Final Assignment

Reflection on Original Project

The title of my previous studio project was "Retreat For an Artist", and it is what it says. The brief included a client who was a painter or sculptor from the 20th century, for whom which we had to design a retreat from their usual surroundings. Depending on the user, this retreat could be a place where they have more space to do their work, or a place where they could get away from their work. My Design was to be made for Jackson Pollock.

The objective for my Design specifically, was to create a space wherein Jackson could detach from his depression, and begin a healing journey. He was at a point in his life where his drinking had gotten to a dangerous level, and so I designed a retreat with the maximal exposure to nature. The original Design was based around having many levels, within the same floor. This was a decision motivated by the painter's work. As his work consisted of many organic layers, so too did my original Sketches.

These sketches, however, proved to be impractical. Upon further scrutiny by myself, my peers and my tutors especially, the design began to take a much more sensible shape. Involving more deliberate use of curves and corners in certain parts of the retreat, to ensure that it had the best user experience, while still remaining as faithful as possible to my intent.

In the end, I feel as though I had achieved my goal of creating a satisfactory user experience, especially considering the complicated inspirations I had to work with. However, this user experience came at the cost of not spending enough time experimenting with the form of the building. I feel as though I could've made it much more interesting while still keeping the same user experience. Generally I felt as thought it could've been more intricate. I wanted to draw inspirations from radical brutalist architecture of the 1940s and 30s, as this was around the era that Jackson pollock began his painting journey, and especially since it's similar to what he does artistically.

Jackson himself is a radical painter who challenged the ideals of traditional western painting. I I wanted to pay homage to him through evoking some ideas of brutalism, mainly in using exposed materials and geometric shapes. Jackson pollock isn't a brutalist painter, but I feel as though brutalism aligns with his philosophy, and time period.

In retrospect, I feel as though these two ideas were pulling at each other. Brutalism leans in the direction of minimalism, and Jackson Pollock is anything but. By trying to marry the two styles I ended up with the perfect marriage of Geometric purity and Chaotic organicism, which is to say, nothing. The design felt a bit empty, with boringly placed windows and openings. That is why I wish to improve the design by leaning more heavily into a more complex design. I would like to adjust the design just slightly, as I feel any other changes would require me to begin from scratch.

AI Workflow & Exploration

My main goal is to improve the form of the building. Since form is a concern of the 3rd dimension, I am constrained to using AI as guidance through my approach, using the help of some image models to create an inspiration board.

My group's findings in Assignment 3 led us to Claude as the most well rounded tool, though, I will be pairing it with Gemini as a supplemental tool, for its superior image generation capabilities.

Initial Prompt Strategy

My initial Prompts were a foray into the different techniques to add details to the form of a building, generally speaking. These weren't specific to the styles I was drawing from or my own inspirations for my design, but they were more general guidelines I could apply to the designs I've already made.

Prompt: What are some techniques to add more detail and interest to a form of a building. specifically along the vertical axis, and without altering the overall shape of the rooms too much

Response:

Here are the main techniques architects use to add vertical interest to a building facade without changing the floor plan:

Layering the facade away from the structure — the most powerful move. The building's skin doesn't have to sit flush with the structure. By pulling the facade plane outward and creating a secondary layer (a screen, a fin system, a brise-soleil), you create depth and shdow without touching interior room geometry at all.

Vertical fins and reveals — attaching thin vertical elements (fins, mullions, pilasters, flutes) to the face of the building creates strong shadow lines that shift through the day. These can be structural, decorative, or both. The deeper the fin relative to the surface, the more dramatic the shadow.

Stepping and offsetting floor levels — small horizontal shifts between floors (projecting a balcony slab edge, recessing a spandrel panel) break the flatness of the elevation and add a strong horizontal datum at each level, which paradoxically emphasizes the vertical reading between them.

Material changes along the vertical — using heavier/rusticated materials at the base and lighter/more refined ones as you move up mimics classical tripartite composition (base, shaft, capital). This adds visual weight and rhythm without any geometric change.

Window proportions and reveals — deep window reveals cast strong shadows and make the wall feel thick and substantial. Clustering windows vertically (punching tall openings that span multiple structural bays) creates verticality even in a wide building.

Applied texture and pattern — brick coursing, precast panel patterns, screen-block arrangements, perforated metal — all of these work entirely on the surface plane and can be varied in density or orientation to create rhythm, gradients, or focal points.

Attached or semi-attached elements — things like loggia columns, engaged columns, structural cores expressed on the exterior, rainscreen systems with protruding clips — these are all ways to "dress" a building in three-dimensional elements without altering the plan.

