TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · Judgment · AGE 18—21 · POSITIONAL · HELD
An image is not a proposal
A frame can hold all six layers or hold two and fake four.
The dilemma
A render arrives that makes the room go quiet. It looks finished. Do you treat it as a proposal, something that could go to a client, or as an image that may or may not contain a proposal?
The choices
Accept the frame as the scheme. Or take it apart into six layers and check each one. Is there an image? (Yes, obviously.) Is there an idea, a stance about how this place should work? Is there geometry: could this be drawn in plan and section? Is it buildable: would the structure stand, would the junctions resolve? Is there atmosphere, the feel of the light and material? Is there a story, something it says about who lives here and how? A frame can hold all six, or hold two and fake four.
The consequence
Treat a strong image as a proposal and the faked layers show up later: at coordination, at costing, in front of the client who took measurements off the render. Take it apart first and you know what you actually have. Perhaps a strong atmosphere and a story, with no geometry yet. That is useful. It tells you what to draw next.
The case
A client in Hubballi forwards an AI image on WhatsApp: "something like this." A cantilevered timber box over a stone base, golden light, a single tree. The student goes through the six. Image: yes. Atmosphere: yes, strong. Story: a retreat, a view. Idea: unclear. Why the cantilever? Geometry: the box has no depth that could be drawn. Buildability: nothing is holding the cantilever up. Two and a half out of six. The client's expectations were set by the three that are missing.
Our position
A beautiful image is not a finished proposal. It is a frame that may contain parts of one. The designer's job is to say which parts, and the six-layer check is how the Lab teaches that reading.
Why we hold it
This rule is older than AI. A hand-drawn perspective could always flatter an unresolved scheme. What has changed is the volume and the polish. Images now arrive looking finished before any thinking has been done, and they arrive from clients as well as from students. The check has to be explicit because the temptation is now constant.
The strongest objection
Every proposal begins as an image, and the best designers have always worked backwards from the picture. Insisting on six layers before an image "counts" may kill the moment when a frame is genuinely ahead of its own geometry, when the sketch knows more than the person who drew it. A rigid check risks marking intuition as fraud.
What would make us revise it
This card does not forbid the image. It forbids calling the image a proposal. We would soften it if we saw students who had been taught the six layers become unable to start from an image, which would mean the check was producing caution where it should produce clarity. We watch for that in workshops.
Try it
Take one image and check it across all six: image, idea, geometry, buildability, atmosphere, story. Then find one case where the image is strong but the idea is missing, and one where the idea is strong and the image is weak. The second is the more valuable frame, and the harder one to defend in a room.
Take it to crit
When the student pins up a render, ask which of the six layers it really holds. Watch whether they already know. A student who has checked will answer in order. A student who has not will defend the picture.
How it works
Image models learn what finished buildings look like in photographs. Atmosphere and story are in the photographs. Geometry and structure mostly are not, because a photograph does not contain the plan. So a model is strong on the layers a photograph carries and weak on the ones it does not. A render that scores high on the first and low on the second is not a flaw in the model. It is what the model was fed, showing through.
What this idea builds on
What this idea opens up
Sources
- image ≠ resolved proposal doctrine (architectural-thinking rule)
- The AI Design Director Protocol
Open this idea on the map · The complete map · Logika · RBDS AI Lab, India · revised every edition.