Here is a "easter egg" for SketchyPhysics2. Buoyancy. It wasn't quite ready for beta1 so I disabled the menu. But it still works and in spite of the problems I am going to turn the menu back on in the next version.

Buoyancy isn't water exactly but it allows objects to float. It is unlike water in that it has no waves and cannot "flow". Also machines like a propeller or a water wheel will not work.

The buoyancy plane is infinite but it can be moved and even rotated. The new version will expose a easier way to create the plane and expose a few attributes like current direction and viscosity.

To create a buoyancy plane type "makeWaterPlane" into the ruby console.
NOTE: Its harder to create a model that floats right.
-Objects that are too small (<1ft) move like they are in a thick fluid.
-For best results you need to change the "Physics Settings" values. The defaults don't usually look right. Changing object density, gravity and scale can make a big difference.
-Compound objects(groups of shapes) will just sink like a stone. Instead you need to connect all of the individual shapes to a single joint. Create a single slider joint that has min=0.0 and max=0.001 and attach all the shapes to that. That will act like a compound object but will float more realistically. Look at the examples.

Chris

water5.skp

SailBoat5.skp

@kwistenbiebel said:

...exactly the same here. I see no errors in the ruby panel though.

Wow. I tried it on 2 different machines and it works. Maybe you are using a Mac? Altho I don't see why that would make any difference..

@unknownuser said:

I can't get this script to work. Nothing happens.
I create a sphere, group it, right-click and do 3x3 FFD, enter the new group with Construction Points, move some to another position, exit the group, right-click the sphere group and choose Update FFD. Nothing happens.

Can you open the ruby console and see if it is printing any errors?

Chris

@krash88 said:

It looks great in it's current form, but I get this error in the console... Maybe I'm doing something wrong...

Error: #<NoMethodError: undefined method definition' for #<Sketchup::Group:0xbec0960>> C:/Program Files/Google/Google SketchUp 6/Plugins/SketchyFFD.rb:224:in analizeLattice'
C:/Program Files/Google/Google SketchUp 6/Plugins/SketchyFFD.rb:60:in updateFFD' C:/Program Files/Google/Google SketchUp 6/Plugins/SketchyFFD.rb:13 C:/Program Files/Google/Google SketchUp 6/Plugins/SketchyFFD.rb:13:in call'

Ooops! I fixed it and updated the script in the original post. Kwistenbiebel this is probably the same bug you were seeing.

Chris

@eldar said:

hi! i was wondering if there's a way to lock the pose in SketchyPhysics, maybe with an example i can explain myself (english is not my first language) i downloaded the model of the robot arm, open the model, push the play physics button and start to move the fingers and made a cool pose, but when i finish and use others sketchup buttons the pose is lost, my question is: there is a way to not lose this pose or save it, stay with the worked pose and continue working with the tools of sketchup, draw something, etc? (maybe save in a layer?)

This is very important also in my work as an architect to design solar protection windows and stuff like this. Amazing tool by the way, thanks!!

I understand. That should be possible. Ill add it to my list.

You could do something similar now if you take advantage of a bug. Just save the model while the simulation is running. It will save it in the posed position.

Chris

UPDATE April 2021
Hi Everyone. I am happy to announce that mind.sight.studios is going to be taking over development of SketchyFFD effective immediately. Please see this thread by Dale here for more information:

https://sketchucation.com/forums/viewtopic.php?f=323&t=73658

And I'd like to thank TIG for helping keep SketchyFFD updated the last few years.

Chris

[Original Post below]

[NOTE: These instructions are out of date. Please see this thread for updates]
https://sketchucation.com/forums/viewtopic.php?f=323&t=73658
Instructions: (somewhat out of date. see updates for changes)
-Create a group with all the geometry you wish to deform. This group should contain just faces and/or lines.
-Right click the group and select one of the options in the FFD submenu. Either 2x2 or 3x3 or NxN FFD. Depending on the size of the model it could take a while (seconds I hope). At the end you should see a new group containing nothing but control points
-Open the control point group and move the control points as you wish. The object should deform as you move the points. NOTE: it can take a long time for complex models. Look at the status bar in the lower left to see the status of the deformation.

Bugs and limitations:
-BUG. If you start a FFD and the control points don't match the object you need to explode and regroup the object. Then try the FFD again.
-Only one FFD at a time. The script uses a few globals for speed.
-Only works on geometry in a group and only one layer deep.
-Sometimes hidden edges don't move right. Dont know why. Might be a bug in Sketchup.