Here's a diagram showing how these layers stack spatially:A few practical notes on how architects typically combine these:

The most bang-for-buck technique is usually fins paired with deep reveals — the fins cast moving shadows across the day and the reveals make every window look intentional rather than punched-in. Together they can transform a flat slab building into something that reads as richly textured.

Material change at the base is almost always worth doing — even just switching from smooth render to brick or stone at ground-floor height gives a building a sense of being grounded and completed, rather than just stopping arbitrarily at grade.

The techniques that involve the least structural commitment are applied texture and window proportioning — you can change a lot about how a building reads just by adjusting the coursing pattern of brick or the depth of a window reveal, without touching anything structural or spatial.

Prompt: Draw a single storey family home whose plan is a series of gentle curves — no hard corners, walls sweeping continuously from one face to the next. The curvature is the plan's given; everything on the facade works with it rather than against it.

Because curved walls resist the logic of flat panels and straight fins, all layering is achieved through techniques that bend naturally with the surface. Vertical timber or metal battens follow the curve of the wall, fixed at a consistent standoff distance so they read as a continuous skin that hovers just in front of the structure behind. The gap between batten and wall — 60 to 80mm — is enough to cast a thin shadow line at each fixing, giving the surface texture without interrupting the flow of the curve.

Windows are cut as simple vertical slots, deeply recessed into the curved wall — minimum 150mm — so the thickness of the wall is legible from outside. Where the curve is convex and faces the street, reveals are deeper still, pulling light inward and emphasising the solidity of the mass. Where the wall turns away, openings can be shallower, catching raking light differently at each hour.

At the base, the wall thickens visually into a continuous plinth — raked render or in-situ concrete — that follows the curve at ground level and separates the batten skin above from the earth below. No sharp transitions, no corners where the plinth meets the wall: everything flows.

The roof edge is a simple continuous soffit that traces the plan curve overhead, its underside a single pale material that reflects light down onto the batten layer beneath.

From the street the house should read as a slow, continuous arc — textured and shadowed, never flat — its surface shifting in character as the eye travels around the curve.

Response:

Refined Prompt Strategy

As I dive into more complex ideas, imbuing the prompt with the context of my floor plan, as well as the styles I was attempting to emulate and draw from, could yield better strategies.

Prompt: What are some techniques to add more detail to the form of the floor plans attached below.

Response:

This is a really interesting design — a chain of overlapping circular volumes, each topped with a clerestory or lantern, set in what looks like a mangrove landscape. Here's how the facade strategies would apply specifically to this plan:

Battens following the curve work perfectly here because each circular volume has a continuously changing surface angle. Vertical battens fixed at a consistent standoff would catch light differently at every point around the circumference — the ones facing the sun read as bright lines, those wrapping away fall into shadow, making the curvature itself legible without any additional ornament.

Window reveals become more dramatic on convex walls. Because the wall curves away from you, a deeply recessed window on a convex face creates a shadow that shifts noticeably through the day as the sun moves. The large glazed panels visible in the elevations would benefit from a projecting hood or thick surround — even 100–150mm — to emphasise the depth of the wall at each opening.

The clerestory lanterns are already doing vertical work — each drum sitting above the main room volume reads as a distinct layer. The transition between the main wall and the lantern base is the most important detail: a strong shadow line there (a projecting ring beam, a recessed band, or a change of material) would make each volume read as tripartite — base, body, crown — even at single storey height.

The junctions between circles are where the most interesting shadow play happens. Where two volumes intersect, the wall folds inward and creates a natural recess. Emphasising this — by setting windows into the concave fold rather than on the convex face, or by keeping the batten layer continuous across the junction — would tie the chain of volumes together while celebrating the geometry of the overlap.

A consistent plinth running beneath all the volumes would unify what is otherwise a very episodic composition. Because the plan reads as several distinct bubbles, a low continuous base in a heavier material — board-formed concrete or raked render — would anchor everything to the ground as a single object rather than a collection of separate rooms.

The main challenge with this plan is that the circular volumes naturally draw the eye around and downward. The strategies above all work to counteract that — pulling attention upward toward the lanterns and creating a vertical reading within each drum, so the building feels grounded but also tall for its scale.

Prompt: alter the design of this building using the strategies discussed below (referring to images and prompt from before)

Response:

Conclusion

Relies almost exclusively on vertical perforations and variations in the facade along the horizontal axes. These designs do achieve the goal, this style would be a very simple way to add some variation to the facade. Using these fundamental guidelines, I will model a simplified mock-up of the form from my final presentation, compared with a mock-up of a new form based on the techniques discussed.

