Although I more or less agree with that definition my personal view of Generative Design is slightly different. For lack of a better example I'm going to go with DNA. In its raw state, DNA contains all of the instructions by which life can be created. It will dictate the growth of cells as well as the type/characteristics of cells. Theoretically you could clone someone from their DNA and get an exact "copy" of that person (remember Dolly?). However, the end result of that clone would not be an exact copy. Why? The growth of cells is not dictated solely by DNA, but also by its interactions with its surroundings. Since the clone invariably develops in conditions that were different from the "source being" there will always be a difference between the two.
To me, this is the characteristic of generative design. Its not just a product of the parameters that generate it, but of the conditions that surround and interact with it. Even within that definition, there's a very thin line that separates parametric design from generative design, since both rely on defining their operations through external values. That thin line, to me, is parametric design being generated by a singular reference to an outside piece of data and generative design being generated by a multitude of references an some interpretation of those references in an interconnected manner. I've moved towards defining GH as a Logical Modeler rather than parametric or generative. The main reason is that most parametric modelers rarely focus on the link between parameters as an artifact of their process. Because of this, their process, though parametrically based, is still embedded within model itself, which is certainly fine. GH does not operate in this manner. The creation of a GH definition is more of a representation of the logical steps used to create a given output, where as with parametric modelers, the representation of the inheritance of all parameters is not nearly as evident. Considering the control, amount of information that is immediately available by looking at a definition, and the interaction with those logical steps I would certainly argue that THAT is an extremely valuable aspect of the GH process. I could possibly argue that the GH definition itself is more valuable than the output it creates. As to the Generative capabilities of parametric modelers and GH, it all depends on how you structure the interactions with external data. A 1 to 1 relationship between a value and an output is more parametric in my book, where the interaction between a multitude of values is more of a generative approach. Just my 2 cents Best, Damien On Mar 4, 9:51 am, David Rutten <[email protected]> wrote: > Wikipedia: > > "The term "generative modelling" describes a paradigm change in shape > description, the generalization from objects to operations: A shape is > described by a sequence of processing steps, rather than just the end > result of applying these operations. Shape design becomes rule > design." > > Well, that's clear enough. I'm still having a hard time drawing a > sharp line between parametric and generative though. > > -- > David Rutten > [email protected] > Robert McNeel & Associates > > On Mar 4, 3:48 pm, David Rutten <[email protected]> wrote: > > > Hi Daniel, > > > a parametric modeler is one that's driven by parameters, as > > disappointing as it sounds. So, yes, on a basic level ALL modelers can > > be considered parameteric. > > However, that wouldn't be a useful term since it wouldn't exclude > > anything, so when we talk about parametric modelers we usually refer > > to indirect parameters. > > > For example, in Rhino you define a line segment by specifying the > > start and end point. This is typically not considered parametric. In > > Solidworks you can define a line segment as a tangent relationship to > > a circle and a certain length and angle. This would be considered > > parametric. One of the major outcomes of this different ideology is > > that parametric models tend to be history based. When you define new > > geometry using existing geometry, it doesn't take much to change the > > starting conditions and have the entire model adjust itself. So, I > > think most people will agree that in a non-parametric environment you > > directly design the model, whereas in a parametric environment you > > design the description of the model. In this light, Grasshopper is > > definitely parametric. > > > However, the term "parametric" has been applied to packages such as > > Solidworks and Catia and Pro-E for so long that it is now tightly > > associated with that particular style of interface. Perhaps this is > > why the term "Generative" has been introduced. To be honest, I don't > > exactly know what generative modeling is supposed to be. > > > -- > > David Rutten > > [email protected] > > Robert McNeel & Associates > > > On Mar 4, 11:37 am, daniel hilldrup <[email protected]> > > wrote: > > > > Hi All, > > > > I don't know if I am in the right place to start this debate, but I am > > > looking for answers > > > to my questions as I am about to undertake research on the field of > > > 'parametric' CAD design and am looking for some terminology > > > definitions. > > > > I am a MA product design student looking to adopt some of Grasshoppers > > > excellent > > > capabilities, and apart from the research would like to know how the > > > software is defined. > > > > OK my questions/statement: > > > > What is a parametric design? Is there a definitive definition? > > > At a fundamental level isn't all CAD parametric, using parameters that > > > state type, insertion point and direction (as with a line). Elsewhere > > > I see, dimension based modelers classified as parametric.... > > > > Is Grasshopper Parametric? I know that the 'blurb' states that it is a > > > Generative Modeler > > > using algorithms, so does it make it parametric? > > > > Would really appreciate any sources of information on the definitions > > > of generative/parametric modeling and the differences. > > > > Right, back to the hopper to see if I can make something from this > > > incredible software!
