Jump to content

Strange Shape How To?


Go to solution Solved by Cerbera,

Recommended Posts

I have completed the “Kelpies” no time to explain Google it! Render is on the way. Next part of the project is the Falkirk Wheel, again Google it!

They are some of the iconic scenes in Scotland (Totally Amazing Engineering).

Anyway, the Falkirk Wheel is a challenge and was the Kelpies were, however the Visitors Centre is doing my head in. It is the shape of the building that is confusing me. I though firstly using splines but I think that will cause problems with the windows. Now I think a Primitive with some sort of a deformer may be the solution.

Sorry but I could only upload 2 images.

 

Any feedback would be great.

 

Cheers Jacobite

2.jpg

3.jpg

Link to comment
  • Customer

That looks like a thin slice of a big sphere.

Why don't you ask them to provide you some top view architectural designs ? I know it might be a bit hard or time-consuming to retrieve such designs but in this case the geometry is very confusing if you try to reference it just from photos from google street view. Even a fire-exit plan could be useful.

And depending on the detail-level and photorealism they hired you for the architectural plans become more and more mandatory. 

image.png.da6b2820378c03c5f42ca6d2ce54ed93.pngimage.png.cb1f969a2692530cd3d1523e4f6a3b54.pngimage.png.a2d98969d21f45e8c8be447b46ad8fca.png

Link to comment

No disrespect to anything people have suggested above, but I don't think anyone has quite nailed that shape yet, which isn't so much like a segment of orangey roundness as we might initially think. If it is, it's a hell of a long orange ! And nobody is doing what I'd consider the most important thing, which is using what we've got (reference-wise) to determine a highly accurate shape, see below...

 

I agree that ideally you need the arch plans to do this fully accurately, but we are helped to a really remarkable extent by the grid panelling on the front, which allows us to establish accurate proportions easily guided by that alone.. There is a raised flat section about half the height of a human at its base we have to take account of, which is a further deviation from the spherical segment plans suggested above.

 

But the great news is that there is some flashing that covers the ridgeline, which, somewhat crucially means your front surface does not have to be contiguous with the rest of it, and there are lots of reasons you wouldn't want it to be so ! So that's all good ! 🙂

 

CBR

 

 

Link to comment

This, I think, is the first couple of moves.

 

We start with a plane, and immediately establish the right amount of segments, colorizing them temporarily so we can see what is windows and what isn't. It's 42 x 10 btw, 9 of which are in the angled section.

 

Even though the building profile might be symmetrical, its doorways aren't, so we are limited to using that only initially, but that can be quite helpful for drawing an exactly symmetrical spline (spline pen) using your front reference to be quite accurate about which panels that intersects where, like so... you could also do this with the arc primtive, but trust me - it's easier to just draw it with 2 clicks and a drag !

 

image.thumb.png.3577355ecae42428b1f2680f591cd980.png

 

We don't have to worry abut either triangles or ngons here, in fact this is one of the very rare situations where ngons are preferable and 'most correct under the circumstances', especially where we need the 'double split' ventilation panels above the windows and there is no need to continue that segmentation anywhere else, connected or not.

 

In the above green arrow is the symmetry line, and red arrow shows the 'hinge point' to which we will snap our modelling axis when we come to rotating this off vertical.

 

Note we have exactly square panels EXCEPT at the base, which will become the flat section, and the L/R edges, which matches the reference with those narrower panels we get at each end. Next step will be either boolean-ing away the excess with a very high poly Operand B so we get immaculate curvature without subdivision or, merely cutting it out based on the spline reference like I did below (it'll be covered in flashing panels anyway). And that should give you everything you need for;

 

a) making the front panel details - you can make the grid using edge to spline and sweep(s) for example.

b) making the upper curved back section from the spline, possibly via a sliced lathe that is subsequently made editable and extruded at the base to match that flat section.

c) making the flashing that bridges the 2 sections (Swept L profile + thickness unless you want to make each ridge tile individually, in which case a cloner setup using that spline might work (not at the ends though - they'd have to be manually completed !)).

 

By the way, the 'angle of dangle' on that front bit is 51.6 degrees off Y, based on an orthographic(ish) side view I found.

 

 

image.thumb.png.45ea6d459a8bd81e991c0c9d21137203.png

 

Then we can delete the polys where the doors will be, and patch those in perpendicular to the floor (also shown above).

This will be the basis for all geometry you need for the front side.

 

Next up, the back. Lathe FTW here. We need to rotate the lathe 90 degrees so X is facing up, THEN child the spline to it and set a sectional angle of -38.2 degrees and set our spline interpolation to a high (16+) uniform type value, at which point we can make it editable straight away (having taken a copy you can go back to ideally) and extrude down the flat section to meet the frontage, giving you this sort of result...

 

image.thumb.png.a3000c29f95f56fa8f504910d4ff6c09.png

 

image.thumb.png.364cce467f02e460ae11539de5400afb.png

 

This is of a suitable density to either pull polys out directly to get the adjoining buildings that come off the back of it, OR to cleanly intersect them with / use as boole operands if that is easier / more optimal to your final plans. I quickly tested the former (and IMO superior) option below...

 

image.thumb.png.03bbebe5483702fb56bda46a2bcb9d91.png

 

Hope that helps...

 

I do hope you are making that barge lift thing in front of it as well, and want help with that at some point ! 😉

 

CBR

 

Link to comment
  • Solution

Ah yes, the grid and flashing rim thing works nicely too...

 

image.thumb.png.3831ab5564f29b12eabf46204a526624.png

 

I think the next move would be to split off the walls into a separate object and then give both those and the glass / solid panels independent thickness so that they react properly to light. That doesn't have to be done on an individual panel-by-panel basis if we remember to make sure the gridwork intersects both sides of the glass, meaning light can't bounce from panel to panel out the sides, if you see what I mean. Intersection here will be very clean, and unproblematic at these sorts of scales, unless you are planning to crawl a camera inches over the surface of them.

 

CBR

Link to comment

CBR, you are a Master. You make the most technical stuff sound so simple. I can't tell you enough how much I appreciate the work and time you put into this. I will follow your instructions as well as I can but please don't delete your project file, who knows what I may end up with.

 

With all the amazing help from the most amazing people around the world "This is what makes Core4d the place to be"

 

Jacobite

Link to comment
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • LATEST ACTIVITIES

    1. 7

      Draw primitives on surface of existing objects?

    2. 0

      Nodes Modifier result isn't updated anymore

    3. 220

      Scene Nodes | Capsules file pit

    4. 3

      Flipped clones in multi-segments curve

    5. 2

      How to create continuous UV texture for irregular wall like shapes ?

×
×
  • Create New...

Copyright Core 4D © 2023 Powered by Invision Community