achieve such a panelization consisting all identical panels, unless your reference surface is very very simple.
The panel clustering functionality of EvoluteTools CLUSTER would be a possibility to analyze the resulting panels for repetition. However they will need to be planarized first, which will definitely result in collisions as mentioned by you. Unfortunately there is no way to avoid that by construction. Scripting will be required to approach this task, and feasibility will heavily depend on the reference surface and panel size.
Summarizing, this will be a quite challenging task, and it’s not guaranteed to work well
The photo you posted suggests that are not flat every but triangulated. Do you know more about that?
Thank you very much for your clear and complete answer (and sorry for taking so much time getting back to you).
It’s a clever approach! I will look at the etMeshPatternMapper function more into detail.
Regarding the panellization, I found another software (Kangaroo plug-in) which has a tool just for that. I also developed something in Grasshopper that sort of wiggles the panels until they reach a minimum of overlap. The last problematic panels can then be adjusted manually.
I got some more information about the building on the photo, it has been apparently made “by hand”: only fixing rails were calculated and placed. The tiles have then been added and adjusted one by one.
Again thank you for your detailed answer, it helps a lot.
Hi @Charligoux_Gatal. I was part of the team which came up with the concept for the panelling of this pavilion (I previously worked for Ramboll). The panels were originally set out using the chebyshev net approach, similar to this. This technique is limited to generating quad meshes, unfortunately. Although, as @snabela points out, you may be able to map your pattern to a quad mesh. The fixing rails were later derived from this method (originally each panel was to be mounted independently). Some of the panels had to be split on the diagonal to accommodate areas of tight curvature, but most are planar (simple square tiles). I wasn’t responsible for the final setting out of the panels – it’s entirely possible that some manual “fine tuning” of the tiles may have taken place!
Sure! I did not yet set it up in Kangaroo so I have nothing to show, so I will just tell you about the functions I found that could do the job (I was a bit enthusiastic when I mentioned a “tool just for that”).
If you don’t know it, Kangaroo is a plugin which adds more tools in Grasshopper, which is itself a plugin for Rhino. To panellize something, I would :
Roughly map the desired pattern over the surface using a strategy like what you described (fit pattern on a square, then meshing the target geometry with square cells, then fitting the square pattern in each square cell).
Use the following functions from Kangaroo (V2): “ClampAngle”, “ClampLength”, “Planarize”, and “OnMesh”. The two first ones allow you to control angles and edge length in your mapped patterns, in other words the global shape of your tiles. It evens everything out. The third one flattens each tile as much as possible, and the fourth one keeps the tiles on the geometry to put the tiles on.
I did not use this approach yet, this part of the project is now in stand by.
Thank you Rajaa, it’s also a very good option to map on our surfaces. Especially since it works on spheres too, which was always a critical shape for some panelling methods I found.
Thank you Will, what a small world Amazing! An architect we know just guessed how it was made based on his experience. I am very interested in the method you described, thank you for sharing some information about it. I will definitely dive into the paper.
Thank you everyone for your kind help, it’s very much appreciated! I can update you with methods and results when we get back to those tiles.
Amazing! An architect we know just guessed how it was made based on his experience. I am very interested in the method you described, thank you for sharing some information about it. I will definitely dive into the paper.