AEC Generative Design and Dynamo Product Briefing

LILLI SMITH: Hi, everyone. Welcome to the AEC Generative Design and Dynamo Product Briefing. My name is Lilli Smith. I'm a senior product manager in the computational design and automation group here at Autodesk in Boston. I am a licensed architect and have worked on many tools at Autodesk, including Revit, Formit, Dynamo, and generative design in Revit.

SOL AMOUR: And my name is Sol Amour. I'm also a product manager in the computational design and automation group at Autodesk. I hail from New Zealand, originally, but now call Boston home and have been here about four years. I have a background in architecture, construction, industrial design, and many other fields. And I like to describe myself as a curious human being. I have been working on Dynamo for about these four years. But I've been deeply immersed in this world for much, much longer.

LILLI SMITH: So we want to give you an idea about how we're envisioning the future. In doing so, we may make statements about the future that may or may not come to fruition. So please don't make any purchasing decisions based on any of these statements. All right, let's get started. So these are the learning objectives for today.

I'm going to start with a little bit about why we think computational and generative design workflows are important. Then, we'll talk about what is new in the products-- hot off the presses. Next, we'll feature three users who are using these products to improve their practices. And finally, we'll talk about our future roadmaps and the principles that are driving these future endeavors. So first, why is computational and generative design important?

The AEC industry owns much of the responsibility for building out the commercial and residential spaces for our rapidly expanding global population and changing climate. The future is a design challenge. Workflow automation and generative design can revolutionize the way we design by speeding up our work, making it more efficient, and using goals and measurable outcomes to help guide us and get all of this work done.

Architectural and engineering services have evolved from drawing by hand on paper and delivering it to others to build buildings. They've evolved to building information modeling and more efficient ways to document and deliver building instructions to the field. But with the serious problems that we're facing, we're going to have to figure out how computing power can help us even more. We want to invest in ways not just to record design decisions, but also project goals, constraints and success metrics.

Now we could throw more people on projects to do more work, do more work faster, and come up with more design ideas. But we could also use automated computing power to help us. What we really want to do next is to pair human intelligence with machine intelligence. Key to all these processes is being able to combine and codify the kinds of knowledge needed to solve building problems-- so that we don't have to spend as long on tedious tasks and so that we can use data-backed design decisions to work together to create a better built environment.

SOL AMOUR: Lilli and I work in the computational design and automation group at Autodesk, where our mission is to provide simple and capable tools for encoding AEC goals and constraints to assist design and analysis with automation. We developed Dynamo, the Dynamo players, and generative design in Revit. So this is the way that we see the world. The sky is the limit of what you can do by just writing some code.

But that is like saying you can just create the Mona Lisa just by making a painting. Writing code-- good code-- is a discipline and an art. And it takes time and effort to learn. It doesn't mean everyone here can't do it. But it does mean time spent away from designing, engineering, planning, and all the other things that as architects, engineers, and contractors you do as part of your domain. We have invested heavily in the open source, visual scripting tool Dynamo because it is a middle ground between direct modeling and a full coding environment. It allows architects and engineers to record their logic and speed up their workflows.

So this is Dynamo. There are nodes to read data in that come from many sources, such as Revit, Civil 3D, or Excel files, which connect to nodes that compute certain things-- in this case, capitalizing the view named text-- which then connect to nodes that write that data back out, i.e. this capitalized text.

So where can I find Dynamo today? Beyond our developer playground in the Dynamo Sandbox, you can find us amongst the awesome host applications that you see onscreen here that take the core consistent experience of Dynamo across many different flavors of design-- exploring rich data workflows and Revit, navigating large design spaces in Civil 3D. We're exploring powerful geometrical workflows in Formit.

So Dynamo itself is made up of two layers-- the background preview that shows the geometry and the graph navigation view that contains all of the nodes. Inside the graph view, you can build your graph using nodes that each have inputs and outputs that are connected together through wireless-- enabling data to flow sequentially through your graph and be computed and transformed along the way. You also get to inspect outputs on every node via the preview or a watch node to ensure that your graph is doing what you intended.

