Revit Design & AI Technology Exploration
Disclaimer: All Rendered Images for this Project were Generated Using PromeAI
REVIT PROFICIENCY
Design Intent
The main overall concept for my design is a three level multifamily housing project  that focuses on the importance of integration of outdoor space and greenspace within urban environments. Sustainable multifamily housing design is an area of architecture that I am very interested in, and I took the chance to explore this as I have not previously completed a project with this premise. Within this project, I wanted to create intentional natural areas for the residence, in this case a courtyard design, with purposeful circulation to encourage the users to experience and enjoy these integrated natural areas. In addition, I was also very intentional about including accessible bedrooms, bathrooms, corridors, and circulation throughout the complex, with every unit including at least one accessible bedroom and bathroom.  
In terms of specific Revit tools and aspects I wanted to explore, I was intrigued by the various window options and curtain walls available and wanted to specifically experiment with how these can be used in multistory projects. Also quite important to my concept was the ability to create groups for different apartment units in Revit. I wanted to play with how this crucial aspect of Revit could improve the workflow of a multifamily housing project that often includes the repetition of modular units and rooms, and thus could speed up the timeline of a project exponentially if creating each unit by hand is not necessary.
Groups & Units
The "Create Group" tool in Revit was particularly helpful in my project as it allowed me to create individual apartment units and then extrapolate these units through the building. I saved a lot of time by only having to make these units once for the project, and my workflow relied heavily on the use of this tool. The first thing I did after sketching out the overall form of my idea with the wall tool was making these groups, fully furnished and detailed, and then used the dimensions of each unit to refine my overall plan. A lot of my layout and design decisions were directly informed by the sizes of these units, and the "Create Group" tool made this process much faster. Once I had the first floor laid out, I then copied the room layout to the second and third floor, having successfully created three full floors of units in a fraction of the time it would have taken to model then manually.

1 Bedroom Unit Grouping

Tools Used: Wall, Trim/Extend, Align, Split, Create Group, Door,  Component, Families, Curtain Wall, Rotate, Move, Measure

2 Bedroom Unit Grouping

Tools Used: Wall, Trim/Extend, Align, Split, Create Group, Door, Detail Line, Component, Families, Curtain Wall, Rotate, Move, Measure


Four 1 Bedroom Units in Sequence : Created Using the "Create Group" Tool

Tools Used: Mirror, Trim/Extend, Move, Tags, Grid Lines


Three 2 Bedroom Units in Sequence : Created Using the "Create Group" Tool

Tools Used: Mirror, Trim/Extend, Move, Tags, Grid Lines

Raw Revit Drawing Outputs

First Floor Plan

Second Floor Plan

Third Floor Plan

Roof Plan

North Elevation

South Elevation

East Elevation

South Section

East Section

Axonometric View

Aerial View

Courtyard Approach View

Entry Approach View

Technical Insights
Overall, Revit is a very useful tool in creating architectural projects. The "Create Group" tool was fundamental to my ability to complete this project in a timely manner and I was in awe of how easy it was to extrapolate the rest of the plan from just two apartment units. While there was some initial challenges with having walls overlap when I copied these apartment units and placed them next to each other, I was able to fix this quite quickly by tweaking which walls were included in my groups for each unit. I also had some trouble with alignment of the units on each floor, but once I discovered this I used the Measuring tool and the Align tool to line up my walls and make sure both wings were exactly opposite each other. Learning Revit has been invaluable to the efficiency of my workflow and I thoroughly enjoyed getting to explore the tools and solving the challenges that come with it.
SOFTWARE EXPLORATION
PromeAI
PromeAI is an emerging generative AI website model that specializes in the ability to turn text, sketches, drawings, and 3D models into stylized or realistic renderings and images. The user has the ability to control a variety of different settings including levels of detail, image style, building style, and liberty that the AI can take in the generation. It is particularly useful in design fields such as architecture, interior design, furniture design, product design, and other design fields.
My aim with this project was to explore AI tools that allow for a more efficient ideation and brainstorming process in the initial stages of design. I chose to experiment with PromeAI primarily because of it's targeted focus on rendering  and architecture-specific outputs, and I hypothesized that such a model would produce the most high quality and accurate images from sketches and 3D models of buildings. 
Key Features
PromeAI hosts a plethora of tools on their platform, with the two I utilized being Sketch Rending and the AI Image Generator.
Both of these tools have the option for inputting a text prompt for the AI to generate from, with the Sketch Rendering tool also having the unique ability to render from an image such as a sketch, drawing, or view of a 3D model. For both tools, you can also input a "reference image" in which the AI will attempt to mimic the rendering style of the image to help the user more accurately achieve their desired output. 
There are three main features in both of these generation tools that allows the user to more carefully control their image outputs.

