Week 02 — Computer Aided Design¶
Global Class¶
The global lecture framed CAD not as a linear skill to acquire, but as an ecosystem of tools, each optimized for different stages of a design process.
A key message was to avoid committing too early to a single piece of software. Instead, we were encouraged to sketch, explore, and test ideas across multiple tools before settling on the ones that best support the project’s intent.
Key Ideas from the Global Class - There is no “best” CAD tool — only tools better suited to specific goals - Different CAD tools reflect different mental models of design - Combining tools across a workflow is valid and often necessary - Early stages benefit from speed and flexibility rather than precision - Precision, constraints, and manufacturability matter later in the process
CAD Tools and Design Intent¶
Several CAD tools were discussed, each associated with a distinct design mindset:
- FreeCAD — open-source, parametric, logic-driven
- SolidWorks — industrial, manufacturing-oriented
- Fusion 360 — integrated design, simulation, and manufacturing
- Blender — mesh-based, artistic, sculptural, animation-focused
- Rhino — NURBS-based, precise yet conceptual, common in architecture and design
Rather than ranking them, the lecture emphasized understanding what each tool is good at — and what it resists.
Parametric Thinking and Design History¶
One of the most important concepts introduced was design history.
In parametric CAD tools, a model is a sequence of decisions. Every operation leaves a trace, and those traces can be revisited, modified, or branched later.
Parametric design was described as:
- Geometry defined by parameters and relationships
- Changes propagating through the entire model
- Design intent being explicit and editable
Beyond Geometry: Simulation and Analysis¶
The lecture also touched on simulation and analysis tools:
- Structural and mechanical simulation
- Integration of simulation within CAD environments (e.g. Fusion)
- Use of online calculators for early validation
These tools were presented as ways to test assumptions early, rather than as final validation steps.
Documentation and Media Tools¶
Documentation was presented as part of the design workflow, not as a separate activity.
Tools mentioned included:
- Kdenlive for video documentation
- FFmpeg for video compression
- Batch image compression following Fab Academy guidelines
These tools will be used later as part of assignment documentation and final project reporting.
Key take away¶
One of the most messages from the lecturer was the encouragement to stay “weird.” The intent was to avoid over-constraining ideas too early and to treat CAD software as a way to expand creative possibilities, opening doors rather than closing them…
Local Classes — CAD in Practice¶
The local classes on Thursday and Friday focused on how these tools behave in practice, and what their differences mean when actually designing something.
Working through the same neutral reference — a LEGO block — across multiple software environments made these differences tangible.
The emphasis was not on mastering any single software, but on learning how to read tools: what they encourage, what they constrain, and how they shape thinking at different stages of a project.
Raster, Vector, and Fabrication¶
A key topic in the local class was the distinction between raster (pixel-based) and vector (mathematical) graphics, presented as two fundamentally different ways of representing information.
Raster images represent the closest digital version of reality, including photographs, scans, and camera images. They are rich in detail but dependent on resolution and scale, which limits their use in fabrication.
- Best suited for references, textures, and visual documentation
- Quality degrades when scaled
- Common tools: Photoshop, Krita, Photopea, GIMP
Vector graphics are mathematical descriptions of lines and shapes, designed for precision and scalability rather than visual realism.
- Infinitely scalable without loss of quality
- Directly readable by fabrication machines (laser cutters, CNC mills, plotters)
- Foundational for digital fabrication workflows
Vector Editors, 2D CAD, and Fabrication¶
Vector editors were introduced as an important link between design and fabrication. To illustrate this, we followed an Inkscape tutorial focused on boolean operations. Working through union and subtraction operations helped clarify how complex shapes can be constructed from simple geometry, and why boolean logic is central to fabrication workflows.
Several practical aspects of vector work were highlighted. Imported raster images should be prepared before vectorization by adjusting contrast and color. Once converted, shapes can be edited at the node level, allowing precise control over geometry and intersections. This level of control is what makes vector files suitable for use with digital fabrication machines.
Extensions and plugins were also mentioned as a way to adapt vector editors to specific fabrication processes, including laser cutting, embroidery, bending, and CAM preparation. This demonstrated how fabrication-related decisions can already be embedded at the 2D stage, prior to any 3D modelling.
2D CAD was then revisited with a focus on efficiency and clarity. AutoCAD was presented as a tool optimized for technical drawings, where accuracy, consistency, and standardization are prioritized over exploration.
This reinforced that 3D modelling is not always necessary. In many cases, especially for fabrication and documentation, a clear and well-prepared 2D drawing is sufficient.