So with Dynamo, you can do many different things, from complex form creation, to extracting data, to coordinating elements, and much, much more. So these are the active Dynamo user groups all over the world-- from Atlanta, through Boston, Catalonia, Shanghai, Ireland, Auckland, and many more. And these are just the ones with some of the coolest logos.

So Autodesk's mission is to bring new technologies to the market so that more people can use them and participate in the design process. In Revit 2021, we introduced generative design in Revit to make design automation workflows more accessible to more people. So here on the screen, you see a designer is performing a masking study to explore the allocation of retail and office space distribution while minimizing cost and maximizing rentable area.

You'll see it as integrated into Revit and the Revit environment. And the designer performing the study does not necessarily need to write the code themselves or even really look at the underlying Dynamo graph. Integrating these tools into the large river ecosystem is our first step towards making generative design processes more mainstream so that people can have the supercharged ability to explore the best possible solutions to their design space.

So let's take a look at what's new. Dynamo has undergone a visual refresh that reimagines the paradigms of what Dynamo is, while still honoring its past. With a focus on clarity, consistency in the removal of hidden behaviors, Dynamo's UI has been updated to give you information at your fingertips-- ensuring that you are well equipped in your Dynamo graph creation journey. And you look. And you feel many quality of life improvements-- but still the Dynamo you all know and love under the hood.

So part of this refresh was the introduction of a whole swathe of new graph authoring tools designed to make your lives easier when building, curating, and documenting your dynamic graphs. We have a new set of wire actions that allow you to hover over wires and see the data flowing through that wire to be at the ability to choose to pin that wire or insert a watch node if the data stream is complex. You can also right click on a wire to have the ability to break connections or remove wires, slit connected nodes to those wires, or even hide the wire itself.

We've also reimagined how groups work in Dynamo, allowing you to more richly describe your groups with both a title and description field, nest as many child groups inside a single parent group as you like, and collapse groups that you do not need to be interacted with, to remove visual clutter and gently lead others through your graph logic. Take note that collapsed groups behave as large nodes, automatically generating inputs and outputs from the nodes contained inside. Epic stuff.

We've also heard that a lot of people complain that they get stuck when trying to figure out what node can feed into what other node. Given that Dynamo has over 700 different nodes out of the box, and other hosts-- such as Revit-- add another 700 or so, learning about them all can be a daunting task. So Node Auto-complete addresses these concerns by providing an applicable list of choices based on object type.

You can think of this as a curated library of nodes that will work with the node you are triggering it from. It still requires design intent, but discoverability is much less daunting. We also have some in-depth documentation that you can read inside of Dynamo around what Node Auto-complete is and how you can use it.

We also continue to improve the generic design tools in Revit, with improvements to the Create Study interface in Revit 2023. We have made these tools more accessible with keyboard navigation and have introduced new input types, such as number inputs. You can now also put detailed information and links in a graph to provide more explanation. And to help users, we have also renamed and better explained the option generation methods themselves.

We have also made improvements to the Explore Outcomes interface. You can now see more than 10 outcomes at once. Awesome. And you can also export these outcomes, including thumbnails and rejected options, so that you can understand how the optimization algorithm is working and so that you can then use these assets in other contexts.

And lastly, we have made some tweaks to how generative design works in Dynamo to author generative studies. You can now work in Dynamo without having to close the Dynamo player, allowing for real time changes that apply to both. You can set graph types to generate design and get helpful information about graph setup. And you can also show an edit graph properties to help describe what the study entails.

In Revit 2023, we have also updated the Dynamo Player to have the same user interface as the generative design tools, including a lot of improvements we have just showed for generative design. You can still access Dynamo Player through the same button on the Manage tab. But the experience has been updated to now have a consistent UI with generative design. Now awesome new features in the UI have come to both of these tools at the same time.

