The parametric design process is a method of coming up with an idea, figuring out how to turn that idea into a sequence of logical steps, and finally using a tool like Grasshopper to turn the logic into commands and data flows that create the desired model.
When architects first start learning parametric design, with Grasshopper or similar tools, they easily get frustrated because while they learn about individual components, tying it all together to create what they have in mind feels challenging. In many cases it’s because they are lacking a clear process.
In this post I’ll cover the design process and provide tips for developing the skill of thinking in algorithms.
The 5 Steps to a Parametric Model
- Step 1: Define design goals and parameters. The first step in the parametric design process is to define the design goals and parameters. This involves identifying what the final output should look like, and what specific requirements the design must meet. Be as specific as possible. A parametric model is highly logical, and a vague goal can’t be translated into a logic.
- Step 2: Figure out the main geometric operations behind the design. Start breaking down the final result into a series of geometric steps and identify the key operations that will result in the desired output. For example twisting, scaling, shrinking, extending etc. At this stage we want to know the key geometric operations we’ll need to run to get to our target outcome. Usually there are just a handful of key geometric operations.
- Step 3: Break down the logic into smaller geometric steps. Take the design logic and break it down into a series of smaller geometric and logical steps. Did we need to make smaller adjustments to the geometry in-between the key geometric operations? Did we need to select only certain elements or all of them? Do we need to orient the geometry a specific way? If so to what? You don’t need to solve the whole logic, but the more granular you get, the easier the next step is.
- Step 4: Turn the logic into components and managing data flow. Once the logic of the design is established, the next step is to open Grasshopper and start building the component sequences. Lay out the key operations you defined and start connecting them. This is where you’ll realize that you need to be even clearer, because you have to ‘feed’ all component inputs. You’ll find more micro-steps to solve. This is also where data structures and data flow become relevant.
- Step 5: Debug your script: Nobody gets it right on the first go. Along the way you are going to run into issues and get error messages. That’s normal and expected. Learn to understand what the error messages mean and how they can be solved. With time debugging will get easier and easier.
How to learn to think in Algorithms
There is one fundamental skill needed to become proficient in parametric design: the ability to think in algorithms. I don’t mean actual mathematical formulas, instead it’s about understanding the underlying logic of a design and breaking it down into a series of small, manageable steps.
Here are four ways that I’ve found help people new to parametric design to practice thinking in algorithms:
- Put it into words. As simple as this may sounds, one way to develop algorithm thinking is to describe the desired output in words. By articulating the image in your mind, your brain is forced to make sense of it which helps to break it down into a sequence of descriptive steps. Its also useful to describe the design to others and see if they understand it. If they have any questions, use them to refine and concretize your idea.
- Create a sketch of your idea. By sketching, more questions are going to pop up naturally.
- Reverse engineer. Think backwards from the final result and ask the question: what’s the last step that was done to create this result, and what does the model look like before that step? By repeating this question, you can reverse-engineer the design.
- Learn to think in relative terms. In Grasshopper, it’s important to describe things in a relative way, such as the second item in a list, or the surface facing up, the longest of a set of curves, or the object closest to a point. This ensures that the script will work for any similar input and even multiple inputs at once, making the design more flexible. So instead of thinking “pick this box” (which is clear to you but not to a machine), think “pick the largest box” or “the first box in a list”, etc. It will make your scripts much more flexible and re-usable.
