Round-Bottom Hull in Rhino - how?

I've got an old table of offsets (probably 30+ years old) for a round-bottom boat I'd like to model in Rhino. I've only ever worked on chine hulls - I draw the sheer, the chine, and the keel, then add a surface between them.

How, conceptually speaking, should one go about modeling a round-bottom hull in Rhino? I've got a sheer curve and a keel curve, and then a whole mess of buttocks and waterlines and diagonals in between. Would the best bet be to draw the sheer and keel, and then the diagonals and then loft or sweep a curve between them? Just trying to figure out how to proceed without drawing a zillion points/curves.

Thanks in advance.

 

Create the stations (just work one side) from the offsets by inputting the points and then fit a curve through each set of points. Then loft the stations.

Use the pictureframe or background bitmap options to project the scanned plans that you have onto a background. Then you can use the profile and sheer curves to check the position of the stations. Your drawings should have the stations on as well? If so you can simply trace them rather than entering offsets.

There are a few tutorials around, maybe someone else can point you to one.

 

I've been busy creating surfaces from offsets and plans for numerous traditional round bottom hulls in Rhino recently, and should probably write up the process I use along with illustrations. Unfortunately I need to get a project out the door so it will have to wait.

Quick summary of the process I use:

1) Use an Excel spreadsheet template I created to input offsets and output a CSV file with point coordinate triplets. Import the CSV file into Rhino. This process generally takes under an hour.

2) Use CurveThroughPt and/or InterpCrv (selecting imported points) to create curves for the transverse sections, water planes, buttocks, sheer, rabbet, etc.

2a) Diagonals are slightly more complicated. I import the points flat, then move and rotate the sets of points into position. Then draw the diagonal curves. My experience thus far is most diagonal points in offsets are not as accurate as other points and are frequently best ignored.

3) Check the curves for obviously bad points and discrepencies. Typical discrepancy are when a buttock and a waterline are close to each other at a station but if the section obviously doesn't actually go through both. Revise and sometimes discard points as appropriate. Buttock curves generally need additional points at the sheer. Create these points by intersecting the planes which the buttock curves lie in with the sheer curve. Occasionally waterline curves need additional points created at the rabbet and transom. Re-create curves which go through altered points.

4) See if the waterline curves and buttock curves intersect everywhere. Usually there will be some misses. Decide if the misses are too large, and if so which curves should be altered. Revise points and recreate curves as needed.

5) Inspect the curves and decide if they are fair enough. If not revise. I generally don't do final, very precise fairing of the curves. This may be controversial for folks used to traditional lofting methods - separate discussion.

6) Decide what curves will be used to create the surface. Decide on surface topology, ie arrangement of the boundary curves and interior curves, and command used to create surface. I usually use sheer, rabbet, keel, transom edge, sections and waterlines with NetworkSrf. I occasionally skip the waterlines and use Sweep2. Topology of the surface is important, and can be tricky if the stem rabbet flows continuously into the keel rabbet. Alternatives include (hull with stem rabbet flowing into transom used as example):
A. Surface has three boundaries: sheer, stem and keel rabbets joined, and transom edge. This can provide a reasonably decent surface but the result may cause problems for further manipulation such as generating offset surfaces.
B. Pick a point to break the combined stem rabbet and keel curve. This point may or may not be at the base of the stem as shown on the plans. Any sections ahead of this point can't be used. Any waterlines below this point can't be used. Surface will now have four boundaries: sheer, forward/upper portion of rabbet, lower/aft portion of rabbet, and transom edge. This generally works but the surface will generally have a singularity at the rabbet break. Also causes problems for further manipulation.
C. Extend the surface past the bow and possibly below the keel. Do this by extending the waterlines and sections, and create boundary curves for the surface which go through the extended curves. Additional sections and sometimes waterlines are needed to properly capture the shape of the stem and keel. Extended surface is created, then trimed using stem and keel rabbet curves. This may need some illustrations to be clear. It usually results in a very good quality surface which can be faired with various methods. Care is required though in extending the curves, and creating the new boundary curves. This is usually my prefered method for creating the surface.

7) Check the surface for agreement with original offsets (less the obviously bad points) using PointDeviation. If not close enough go back and revise what is causing the problem.