Dynamo Player provides simple graphs for Revit, Civil 3D, and our other hosts with Player that you can use to customize your project. You can manage and save your folder locations, enabling you to create your graph library, add in descriptive information to the graphs-- such as thumbnails, descriptions, and authors-- and add clickable links out to external information, empowering you to create video and textual content to enrich your user experience.

Features from inside Dynamo-- such as the graph properties, extension, containing the metadata or pinnacle notes to nodes-- now transfer to Dynamo Player to enrich that experience even further. Back to you, Lilli, to showcase some of the Dynamo Ecosystem tools in action.

LILLI SMITH: Thanks, Sol. That's really exciting stuff. So now, I want to transition to show you what architects and engineers are doing with our tools and how they're using them to innovate and improve their practices. I am thrilled to welcome three special guests from all around the world.

We have Karam Baki, who's a senior principal at AECedx, out of Turkey and Jordan. Dana De Filippi, who is a computational designer at the Smith group in Washington, DC. And Edgar Pestana, who's a BIM engineer at Basler and Hofmann in Lucerne, Switzerland. These innovators are going to show us examples of how they use computational and generative design workflows in their practices.

First up is Karam Baki of AECedx. Karam will show us a Dynamo workflow where he uses geometrical tools in Dynamo to automate improvements in his Revit models for better drawings and more accurate material takeoffs.

KARAM BAKI: Hi. This is Karam from AECedx. I'm very thrilled to present this AU user case. We usually work on solving challenging problems with some high profile projects, such as [INAUDIBLE] cell tower with IBECE group in Dubai. And typically speaking, each project contains its own set of problems that requires its own set of solutions. But when a solution works on every project, then this solution is worth talking about.

And today, I'm going to present one of these solutions. So as you can see, we have a sample structural file that contains structural framing elements and structural columns. We also have our sample architectural file that contains walls, and floors, and some architectural elements. Of course, the structural model is also linked in this architectural model. Now right away, you will see an obvious problem between the structural model and the architectural walls.

The relationship between those elements isn't consistent. And that's an expected behavior in Revit links. There are workarounds that some people actually do, such as copy and monitor the columns. But it's really hard to track them. And it doesn't work on every category, such as structural framing. For some companies, this is a normal behavior and they accept it as it is. However, this will cause some serious inaccuracy in the material take off and will also cause some inconsistencies in the drawings and deliverables, such as this section.

As you can see in here, I cannot obtain the material of the structural frame with the material tag. It's simply unrecognizable. And in such circumstances, people tend to use dump tags that are text based. And for the structural framing itself, they either use fill patterns or detail item elements. So in all cases, this is not proper BIM. So how do we solve these issues?

Simply go to Dynamo Player. Assuming you have Synthesized Toolkit package installed, you have to add the extra folder of the package in Dynamo Player. Now, just search for link, and you'll find Links Copy Elements from links by category. Click on it, and you will see some inputs. The first input is Link Names to Copy from.

The link name I want to copy the elements from is called Struc.rvt, so I'm going to just write down here Struc. The category to copy, I'm going to copy the structural columns, then the structural framing. The next option is Generate Model Group and Generate Void Family. I do not actually want to generate a model group from those elements. So I'm going to turn this off.

But I'm going to leave Generate Void Family on because the concept of the solution is that I'm going to create a void family based on these elements, and I'm going to place them in this project. And I'm going to hit Run. And we'll give it a couple of seconds. Of course, that might take a little bit more time in big projects, but it's generally fast because it does not recreate the geometry in Dynamo. It simply takes the existing Revit geometry.

So I have a void family, as you can see. These are the columns as a void family. And next, I'm going to copy the structural framing. Here it is, and I'm going to hit Run. The next step is actually to cut our host elements from those voids. So we go back and simply search for another script. Search for void, and you're going to find Void Auto Cut Voids with Elements Type. Click on it, then Select Void Elements. I'm going to select those.

And I'm going to cut from element type the walls. And I'm going to hit Run, of course. When voids do perform a cutting operation, they disappear, which is a very good behavior. And next, I'm going to cut them again from the floor element type, and Run. And now we're finished. As you can see right away, there's an improvement in the relationship between the structural elements and the hosting walls. Here they are interacting as expected.