1. Style
This feature allows the user to pick an image style from hundreds of options, including realistic, stylized, and illustrative references. This helps the software to anticipate the style of image they are generating
2. Scene
A more specific stylistic option, this can help the AI anticipate the particular genre of image you desire, particularly for different design disciplines such as architecture.
3. Mode
This tells PromeAI how much liberty they can take in their generation vs. how much they should adhere to the guidelines of the input sketch/image. For my purposes, I found that the "Outline" preset worked the best as it kept the general shape of my design while still allowing the software to take creative liberties. Changing this setting can yield wildly different output results. 
Initial PromeAI Generation
Prompt & Settings Refinement
After my initial round of generation, I had a few that I found intriguing, but disliked some of the aspects such as the curved balconies and facades and a lack of balance between materials in some of the images. I also disliked how some of the concepts looked "overrun" with plants instead of seemingly intentionally placed and integrated. It was at this point that I began to consider ways to refine my results and more accurately generate my vision for the design concept.
One of the most influential aspects of a quality AI generated output is the prompt. Without a highly detailed and specific prompt that describes your desired outcome, the AI is merely guessing on what it is you would like to have visualized. Below is the refined prompt I curated after several rounds of generation. 
“Generate a three story contemporary, nonsymmetrical, sustainable apartment building made from wood, stone, and an abundance of glass curtain walls. There is an open courtyard in the middle of the complex that serves as a community space with seating areas and open green space. There are balconies lining the view into the courtyard and outward. Lots of greenery and plants on the roof and balconies. Set in an urban environment. The weather is clear and sunny. “
At this point I also began playing around with the settings, playing around with different detail levels, how strongly the AI kept to the prompt, and the inclusion of reference images which I used to curate the materiality I wanted represented.
My favorite images I used in the next sketching phase are pictured to the left.
"Failed" Sketch to Render Generation
At the start of the project, I was planning to sketch out my new ideas after my initial round of generation and then input them back into PromeAI’s Sketch Rendering tool. This tool can take sketches of buildings or designs and render realistic views of the project, and could help with visualization after initial rounds of brainstorming and sketching. 
However, I ran into some difficulties with getting decent results. It seems that very refined sketches are needed in order to produce good results, not the more “messy” sketches used more often in the idea process. It was difficult to get the AI to output realistic forms from the sketches as it wanted to follow the curvature of the drawings and it was very difficult to get a product that looked at all like a realistic concept. Particularly, the second set of images to the right are products of a sketch that closely resembles my final form. The results ended up being skewed, warped, and simply unrealistic as the software attempted to follow the sketch closely. 
With this set back, I opted to wait and try this tool again at the end of the project and test it’s rendering capabilities of my Revit model once my project was completed. 
Rendering Generation with PromeAI
For each of the renderings with PromeAI, I took the raw images from Revit and input them into PromeAI's sketch to rendering tool to achieve my results. With each of them, I wanted to try producing a more "realistic" style and a more stylized "watercolor" version that could showcase the AI's capabilities to generate images that look more analog in style. To achieve the more realistic styles, I chose the "realism" preset in the settings and input a reference image of a building similar to my target. For the watercolor, I simply tried out a few of PromeAI's watercolor rendering presets. Overall the results were fairly successful, with my favorite outcome being the aerial view in the watercolor style. 
North Elevation  Renderings
Courtyard Approach Renderings
Entry Approach Renderings
Aerial View Renderings