Implementation

Before changes

After Changes

Revised Design Proposal

Via my own judgement, I decided where to apply these ideas and how. Especially through the criteria of the Keywords I had to keep prominent through the design process (Bold, Layered, Controlled Chaos, Organic, Large Scale).

Overall, I feel the revised form has created much more interest, pushing it more comfortably into the Keywords and especially keeping more with the style of the painter. It keeps it's organic forms while still adding complexity and layers to the building. While I feel that AI was useful somewhat in expediting the process of adding this complexity, I feel as though I just needed more time to refine my design eye and, especially, some time away from the design to be able to see its problems more clearly. New solutions help to improve efficiency but they will not work without old fashioned techniques. Occasionally, the best way to improve your speed and proficiency in design, is to not design at all for some time.

Critical Reflection

Artificial Intelligence can only really be useful to me in the way of guidance. The efficiencies it provides for me, are in that of removing the busy work of refining your design eye, or in Rendering a mock-up sketch into something presentable. It's quite useful for gathering references and finding different sources for historical techniques with incredible specificity. They can be used as a research tool, and as a replacement for a traditional search engine, they aren't the worst. in terms of rendering they can provide many avenues to speed up the ideation process. If it takes you only a few minutes to be able to picture what your final design might resemble, you have more time to iterate and refine.

That being said though, I would not trust AI to render something remotely polished. Even now it still has many limitations, I was dealing with inconsistencies in the form of the picture I sent, many distortions and some hallucinations.

Some of the issues with AI cannot be solved through better trained models, updates or more data centres. Such as those of the actual tech itself being too good. It's easy to use as a crutch. One might claim it's only helping to reduce the friction of getting tasks done, but I personally find that that friction is what causes your design eye to develop and what helps you to work faster and with higher quality. If you train your skills in drawing all your plans by hand, you will train far more than you draftsmanship, and these skills build up over time.

We need to train our traditional skills, not in spite of new technologies that might replace it but specifically due to them. AI is limited, in that when it's used well it can only replace unneeded work, but it is so often used to speed up or circumvent needed work. Not unjustly, that is what its designers intended. Though we have to push against this vehemently.

I sincerely fear that Artificial intelligence as a tool to create more efficiency is destroying that. It is undeniably a useful tool in our current landscape, though. Obviously I am not suggesting that we should all leave all new technology and switch to drawing things by hand as if it is the 1950s, but even Computer Assisted Design keeps all of that skill in your own hands. I am not someone who feels that new technology is something to fear. I am actually quite fond of technology, or at least I was back when there were things to be fond of. I say this to mean that I am not speaking from a place of ignorance about AI. I keep myself updated on most AI news and in general like to stay aware of new developments. However the ways in which AI is used do turn me off on a strict moral level, regardless of how it could be used. It is undeniable that AI is an incredibly impressive technology. One that is useful for us and many other fields. However with time it becomes increasingly obvious the people designing these tools are not making them to help designers.

Ethically speaking there aren't many concerns in terms of the safety of the building. All of the safety and the majority of the design was made by me, and all the changes from the AI have been overseen and checked by me. I do remain responsible for whatever is made under my name. The ethics of the AI system itself is a different question, though. This very much is not similar to the advent of technologies such as CAD or BIM software, or 3D rendering software. They resemble each other at a glance, in the sense that they both made obsolete work of what beforehand had been an industry-wide standard. AI is different though, with BIM and CAD software, what was being replaced was the desk at the architects office, and the environment the office itself is in. AI is replacing the actual role the architect takes in his own design work. if AI is what provides answers it is very starkly taking a position that had been traditionally our own, as architects.

The judgement of architects is still valuable. Our knowledge is still valuable, but these are attempts to replace our expertise.

My judgement of the performance of the AI is middling. On the one hand, It provided me with useful information, and it undeniably sped up my workflow by a considerable amount. On the other hand I ethically cannot use it without directly helping the people actively working towards making my job obsolete, because they don't want to pay another worker's wage. I cannot, with sound conscience, bear that on my self. I do not harbour ill will to anyone who uses it. After all it is a tool, and you do not blame a carpenter for replacing an axe with a chainsaw.

Despite this all I remain optimistic on the future of the industry. I don't believe that people will be replaced entirely, and perhaps there's even a future where AI is seamlessly integrated into the workflow. As it is right now, though, the development of AI is directly at odds with everything I stand for. Unless the industry changes very soon, I do not foresee myself using AI in the near future.

Page updated

Google Sites

Report abuse