And if we go to the section, we're going to see that it recognized the material correctly. The element also behaves very good. And in the material take off, we're going to see a decreased value of the concrete material. So Dynamo did not only help us to automate a custom solution, but rather it helped us to access some tools that are usually not accessible from the UI, such as creating these free form void elements on the fly. I hope you enjoyed this Dynamo user case. And see you in other sessions.

LILLI SMITH: Next up, we have Dana De Filippi of the SmithGroup. Dana will show us a workflow she developed to ensure incorporation of her company standards in Revit workset creation using Dynamo Player to allow more people to use the tool to eliminate manual transcription and to save time.

DANA DE FILIPPI: Hello, my name is Dana De Filippi. I'm a computational leader with SmithGroup. YouTube channel, DanamoBIM. Today, the problem that I'm going to be discussing is the incorporation of SmithGroup standard worksets across Revit models by discipline.

There's quite a few considerations when facing this problem at hand, such as the naming conventions that we want to match, the default visibility settings that we have, so those kind of default standards that we have that align with some of the standard workflows that we already have in practice. And by doing this, building efficiencies that will save time in the creation process and of course eliminate errors through the manual transcription process.

In our template we actually have a workset standards view for our template instructions that goes through each workset by discipline. As SmithGroup is multidisciplinary, each discipline has a list of worksets and its related visibility settings to incorporate in the early stages of model creation, essentially right as the model is collaborated. You can see, there's optional worksets as well for some of our disciplines.

The solution that we came up with is a Dynamo Player script, essentially allowing you to specify whether you want the optional worksets or just the default worksets-- by default, you will get all worksets-- and which discipline you align with. By running the script, it will automatically create all of the worksets as well as set the visibility. Behind the scenes, we have a Dynamo script that incorporates some Python.

Very easy to get this shared across the office, as it does not incorporate any custom packages that you need to ensure all users have. Also it'd be very easy to share from the client or owner perspective or across consultant teams. Showing the process of a user running the Dynamo player script, typically you create the collaboration very, very early in the process.

And on that collaboration function, you're going to get two worksets, workset one and shared levels and grids. When running this script, it will automatically name all of the worksets for you by discipline as well as set the necessary visibility settings. Dynamo script, upon completion, will also display the not viewed worksets, the worksets that are not visible by default, as well as the worksets that were created, the total list, that is.

And in opening up our workset dialog box, you see that list in addition to those visibility settings. If you are interested in learning more on workset creation, please check out the DanamoBIM episode "Python Tools for Revit, Episode 1." It's actually how we all got started. Ehsan and I created a video on how to not only create worksets, but how to update visibility settings.

Other workflows as well could be updated in this way, from finding and replacing view names, all the way to updating your title block key plans. Make sure you subscribe to DanamoBIM and check out all the videos that we have on there. This particular script for workset creation, we have found that it is varied across disciplines, which is really great. You get a pretty equal distribution related to how the relative size of each of those disciplines in our firm.

Not everyone creates Revit models, therefore not everyone is going to need to collaborate the Revit model up and create worksets. So we do see not all Revit users using this script, but it does align with those who create Revit models. Thank you so much.

LILLI SMITH: Thank you so much, Dana. And thank you for coming and inspiring us with this one use case and also all these cases that you've posted on your YouTube channel. Really great. Make sure you go check that out. So lastly, we welcome Edgar Pestana of Basler & Hofmann.

Edgar will share with us an MEP design optimization workflow, which he calls spacing. So this workflow uses Generative Design in Revit to minimize technical space while ensuring feasibility of mechanical system installation.

EDGAR PESTANA: Hello, everyone. I'm Edgar Pestana from Basler & Hofmann. We are a planning engineering and consulting firm in Switzerland. And I would like to present you one of our digital products. We call it spacing. So what is spacing? Spacing is a Generative Design product that digitalized the norm-based planning service for MEP design optimization in early phase.

