Module 8 Questions:
Part 1: Study one talk.
The video I watched was option B: The Future of BIM is not BIM, and it’s Coming Faster Than You Think - the Sequel. The issue the speaker discusses revolves around the idea that we continuously recreate the same things in programs over and over again. This occurs not only within the same program, but also when it is required to translate elements of a model or information into another program or set of documents. That aspect is deemed the model drop chasm. Additionally there is a data driven chasm, where we consume information in rows and columns in spreadsheets. The main solution to solving these problems is coming up with ways to reduce friction within projects.
Some solutions the speaker has worked with were utilizing dynamic dashboards to fight against the data driven chasm. With dynamic dashboards, information can be visualized easily using sundials or adjacency matrices. Another solution they’ve worked with is project refinery in dynamo as a way to optimize outcomes. They incorporated optimization based on goals and outcomes as a method of evolutionary problem solving. A method of reducing friction within the industry for transferring information over into shop drawings is to use generative modeling to recreate aspects of the model into piece drawings. Fabrication drawings can also be auto dimensioned and tagged.
They ask the question of how we can apply computational and industrial processes to construction. The challenge being that not all buildings are new and most are unique. Computational design can be applied to materials. For example, panel flattening can be done within a software to properly obtain material quantities. Robots are also being utilized in construction for brick laying or laying out rebar. Although, there are still workers on site. I was surprised to hear about the robots laying out rebar in Zurich. I had never thought of that application before. One thing the speakers mentioned that made me a bit nervous was fully autonomous construction vehicles. While this sounds extremely useful, I’d have to hear more about it to be totally clear on it because it sounds a bit concerning as I don’t trust a robot to know exactly what to do if an incident were to occur on site.
Part 2: Reflect on the quarter.
Over the past 8 weeks, I experienced a lot of friction when I was modeling complex images or structures. It took a large amount of time for large numbers of small panels to be created or images to load into the system. Oftentimes, I found that dynamo/Revit would crash as I was attempting to run my model. I then would have to tweak my model to be less precise in order for the tool to run. Small changes in the model often required rerunning large amounts of the workflow and reorganizing nodes and outputs. A lot of those small tweaks felt repetitive and like they could have been automated.
An AI version of the workflow would be able to suggest changes and identify errors within the nodes. I would definitely want that kind of AI tool because it would greatly reduce the amount of time I would spend on troubleshooting the technical issues in my code. Despite this, I’d want to still create the actual code myself so I could understand how the variables in my code impact the structure or aspects of structural performance I was aiming to create.
Part 3: Scout the frontier.
Tool 1: TestFit
TestFit was actually a tool discussed in the option B lecture. TestFit is a tool that can automatically generate layouts or floor plans for parking, sites, apartments, and buildings in general. It does so based on a series of inputted site constraints and project requirements. The tool is interesting because it automates testing out a bunch of different designs instead of manually adjusting geometry and rerunning calculations. It would have changed how I approached the building form parametric models in dynamo because I would be able to create alternatives rapidly and bring them into Dynamo instead of just selecting one and running with it.
Tool 2: Maket.ai
Maket.ai is an AI-powered platform that can create residential floor plans from text prompts, site constraints, and design requirements. The tool is interesting because it uses generative AI through text prompts. Instead of manually drafting plans, you can just describe what you want the model to create and it’ll spit out some design alternatives. While I don’t think I would have utilized it this quarter, it could greatly reduce the time spent creating initial concepts before refining them.
Tool 3: Veras by EvolveLab
Veras is a plugin tool that works inside of Revit, Rhino, or SketchUp. It can use AI to transform conceptual models into realistic architectural renderings. This is interesting because it demonstrates how AI can be integrated directly into BIM workflows. It could drastically reduce the effort to communicate design alternatives generated in parametric modeling. I don’t think I would have used it this quarter, but it would be more used for the outcomes of the parametric models.