Experimenting Using ChatGPT 4o with canvas in Grasshopper

[alghifary_alshayeb]

Hello Rhino Community,

I am working on an experiment where I use ChatGPT to create a parametric design with almost zero knowledge of how to use Grasshopper.
The ultimate goal is to lower the barrier for using the capabilities of Grasshopper, making it easier for new users.

So far, I’ve managed to achieve the following:

• Generated the roof surface and wood strip curves using ChatGPT to guide me through Python scripting in Grasshopper.
• Created a manufacturing workflow that involves flattening the wood strips, labeling, and exporting for CNC cutting.

Here is a bit more about how ChatGPT helped:

Capabilities Used: ChatGPT provided Python code for Rhino.Geometry, assisted in visualizing the workflow, and helped me understand the use of different components.
Tricks Employed: The model suggested using flowcharts to better understand the workflow, which I actually picked up from reading the forums here. This visual representation made it much easier for me to build the sequence of nodes in Grasshopper. In the beginning, I would take an image from Google and ask ChatGPT to generate Python code that would help me recreate a similar result to the image provided.
Challenges Overcome: With ChatGPT’s assistance, I managed to navigate through creating parametric curves, flattening surfaces, and preparing the .DXF files for CNC.

My workflow looks like this:

1. Generate roof surface and wood strips in Grasshopper.
2. Flatten the wood strips using the Unroll component.
3. Label each strip for easy identification during assembly.
4. Export the flattened strips as .DXF files for CNC or laser cutting.

My question is: Are there any recommended approaches or specific components/plugins I should use to make this workflow more efficient, especially when it comes to using Python scripting as a starting point for model generation.

Also, has anyone tried something similar using ChatGPT-4 with canvas mode? I’ve noticed that when using other models, it’s more likely to encounter errors, but with canvas, it’s much easier to debug the code.

Thanks in advance for your help!

Best, Alghifari

[Jos_Singer (jos)]

Instead of writing the code in the Python component, you can instead write the code in an AI powered IDE like Cursor or VSCode w Github Copilot. That way you will have AI modifying the code directly. Then you can enable auto save on the file and get the Python component you have there to run the saved file, you might be able to improve the speed with which you can iterate.