What it enables-- it enables architects, investors, and other stakeholders to make self-sufficient analysis of technical space requirements. As inputs, we need to propose architectural layouts in IFC format. It uses the Swiss standards to perform the calculations. It drives the MEP system types and the technical cell installations position.

The primary goal is to find a strong balance between minimizing technical space used and maximizing feasibility of technical installations. We add value to the market by saving 60% of time compared to the conventional planning process, enriching the design options by providing multiple variant solutions, and improving design communications between stakeholders by simplifying the decision making process with concrete information, which is well representations.

Now I'll present you a video where we briefly explain in more details our ideas and shows how spacing works. First, the spacing version. As you know, many specialized planning services are based on norms and standards. Spacing is a Generative Design tool, which automates the repetitive norm-based planning and improves the early phase design. We start prototyping with MEP as part of the shipment and we are in process of considering others.

Now, our inputs that we need. Fundamentally, we need the architectural layout in IFC format with apartment group as IFCZone, and the MEP shafts and technical rooms defined as IFCSpace. Our results, spacing delivers a report with quantitative information of each optimized solution and the respective IFC model. The design course-- so how is the Generative Design applied and what variables are optimized?

In this paragraph, you can see the five Generative Design goals that we defined. All are related to a maximized use for space and the technical feasibility. An example of how the algorithm can propose a better solution in terms of installation coordination by changing the ventilation technical room from one side to another of the building. So what is behind? The software tools used are Revit, Dynamo, Generative Design-- all from Autodesk.

Now a short demo of how the background works. We have here prepared a model. And we open it, the Revit, Generative Design study. We input and modify the parameters, fitness functions, and settings of the study according to our use case and let the study run. Normally, a study takes about 10 to 12 hours to provide us with the results.

We can then interface with the multivariable graph at the front to narrow in one feasible solution based on our variable choice. The respective elements are then create in Revit according to our defined template. The outputs, IFC models then differentiate to technical volumes and other useful volumes to compare with the initial proposal.

In addition, concrete suggestions for large equipment and horizontal and vertical networks are exported so the architects can use it as a base for further planning. Spacing helps you decide and design better. Thank you very much. I really appreciate the time and the opportunity. For any questions, please contact us.

LILLI SMITH: Thank you so much, Edgar, for sharing that with us. It was a very inspiring use case. So in our last section, we want to look at what's next. So I'm going to hand it back over to Sol to talk about some of our focus areas and what's coming up in these tools.

SOL AMOUR: Thanks, Lily. So the computational design and automation groups roadmap focuses on three key areas-- accessibility, where we want to make Dynamo easier for new users to learn and easier for more people to use automation routines.

We are also looking at sharing, where we want to make it easier for you to share your Dynamo graphs, both inside and outside your firm, and performance, where we want to make Dynamo faster and much less prone to missing dependency problems. So let's talk about improving accessibility. We are improving the extended note help, which you may have previously known as the Dynamo dictionary.

And Dynamo, by making it interactive, you'll be able to now more easily read salient information on any node, explore where this node lives inside of the library to orient yourself, review and learn from the sample graphs by zooming in and panning around and even dragging it into your graph, with the nodes we created as a group for you to use in situ. You can also interrogate the inputs and the outputs of the node to better understand how it's built and what it does.

We've also heard that large graphs are hard to navigate and especially difficult to find problems in, or potential problems in them. So if you've spent calories on allowing-- we've spent calories on allowing you to swiftly understand graph node states through zoom states, which show you at a glance what states these nodes live in inside of your graph, a run footer bar that collects quick counters of node states, and then the graph node manager, a more in-depth one-stop shop that empowers you to search any node that lives in your canvas, understand all of their types, their states, issues, and outputs even if they contain things like null values or empty lists as well as even more pressing errors and warnings.

We're also going to allow you to read all of this warning data at a macro level and then export all of this information out to an Excel or JSON file format. So the node autocomplete feature that you saw previously in the what's out right now section is based around object types, matching lines to lines, strings to strings, and so on. This is a great step in the right direction, but we can do even more.