8) Fair the surface (if needed) using a variety of techniques which need their own description.

How long does it take? Depends in large part on:
1) Purpose of surface. Requirements of a surface for analysis are not nearly as tight as if molds will be made from the math data.
2) Complexity of the surface and how many offsets their are.
3) Quality of the offsets. This can be a major factor.
4) Other information which needs to be developed. For instance if the stem rabbet is given as a drawing, or if the offsets for the waterlines and buttocks are to the inside of planking but the stem and keel are given as outside rabbet locations.
A quick surface for analysis from good offsets has taken as little as two hours. A high quality surface for a complex shape from not so good offsets may take me several days.

 

I should add another major factor in how to do and the time it takes is how accurate does the surface need to match the offsets. Just needing to look about the same is very different than if high precision is required.

 

Thanks for the help. I started with MikeJohn's method above, entering points for each station and creating curve through those points. When finished with the stations, I'll create a surface through each one.

I'm curious about fairing, though. Do you fair each station first, and then create a surface through the faired stations? Or just get the stations close, create the surface, and then fair it? When doing a hard-chine boat, I always draw and fair the sheer, chine, rabbet, and keel first, then create the surface (although admittedly I've never actually tried to build a boat from surface produced, so I've no idea whether my method works or not. Looks nice on screen tho'.)

 

as you are concerned about fairing,before going to surface generation, you should analyze your curve fairness by using the command Analyze>curve>curvature graph on ,,, the curvature graph will help you in understanding whether your curve is fair or not and in fairing it.

To understand the basic of curvature graph you can go through the manual,

 

Thanks, but I understand the concept of fairing and how to use the analyze curvature feature. My question was more about process: Do you typically analyze and fair each individual station curve first, and THEN create surfaces through all of them, or do you use the rough (unfaired) station curves to create surfaces, THEN do the fairing.

As mentioned, on hard-chine hulls, I use the former method - but I'm new to round-bottom hulls.

 

adt2,,, you should do the both...because, sometimes, provided that the curves are fair, the surface may not be fair,

after crating the surface you should check the surface fariness, by taking sections in different possitions(seeing if the sectional curves are fair), and by using the command Analyze>Surface>Curvature Analysis

You should keep the control point as low as possible in the surface , so that you can do the control point manipulation for fairing. If you keep less points in the curves, lofting those curves will result in a surface which has less control points,

 

As pavel915 said, you should fair the curves before creating the surface, but fairing of the surface may still be needed. When fairing the surface I will go back and revise the curves if that will solve the problem before getting into moving surface control points.

Other good tools for fairing are Zebra, and EMap (Enviormental Map) with the fluourescent tube. When using EMAP move the surface around in Perspective View and watch how the image on the suface distorts.

Keeping the number of control points low is good advice but can be difficult to do when creating curves from offsets because with N offset points creating a curve you will have N+2 control points. Fit, Rebuild and RebuildCrvNonUniform can be used to reduce the number of control points. Be careful not to alter the curve so it winds up a different shape.

 

If you have N number of offset points and N is a big number then, you can create a control point curve using a few control points around the offset points, which fits those points approximately and then you can adjust the curve by draging the control points to fit the curve to those offset points. This will give you a smoother curve using less control points.

 

Excellent suggestion!

Rhino also has CageEdit which allows objects to be edited by capturing the objects with a another object which becomes the control. Cage may be used to create a bounding box as the control object, but almost any curve or surface can be used. Control objects do not have to contain the captured objects.

 

When creating curves using CurveThroughPt and/or InterpCrv from offset points which are not evenly spaced, for example section curves, the Knots = Chord option will usually provide a better curve.

 

I recently rebuilt a hull by using a single surface with direct manipulation of the control points. I used the pointdeviation command to give me a clue of how far out I was. I was quite happy with the way it turned out (mean of 50mm deviation on a 130m hull). There is a hull built using the single-surface method here: http://www.boatdesign.net/forums/design-software/rhino-sweep-2-rails-vs-loft-hull-surface-17697.html

I used to have a tutorial around somewhere, but I can't find it now after many years of neglect.

Tim B.

 

Thanks, the pointdeviation commands look cool...