Introductory 3D Modelling — Tinkercad¶
Tinkercad was introduced as an entry point into 3D modelling, particularly suited for beginners and educational contexts. Rather than exposing underlying geometry types or design histories, it focuses on the direct manipulation of simple primitives.
Using the LEGO block as a reference, Tinkercad made it easy to assemble basic forms quickly and develop an understanding of spatial relationships. This simplicity lowers the barrier to entry, but it also introduces clear limitations. Tinkercad does not support parametric relationships, design history, or advanced surface control, making models difficult to adapt as complexity increases.
A personal moment connected directly with this part of the class. My 10-year-old son attends an alternative school where a 3D printer is part of their learning environment, and that same evening he coincidentally learned how to use Tinkercad. We opened the program together, explored its tools, and created a simple chicken model as a playful exercise.
Seeing how naturally he engaged with the software highlighted how accessible these tools have become, and how quickly 3D modelling concepts can be absorbed when the interface removes unnecessary barriers.
Parametric Modelling — Fusion 360, Onshape¶
Parametric modelling was explored through tools such as Fusion 360 and Onshape, again using the LEGO block as a common reference. In this paradigm, geometry is constructed through sketches, constraints, dimensions, and ordered operations.
The model is not treated as a static shape, but as a structured sequence of decisions. Changes propagate through the entire design, allowing dimensions or relationships to be adjusted without breaking the model. This made parametric tools feel particularly robust for my final project when precision, repeatability, and mechanical logic are required.
At the same time, this approach demands more upfront thinking. Early decisions strongly influence what is possible later in the process. Parametric modelling therefore felt more appropriate for later stages of a project, when form has stabilized and attention shifts toward manufacturing, tolerances, and assemblies.
NURBS Modelling — Rhino¶
Rhino was presented as an example of NURBS-based modelling, positioned between free-form exploration and strict parametric control. Using the LEGO block, Rhino demonstrated how geometry can be constructed through curves, surfaces, and reference geometry without relying on a linear design history.
Strong snapping tools and a command-based workflow allow for fast iteration while maintaining dimensional accuracy. Unlike mesh modelling, surfaces remain mathematically smooth and editable; unlike parametric CAD, changes do not automatically cascade through a dependency tree.
This made Rhino feel well suited for conceptualization, exploration of forms, proportions, and spatial relationships while still operating within real dimensions. The number of commands and options involved made the learning curve immediately apparent.
Mesh Modelling — Blender¶
We moved on to Blender as a mesh-based modelling, using polygons as the fundamental building blocks of geometry. Working with the LEGO piece highlighted how well the tool supports sculpting and free-form manipulation.
Mesh modelling does not retain a sketch or parametric history, so changes require attention, and small mistakes can propagate quickly without an easy way to recover earlier intent.
At the same time, Blender’s strengths were very clear. It is open source, supported by a large community, and offers a rich ecosystem of add-ons. Geometry nodes can be shared and reused, allowing complex setups to be built without starting from scratch. Its strong orientation toward animation and visual workflows makes it particularly powerful for storytelling and presentation.
It was interesting to learn that Flow, one of our family’s favorite films last year, was produced entirely using Blender. This served as a concrete example of how distributed, open tools are being used in industries that were historically centralized and vertically integrated.
The level of skill and proficiency required to work fluently in Blender also stood out, along with the strong community that has formed around the software.
Final Project — First Attempt¶
After experimenting with different tools, I decided to move forward by developing an initial concept for my final project using Rhino, with the intention of later transitioning to Fusion. The idea was to use Rhino during the exploration and discovery phase, and once a direction emerged, move into Fusion to develop the mechanical aspects of the design.
I began drawing in Rhino and immediately became aware of my limited familiarity with the software. To move forward, I initially turned to ChatGPT for guidance. While it provided step-by-step instructions, the interaction quickly became procedural. Progress depended on following predefined sequences of commands, often interrupted by corrections, mismatched menus, or unclear steps.
This approach required me to already have a fairly clear mental image of the final geometry. As a result, there was little room for discovery. Design felt closer to executing instructions than to exploring possibilities.
First Attempts — Arcs, Revolves, and Ellipsoids¶
My first attempt was to model a dome. The initial workflow suggested creating arcs and revolving them, but this quickly led to confusion: incorrect menus, ambiguous directions, and commands that did not behave as expected.
After asking for alternative approaches, I tried modelling an ellipsoid instead. This led to my first hands-on use of boolean operations to cut and extract geometry. While this did produce results, progress was slow and required constant back-and-forth: asking for step-by-step instructions, sharing screenshots, and correcting misunderstandings.
