Whoa, cool video!!! That caught me off guard too when the bridge and tunnel elements just sprang up like that...
As Michael said, the GH developer has basically coded a parallel version of Rhino commands and script functions. But Houdini was built as a nodal system from the ground up, where everything can be accessed by procedures. So GH functions do not go as deep as that yet. It may never reach that kind of depth... But this video is only showing procedural geometry and not using any special deformers or animation effects, so I think the entire project could be duplicated in GH. GH can also reference standard Rhino objects. So a curve could be used to define the road, then move the curve's points and GH will redraw all the geometry based on that curve. And then account for where the curve intersects the ground mesh, it's height relative to the ground, the length of spans for each elevated section, etc... It must be a tremendous amount of work using either GH or Houdini. I wonder really how much time was spent to set this up...
I've used GH to model a few parametric objects, and you almost have to forget what you know about modeling. Then start to think what is really happening within the program which allows you to model each element? How does an object get replicated along a curve and maintain orientation related to the curve? then you have to start dealing with planes, vectors, frames, domains, lists, etc... and the list seems never-ending. Even something simple like extruding an edge by a specified length requires a lot of steps to set up... It will give you a whole new respect for programmers and those GUI's which they create to do all the heavy lifting for you...
Generally speaking the time spent modeling something like that in video is pretty much the same as modeling it the traditional way(in houdini not grasshopper)
for some things you can actually do a lot faster than the traditional way.
It does require a different thought process and like Michael said it is practically "visual programming" which most people do not like. It takes getting used to.. I did not like that process as well when i started using hodini but have grown to like it a lot just wish it had better more nurbs surface operators...though it is actually possible to make those operators within hodini as well if you have the patience for it
GH is not gona get close to houdini for a long time. It is not as flexible and as integrated as Michael pointed out. For certain architectural concept modeling however it works great.
> It does require a different thought process and like Michael
> said it is practically "visual programming" which most people
> do not like.
Another possibility is that since a system like that is so similar to programming already that it could be good to actually emphasize that aspect even more so...
Some things are probably just easier and more compact overall when they are directly written as code instead of every single thing having to be a visual node with connector lines.
Things like loops, conditional statements, stuff like that can add up to a big soup of nodes and connectors while in text form them are just one line statements.
Folks, just a note if anyone still reads this thread: you can now use Grasshopper with Rhino and Archicad and have your work seamlessly transferred from one application to the other via the (beta) Connection tool. You can test it here: https://www.graphisoft.com/rhino-grasshopper
This is an intersting discussion. There is another node base visual programing called Dynamo. This is an open source and currently works
as stand alone ( you can export your model in STL format) or with Revit integration. Someone also has created the integration between Dynamo and Rhino called Rhynamo.
You must decimate your model with free MeshLab by example ! http://meshlab.sourceforge.net/ (525 000 Faces --> 25 0000 faces 5 seconds)
before use the _ImportOBJ by Max Smirnov!
Today I wrote a test-script which can create and manipulate objects by using html5 graphics interface.
Everything works great. Html5 canvas, window and mouse events e.t.c.
So it is technically possible to write the grasshopper clone for MoI.
Years ago I wrote a simple 2D tool which would generate "aesthetically pleasing" designs based upon a set of rules and a genetic algorithm. Using a simple neural net and GA, it would generate lots of 'base level' designs, then allow the user to select the ones they liked and the program would 'learn' preferences and try and generate more.
The rules were basically very simple. Things like golden mean, circles (few), 3 levels of details, symmetry and some other basic layout principles, like edge detail are all used in constructing a simple graphic representation.
I could see how you could do something very similar, but in 3D (2 1/2 D) and create interesting forms like this:
Frenchy Pilou
>> It's just a gadget or you will want really code this sort of nodes process ?
I would like to write such a script, but it will take a lot of time.
The work can be splitted in three parts:
1. Main core ( I can write it )
2. Design ( I need a help of someone skilled in html5/css )
3. Modules ( There will be hundreds of modules. I have not time to write them all, so I need a help of our community )
If I find someone who can help me I will start the work.
chippwalters
>>I could see how you could do something very similar, but in 3D (2 1/2 D) and create interesting forms like this
Yes, it's very interesting, maybe someday I'll write this script, but not now.
Hi Pilou,
I don't understand what you want to say.
You can find a link to demo in my previous post, so you can test an online version of this script.
However I've made about 50 changes in the code to make it work with MoI. Offline mode of course.
This is something! As a maya user, I can say for many operations, like procedural modeling etc, nodes are best way to buld things. Surprised it can be done in MoI :)
.
http://www.youtube.com/user/GeekatplayStudio/playlists