Potential Impacts & Integration
PromeAI has the potential to be a very useful tool in the design process, both in the ideation stage and the rendering phase. Currently, it is a very good tool for when you are brainstorming ideas at the very beginning of a project, as it can give quick life and visualization to fleeting ideas. Given enough time, the software could get to the point where rough sketches could be turned into full and detailed renderings of project iterations and give designers immense tools for visualization in the brainstorming process. With how it currently stands, I think it's most useful potential would be in the rendering stage. As mentioned, the sketch to rendering feature only really works with super detailed and crisp sketches, not those that would likely be made in the early ideation stage. It is more likely that this tool would be used after the 3D model is made and views taken from programs such as Revit could be input into PromeAI and could output renderings from this process as I did. 
Personally, I would only use PromeAI at the very beginning of the ideation phase, just to get some initial ideas and form inspiration. At that point, I wouldn't use it again until my design is completed where I may use it as the initial stage of some of my renderings. However, I would have to experiment with this to see how much time it may really save as there would likely have to be a large amount of post processing to the images. AI is currently not precise or accurate enough to generate fully realistic building renders, and there may be quite a bit of clean up to do on the renderings.  
REFLECTIVE ANALYSIS
Impacts of Emerging Technologies
New technologies in our field can have huge effects on how we as designers work and function, both positively and negatively. For example, Revit made hand drafting nearly obsolete when it came on the scene, but it also made making these drawings much easier and more efficient. In terms of AI, it could have detrimental effects on the integrity of the design process and could possibly replace many human workers in our field. However, it may also help designers speed up many parts of the design process such as ideation and renderings so they can focus on the quality of the actual design itself. You will always have trade-offs with new technology, and it is yet to be seen whether the positives outweigh the negatives when it comes to AI in the architectural field.
In my personal opinion, I think there's simply no choice other than to embrace AI in our field and foster it's more positive impacts. I have no doubt that the integration of AI models and software into our field is inevitable given its current exponential growth in popularity and its ever expanding integration into other fields. It's only a matter of time until it seeps its way into how we as architects craft drawings, diagrams, renderings, and other visuals for our projects. I do believe their are positive aspects of AI such as making the more tedious parts of our job easier, but even if we believe it may be detrimental to our positions as professionals, we have little choice other than to keep up to date on the technology so we can work alongside it once it becomes integrated. 
There are several reasons many are concerned about AI in architecture and design work in general. The first of these is the inherent emissions that are associated with AI and the training of these programs. The amount of computer power and energy required to train just a simple AI model is immense and far more than anyone may suspect when using an online tool. As designers I believe we have an inherent obligation to mitigate the effect our work has on the environment, and the use of AI certainly has a large one. Many are concerned that the implementation of AI into the architectural field could go directly against our responsibility as designers of the built environment to mitigate these environmental impacts as much as possible. There is also the question of if it is right to have machines designing spaces for humans and whether the lack of human input will be able to conceptualize a space that designs from a human perspective. This could greatly decrease the experience or even the quality of life that users of the spaces experience (considerations of culture, site context, and accessibility all apply). Any such integration of AI would have to be carefully monitored by humans in the design process and work with human ideas and context and not alone without any human-centered input. Additionally, AI so far has inherent biases based on how the algorithm is trained, not only in the styles of buildings and concepts they can generate, but also in the demographics of people they might depict in these spaces. Since AI is based on the material it is trained on and a large portion of human-made materials contain biases, these biases can show up in outputs from such models.  
On the contrary, there are others who view AI integration into design fields as opportunities for more complex designs to be completed in much shorter windows of time and for the more mundane aspects of the design process to be alleviated. AI could aid in the brainstorming and initial ideation process by providing new ideas that a team has not considered, and even help with iterating an existing idea. It could also complete cataloging and documentation tasks much faster than a human and could eventually be trained to complete architectural drawings to a certain style or specificity based on human input. There are a plethora of ways that AI could be used in architecture, and there are possibilities in store for having more time as designers to focus attention to the more human-centered aspects of design instead of documentation. 
Ultimately, we should accept the fact that AI will be a part of our work in the very near future. This may mean that the more tedious parts of our job such as drafting of plans, documentation, and construction documents no longer need as much human input as they did before, and thus the need for human man-power may greatly decline and the search for jobs in our field may become increasingly difficult. However, this might also mean that designers can simply shift focus to the schematic and ideation process of the actual design itself and make a more thought out final product as well as being able to get projects out quicker. I can see the future of the architecture industry including AI software and tools built into programs such as Revit and that the norm will become designers using these as part of their normal workflow, including myself. Only time will tell how AI pans out in the architectural field, but I am certain it will be integrated, it's just a question of how much and if the outcome is beneficial or detrimental.

Back to Top