Changing Strategy — Learning by Observation¶
At this point, I changed strategy. Instead of trying to construct geometry by following instructions alone, I started watching videos of experienced designers working in Rhino. Observing how they sketch, reshape, and iterate felt much more informative than trying to build everything from scratch.
Looking at examples such as geodesic domes and how they are constructed provided a clearer sense of how complex forms can emerge from relatively simple operations.
Exploring Structure — Square vs Geodesic¶
From there, I managed to draw a basic sphere positioned on top of a rectangular flat surface. I initially explored a square-based structure, focusing on alignment and proportions.
I then experimented with a triangulated, geodesic-like structure. This shift introduced a different way of thinking about surface subdivision and structure, even though the geometry was still very rough.
Much of the effort at this stage went into understanding how to “communicate” with the software: what inputs it expects, in which order commands need to be executed, and how different operations relate to each other. The learning curve was evident, and this phase felt more about orientation and familiarity than about producing a resolved design.
Moving to Fusion¶
The intention at this stage was to understand how a parametric, engineering-oriented environment approaches similar geometry.
I initially attempted to follow a similar approach to Rhino: drawing arcs and trying to revolve them into a dome-like structure. Once again, I experimented briefly with using an LLM as a guide, but encountered similar limitations and moved away from it immediately.
I struggled to find clear guidance online for creating a rectangular-based dome, which is closer to my actual project constraints than a simple circular dome. Drawing arcs on different planes and attempting to revolve them did not work as expected, largely due to misunderstandings around sketch planes, profiles, and valid axes.
At this point, it became clear that trying to force a specific outcome without understanding Fusion’s underlying logic was counterproductive.
Shifting Strategy — Learning the Tool First¶
I changed approach and focused on learning Fusion on its own terms. Instead of aiming for a specific final geometry, I followed tutorial videos to understand how sketches, constraints, profiles, and solid operations are meant to work together.
This shift made a noticeable difference. By recreating basic dome geometries, I started to internalize Fusion’s workflow and design logic.
Exploring Solid vs Hollow Forms¶
Once I was able to reliably create a solid dome, I experimented with shell and hollow operations to understand how Fusion handles wall thickness and internal volume—both critical considerations for a thermally active component in my final project.
This step felt more aligned with the mechanical and manufacturing intent of the project.
Weekly Reflection¶
The global class revealed CAD as a vast landscape of possibilities that amplifies the goals, intent, and mindset of the person using it. Tools are not interchangeable; they shape how problems are framed and how solutions emerge.
For my final project, this points clearly toward a layered approach:
- Fast, flexible tools for early concept development
- More constrained, parametric tools once decisions begin to harden
- Clear documentation to preserve design intent across transitions
What stood out during the week was how strongly CAD tools develop their own identities, communities, and advocates. Each ecosystem carries assumptions about how design should happen. As in many other domains, there are opposing camps and strong opinions. What interests me most is understanding how different perspectives can cross-pollinate in relationshsip with the tools used.
My experience using ChatGPT as a design aid surfaced an important limitation. Without the ability to interpret drawings directly— or without me yet having the skill to prompt adecuatelly— the interaction often led to dead ends. Time and energy were spent translating intent into instructions, rather than engaging in discovery. This highlighted the limits of instruction-based guidance in a visual, exploratory design environment.
The main takeaway from this week is my own limitation in tool fluency. For now, continuing with Rhino and Fusion feels like a solid starting point. Adding other tools later for specific purposes makes sense, but expecting mastery across many environments would be unrealistic.
File Compression¶
To keep the repository lightweight and compliant with Fab Academy documentation guidelines, all images used in this week were compressed before committing them to Git.
I used same online tool as last week as it seems to work efficiently.
For design files, I compressed the Rhino files into .zip folder and included the Fusion file in the repo.
Use of AI Tools¶
Prompts¶
Act as an experienced CAD designer to help me create a geodesic dome above the rectangle base flattop.
- Read the transcripts of these 2 videos carefully (https://www.youtube.com/watch?v=2I-Q5OHG974) and (https://www.youtube.com/watch?v=dU2C5t_UWIg&list=PLRgpbzA30vaPgk2xvlX66D68-a8j-654v) to:
- Plan the design and provide an overview of the different execution options.
- Use your criteria to select the simplest and most straightforward alternative.
- Provide detailed instructions and step-by-step guidance at my beginner level to draw the dome correctly.