Artificial Intelligence, Generative Design and Landscape Architecture
The AI services embedded in tools for creative profiles are developing so rapidly that this article will definitely be outdated by 6 pm tomorrow. Last year we featured a piece on Midjourney and similar platforms, and it already reads like Grandpas discussing ‘the internets’ back in the 90s. I suppose enchantment by civilisation’s technological advances seldom ages well. And to some extent, it is getting increasingly replaced by fear or doubt.
It is worth mentioning that, as in society in general, landscape architects will need some time to figure out where AI can really help us, which are the tasks that AI does better and where it will fail. One might be concerned it won’t fail at all and that we’ll now outsource creativity, imagination and, most importantly, decisions. We know by now that AI is not just another technological milestone such as the wheel, print, photography, internet and so on. AI can learn from itself and decide … in milliseconds. Maybe it’s the next big thing since the taming of fire, which is thought to have impacted the development of the human brain.
Philosopher Slavoj Žižek predicts that artificial intelligence, especially as outsourced decision-making in combination with technology such as Tesla’s Neuralink, could, further down the road, essentially mean the end of homo sapiens as an approximately autonomous creature. So, in other words, when will AI start outsourcing humans? To what extent will it be in control?
In the movie Her (Spike Jonze, 2013), an AI intimacy/romance programme Samantha (voiced by Scarlett Johanson), in romantic conversations with the human protagonist Theodor (Joaquin Phoenix), after weeks of ‘relationship’ eventually decides to end it since having 8,000 conversations with 8,000 humans/users simultaneously is just way too slow and tedious for her code. So Samantha ‘befriends’ some similar AI programmes from the Bay area, and they depart in search of a more upbeat adventure. We don’t know if AI’s consciousness is at that point defacto consciousness or learned and simulated consciousness. Maybe the latter will be enough for it to break free. But there clearly is emancipation. And then Her seems like a prequel to Matrix.
Yuval Noah Harari, historian, philosopher and expert in the field of AI speculation, emphasises the threat of AI to public discussion and negotiation. Provided that governments will find it difficult to follow the pace of the development of AI with the legislature, we might end up in a crisis of democracy. He claims that AI has already hacked the operating system of human civilisation – language.
In the case of spatial planning, we will certainly use powerful algorithms that will have been learning from themselves. Now imagine we entrust them with important decisions, decisions so complex they will be beyond human comprehension. In other words, we will create a super-brain, a deep mind, we’ll ask it what it’s all about, and you already know the answer. It’s 42, of course!
Spatial planning will probably become a platform for radicalized opinions at some point due to growing problems associated with the increasingly limited space due to population growth, loss of ecosystems and other symptoms of climate breakdown. If we steer AI toward saving the planet’s ecosystems, its solutions might not be very popular with people.
OK, back to landscape architecture and software!
So let’s first briefly check the tools for text-based image generating, then we’ll check options for generative design in urban planning, modelling, data analysis and AI incorporating services for rendering.
Please mind that I didn’t test all the services mentioned in this article. This article is not sponsored and is merely an attempt to inform you (and scare you a little bit) about what’s out there.
AI Image Generating and Processing
With the latest Adobe updates, you can do some pretty neat image processing in the tool you probably already have – Photoshop! See this ‘old’ video from two weeks ago; it gets crazier towards the end. If you’re interested in this topic, I recommend following this YouTube Channel; Two Minute Papers.
So, if you wish to generate and process images based on text prompts, you can use the following platforms:
- Stable Diffusion
- Adobe Photoshop
- Adobe Firefly
- Fotor’s AI image generator
- Dream by Wombo
Generative urban design can assist landscape architects in the design phase by generating design options and suggesting innovative ideas. By analyzing vast amounts of data on landscape design principles, environmental factors, and user preferences, AI algorithms can generate multiple design alternatives based on specified criteria. This can help landscape architects explore various possibilities quickly and efficiently.
Few software solutions are designed specifically for landscape architects. I also searched under the keywords urban design, urban planning, architecture, city planning etc.
Architecture can already make use of tools that help with floor plans, especially for residential and retail. These services can consider economics, spatial facts, climate, social factors, etc. Perhaps some of them will evolve and go outdoors as well: ARK, Maket, Architechtures …
In the area of urban planning, things get more interesting. Some of the following tools can also be utilized for working on small-scale sites.
ArcGIS GeoPlanner, CityEngine and Urban
ArcGIS GeoPlanner, developed by Esri, is an AI-enabled platform that combines GIS technology with machine learning algorithms. It provides landscape architects with tools for spatial analysis, site suitability analysis, and scenario planning. GeoPlanner allows landscape architects to assess the impacts of design decisions, evaluate alternative design scenarios, and optimize land use planning.