Update Feb 12. New version:
-Support for NxN deforms.
-Mesh automatically when you move control points.
-Added status text progress when it is weighing and deforming the mesh.

Update Mar 21.
-Moved all FFD right click menu items to a sub menu.
-Fixed a bug that forced you to have a least 2 control point in each direction. You can now have do 2D grids. IE. 1x4x3.
-Added an option to create a NxN 2d patch. A grid with a control point at each intersection.

Updated May 9 2008
-Added ability to lock edges in the mesh so they don't move during FFD. Select edges then right click->FFD->Lock edges.
-Added an option to subdivide the mesh when doing a NxN deformation.

Updated Aug 24 2009
-Optimized the initial weighting to be about 3x faster. Idea by Stefan.

Updated 20110810 by TIG
-Code is now inside a module, all $ variables now @@ or @ types, the flaky observer code is fixed.

Updated 20130203 by TIG

• Recently reported NxN issues with NaN [0.0/0.0] & Infinity [1.0/0.0] fixed, which occurred when selected group was 2d & NxN height=1
  and the NxN dialog now remembers the last used values that session for the NxN lattice [width, depth, height, subdivide]...

@wikii said:

Add a RB script to:
Drop down menu? difficult to use~
Toolbar? cost a lot screen area to display~
Context menu? it is already too long~

SUTool organises RB scripts in a AutoCAD_screen_menu_like sidemenu ,as the Gif show.

SUTool v0.06 has ability to add in a new RB script automatically,as the Gif show.

In chinese:
SUTool是一个以类似CAD屏幕菜单方式组织SU插件的工具，用户可以通过自定义INI文件，增加或者减少项目。让大家能够方便的管理自己喜欢的插件！

SUTool v0.06 可以自动添加新的插件：直接对RB文件进行分析，并将其中的命令自动添加到用户指定的ini文件中，只需Reload SUT一下，即可使用新的插件。 注意，被分析的RB文件不要放在含有中文字符的路径下，比如桌面。

Looks like it could be a nice script. Can anyone translate?

Chris

@wright said:

I made a ramp using regular sketch up. then I made a sphere using sketchy solids. but when i turn the ramp to a componet and but the sphere on top, the ball goes down a little then just stops. So i move the ball so its right on top of the ramp. then the ball floats upward. any idea how to fix this?????

Post your model and Ill take a look.

Chris

@pixero said:

It looks great. Is it possible to have lines connecting the dots? That would in my opinion make it easier to see the deformation and which points are moved where.
I think it could be hard to see sometimes otherwise.

Would it be possible also to see the deformation in wireframe like in the new sear transformation script?

I have a version that shows the lines, but in practice the lines get in the way when you are trying to move the control points. I think it could be done right by writing a custom tool. But that is a bit more than I can take on just now.

Generating a wireframe would probably be too slow, but I will check it out.

Chris

@jclements said:

Hi Chris, as always, very impressive.

A few questions:

Are the deformations restricted to just the SU default axis?

If the group/component is rotated manually before being deformed, what determines the deformation "planes"?

Is there an ability to deform relative to a specified point, (ie., center, opposite handle, or inference point). Is there concentric deformation relative to a specific point?

Regards, John

Sorry for the delay responding. Missed this post.

When I initialize the Free Form Deformation (FFD). I create a 2x2 or 3x3 array of construction points that span the bounding box of the group. For 2x2 its one construction point at each corner. 3x3 has one at each corner and each midpoint.

Then I calculate a weight for each vertex in the group based on the distance to the construction point.

The user can then move the construction points any way they like. Singularly, in groups, scaled, rotated whatever.

Once you are done moving the points you select update and the script will move each vertex in the mesh based on the pre-calculated vertex weights.

Does that answer your questions?

Chris

Here is an example with more control point manipulations.

@unknownuser said:

It seems that at least you triangulate the faces, from what I could judge with the deer's kees (and the few faces that were not triangular in the original model).
This is a great tool, and judging from the video, the performance are more than "better"! (compared for instance to a simple shear of the model, that would take 40 seconds with a Ruby script). I imagine that you use a compiled version in C++).
I would encourage you to continue on this path, really!
If you need some help for the GUI, just post a description of what you need and I am sure we would find a way to contribute.

I don't triangulate but I do explode curves. That is probably what you are seeing..