So we are working on empowering node autocomplete with machine learning, which will not only return a set of results, but will rank them based on the most likely node that you will want to place and save your latest used node as a form of favoriting. This means more efficiency, especially around the more generic nodes that can contain hundreds of possibilities. So let's explore a little bit about data sharing.

We are bringing trusted locations to the entire Dynamo ecosystem, ensuring that you control where your graphs are loaded from and whether or not you trust these locations. This gives Dynamo, Dynamo Player, and Generative Design and Revit more security and you more control over how they interact with your graph. Now that we're combining the backend systems, these updates can swiftly propagate across all flavors of Dynamo.

To enhance sharing even further, we are introducing a Dynamo splash screen that not only shows you that something is happening when you open up Dynamo, but also enables authentication to grant you access to the package manager in all Dynamo instances and the ability to input settings before loading Dynamo. That empowers you and your firm to have a common baseline to work from. That can be a part of your Revit custom installer with more hosts that come swiftly afterwards. Inside Dynamo, from the preferences panel, you'll also be able to import and export your Dynamo settings. And these settings are also going to be able to map to the Dynamo Player and Generative Design experiences also.

And we also want to make Dynamo graphs a first-class citizen inside of Autodesk Docs. This means that you'll be able to read the properties of their graphs, such as thumbnails, the graph author, or a description, explore the inputs and the outputs of a graph, and take advantage of innate Autodesk Docs, things such as versioning, removing the need to control this manually, easy sharing, allowing you to control who has access to it, the ability to lock files that are in use, and a whole other swathe of other metadata associated with usage and editing your Dynamo graphs.

So finally, let's explore a little bit around better performance. Soon to be released versions of Dynamo have had an extensive work done on performance, taking a test graph that completely ran out of memory before completing the graph, essentially failing, to a graph that ran in 5.4 seconds on the exact same graph, with a massive reduction in memory use.

Another test graph had performance gains across both the first run experience that resulted in execution being 3 and 1/2 times faster, and also the update experience resulting in a four times swifter execution of that graph, as well as a dramatic reduction of memory, removing over half of what was used in Dynamo 2.1.

We are also working on getting all of the awesome features you saw before, the graph node manager and workspace preferences extension, into the Dynamo Player, enabling you to better understand issues that arise and take action to resolve them right there in the player UI. This is things like understanding what warnings are propagating up from your nodes inside the graph, to what packages you may be missing and allowing you to install them right there and then.

So we are also diving deep into Dynamo's engine, looking to refactor away some of the custom engine layers in favor of more foundational micro software. This means that we can remove some immediate or intermediate translations and get directly to the execution of graphs, resulting in faster performance, less places for things to go wrong, and more streamlined code. Think of a more nimble, light, and punchy Dynamo execution.

And then finally, we are bringing in native polycurves to Dynamo, written to work cohesively with the geometry kernel backing Dynamo, ensuring a robust, performant, and cohesive experience when working with polycurves, a highly popular and impactful way to work with aggregated curves from models in the AEC space. Now back to you, Lilli, for the wrap up.

LILLI SMITH: Thank you so much, Sol. It's really exciting to see what's in store. So in conclusion, I want to leave you with one more thought. So think for a minute about how robotics and automation have really changed automotive production and the resulting improvements in the cars that we drive today. Think about the automotive manufacturers that are embracing new technology and really reaping success from their automation.

Now imagine leveraging more automated and digitized ways of designing and construction and how that might impact your business and help you address the challenges and opportunities that you are facing. How much more might you be able to do with a workforce that you have already? It doesn't necessarily have to be super complicated. This is kind of a silly automation that just sorts red and green tomatoes. I bet it's a lot faster than what I could do by hand.

How might you transform your business to help create the sustainable communities of the future? I want to thank our special guests for joining us today. I find great hope in the stories of innovation and automation that they have shared with us. And thank you all for watching. Keep innovating and keep in touch. We'd love to highlight your work at AU 2023. Thanks a lot.