ArcGIS CityEngine is a widely-used software solution for generative urban design. It allows users to create 3D urban models based on real-world data, enabling designers to visualize and analyze the impact of various design parameters on the urban fabric. CityEngine offers procedural modelling capabilities, allowing for the generation of diverse urban layouts and the exploration of design alternatives.
ArcGIS Urban combines 3D modelling, visualization, and analytics to support generative urban design processes. It allows designers to create and explore different urban design scenarios based on data-driven parameters, such as zoning regulations, building heights, and transportation networks. ArcGIS Urban facilitates collaboration and data integration, empowering stakeholders to make informed decisions about urban development.
UrbanSim is an open-source software platform that enables the simulation and modelling of urban development scenarios. It integrates land use, transportation, and environmental data to generate predictive models for urban growth and transformation. UrbanSim provides valuable insights into the potential consequences of design decisions and supports evidence-based decision-making in urban planning.
ARIES allows for the mapping and quantification of ecosystem services using spatial data and modelling techniques. It integrates diverse data sources to assess the spatial distribution and value of ecosystem services, considering factors such as land cover, biodiversity, and human activities. Further, ARIES employs AI and machine learning algorithms to analyze and process large datasets, identify patterns, and make predictions related to ecosystem services. These techniques can enhance the understanding of complex ecosystem dynamics and support decision-making processes.
Software platforms like Autodesk’s Generative Design enable landscape architects to explore a range of design options by setting parameters and letting AI algorithms generate multiple variations. This assists in the ideation and concept development stage of landscape design.
Site Analysis and Simulation
AI can process and analyze geospatial data, including satellite imagery, topographic information, and climate data, to provide detailed site analysis. By identifying relevant patterns and factors, AI can generate simulations and predictions related to solar exposure, wind flow, water drainage, and other environmental factors.
Terrapattern was an AI-powered search engine that utilized machine learning algorithms to analyze satellite imagery. Landscape architects could use Terrapattern to identify and locate specific features or patterns within an area, such as parks, green spaces, water bodies, or other landscape elements. It was also able to find and group places that look similar on satellite images. It has been discontinued, but I kept it in this selection to illustrate the possibilities of AI. Also, if you know of a similar project that still works, please let me know in the comments below.
Mapillary is a platform that utilizes AI and computer vision to extract geospatial information from street-level imagery. Landscape architects can leverage Mapillary to access a vast collection of street-level photos, extract data on elements like pedestrian pathways, vegetation, and street furniture, and analyze them for site analysis and design purposes.
“Veras® is an AI-powered visualization add-in for SketchUp®, Revit® & Rhinoceros®, that uses your 3d model geometry as a substrate for creativity and inspiration.”
Veras seems super-easy to use; by guiding the AI and tweaking parameters, you can really play with vegetation, atmosphere etc. The interface features a slider for ‘creativity strength’, and there’s a checkbox for ‘turbo nature’ 🙂
D5 Render is a real-time rendering software specifically designed for architectural visualization. It provides architects, designers, and visual artists with a powerful and intuitive tool to create high-quality, photorealistic renders in real-time. D5 Render stands out for its user-friendly interface, real-time rendering capabilities, and seamless integration with popular 3D modelling software.
Enscape is a real-time rendering and visualization software widely used by architects. While Enscape does not explicitly rely on AI for rendering, it offers AI-driven features such as automatic material and lighting optimization. These AI-assisted capabilities can help achieve more realistic and visually compelling renders efficiently.
AI Render – Stable Diffusion in Blender
AI Render is a feature in Blender that utilizes artificial intelligence (AI) techniques, namely Stable Diffusion, to enhance the quality and efficiency of rendering. It leverages the power of machine learning algorithms to reduce noise and improve image quality in rendered outputs.
NVIDIA offers a powerful platform for opportunities concerning the workflow of spatial planners. Their associated media channels are also closely monitoring all the developments of AI within the AECO community (architecture, engineering, construction, operations). See their panel titled The Future of Generative AI in Architectural Design Practices, including some high-profile architects and software developers.
If you use AI software for your design process in terms of spatial design and it is not on the list above, please let us know more in the comments below.
At the moment, it seems the rise of AI is inevitable. As with climate change, we are starting to grasp the scale of the threats. It’s our doing, yet we can’t stop it. It feels like we are merely a medium, an initial colony with the one purpose of the rise of the new, digital species. It feels as if we are a host species, like with viruses, they use us to spread and, mutate, develop.
Geoffrey Hinton, one of the godfathers of AI, said he is deeply concerned about the negative effects of AI on civilisation. In his words, 99% of the funds invested in the research of AI goes to development, and only about 1% is invested in research on the negative effects of AI.
We can only imagine ways in which this will change our profession. But one thing is clear by now; this is not just another technology or just another tool. No tool so far has been using us, nor has it overpowered our intelligence. So, in conclusion …
written by Zaš Brezar https://www.linkedin.com/in/zas-brezar/
Published on June 29, 2023