The animated gif is about 2x playback so its a bit slower than it appears. But it is all written in ruby. I expect it would be a bit faster if I moved some of the calculations to C++. But I think the bottle neck is the moving of each vertex.

I might take you up on the gui help.

Chris

@rayl said:

I'm having a slight problem with the use of a controller. I am using a wired X-Box 360 controller and seem to be missing one of the input commands. Here's the set-up:

Action joyLX and joyLY are associated with the left joystick as they should, but joyRY seems to be associated with the two top analog triggers. I seem to be missing joyRX, as the right joystick has no function.

I know this version is still in it's beta stage, so if no fix is available I completely understand. Or it could just be something obvious that I may be overlooking.

Yeah, the 360 controller is a bit different from most. I am going to make a configuration utility for the next version.

Chris

@rayl said:

About readback collision geometry, how would you fix that if I ever came across such a problem? With the first model that I had that problem when I turned readback collision geometry on and noticed the wiremesh was not consistent with the actual tire model, it seemed to be floating a bit higher than the tire itself. Would it be a matter of just exploding and re-grouping the model in question?

Yes, That is the easiest way to fix that problem.

@rayl said:

Thanks once more for the time you've taken to help and for developing such a great program for Sketch Up users. I am enjoying it very much.

Your welcome.

Chris

@rayl said:

I tried to upload the file but it surpasses the file size, so I tried removing part by part trying to lower the size to 512kb, and when I reached the limit allowed it just happens to be that the problem goes away. I'm going to assume it has to do with the amount of weight applied to the front end.

If that's the problem, is there any way to add additional accessories or additions without adding weight to the main frame?

I included a file with 3 pictures showing the problem. Thanks again.

Looking a the pictures it might be a weight issue. You could set some of the objects in the group to ignore. That way they don't weight anything. But they wont collide either.

Try this: the big box in the back of the cab. Mark it ignore and then rebuild the shape of it as a hollow box. 5 or six thin boxes, one for the front,back,top and each side. That should make it weigh a lot less.

You might also want to make sure that everything is shaped the way you expect. Select all, right click, select Debug->readback collision geometry. Then press play. You should see what the physics engine believes is the shape of everything. Are there any surprises?

@rayl said:

I am having problems with a truck I created, the tires seem to go half way through the floor, I tried adjusting the gravity values and the floor is set to static mesh, and I can't seem to find the problem. This only seems to affect the front tires (the ones that steer). I downloaded a couple of examples from the 3D warehouse and they don't seem to have this problem. I'm thinking it might have something to do with uneven weight distribution, but I'm not sure. By the way it's an 18 wheeler, with a trailer attached. Any help on this matter would be appreciated. Thanks.

Sounds like the tires are either the wrong shape or maybe they have flipped/scaled elements.
-Try setting the shape of the tires to "sphere" or "convexhull". If that doesn't work explode and regroup the tire.

If you post the model I can tell you exactly.

Chris

@whaat said:

I've been playing around with a new script idea lately. I am not aware of any existing script although I heard that RickW was working on a similar script at one point in time.

If I can perfect this script, I see a lot of potential merging the functionality with other Ruby scripts such as 'Free-form deformation balloon' (CPhillips) and 'Skin' (ArchitectBoy).

It seems like it would also be possible to use the same logic to subdivide and smooth geometry during the export process to a rendering app.

Whaat

Looks cool! And it would be a good compliment for a Free Form Deformation plugin.

Chris

@daniel s said:

Hi Chris!!!

Great Script!!! I wonder what happens in very simple models like a cube... Subdivides the faces? or not working in this cases?

Daniel S
PS. I attach a image.. the result of moving three control points on a cube will give a result like this?

It doesn't subdivide faces. That would be even slower.

The more faces you have the better the results.

I worked on it a bit tonight and got the speed up to the point it might be useable. Here is an example using a model I use as a test case.

It still needs a lot of work. Starting with the UI.

BentDeer.zip

I did some more work on this a month or so ago. Unfortunately the approach I took was really really slow. Minutes to process a model with a few thousand verts. I need to go back and try to optimize it at some point.

Chris

@kobus said:

Contextual rightclick would open a list of possible commands. Manageable settings would be nice, too. Is this a possible ruby ?

Look at UI.add_context_menu_handler. Or if you are doing a tool define the getMenu function in your tool class.

http://groups.google.com/group/SketchUp-Plugins-Dev/web/Ruby-UI.html
http://groups.google.com/group/SketchUp-Plugins-Dev/web/Ruby-Tool.html