Midjourney for Architects: Practical Prompts, Workflows, and Honest Limitations

Midjourney is an AI image generation tool that creates images from text descriptions (prompts). It runs through Discord - you type a text prompt in a Discord channel, and the AI returns four image variations within about a minute.

That’s it. It’s not a design platform, it doesn’t have CAD integration, and it doesn’t understand building codes. But for the specific task of generating architectural concept imagery, mood references, and early-stage visualisations, it’s genuinely useful when you know how to prompt it properly.

---

How Midjourney Actually Works for Architecture

You interact with Midjourney by typing /imagine followed by a text description in Discord. The AI generates four images based on your prompt. You can then:

• Upscale (U1-U4): Generate a higher-resolution version of one image
• Vary (V1-V4): Generate variations of one image
• Pan/Zoom: Extend the image in any direction
• Re-roll: Generate four entirely new images from the same prompt

The quality of output depends almost entirely on your prompt. A vague prompt produces generic results. A structured prompt produces images that are actually useful for design communication.

---

Prompt Structure for Architectural Images

The key to useful Midjourney output is a structured prompt. Here’s a framework:

[Subject] + [Materials/Style] + [Context/Setting] + [Atmosphere/Lighting] + [Technical Parameters]

Example Prompts (With Explanations)

Exterior concept:

/imagine a two-storey residential house with exposed timber frame and
white rendered walls, surrounded by mature oak trees, golden hour
lighting, architectural photography, wide angle lens --ar 16:9 --v 6.1

Interior mood:

/imagine minimalist open-plan living space with polished concrete
floor, floor-to-ceiling glazing overlooking a forest, warm afternoon
light casting long shadows, Scandinavian furniture, architectural
interior photography --ar 16:9 --v 6.1

Urban context:

/imagine mixed-use building with ground floor retail and residential
above, red brick facade with steel balconies, narrow European street
with cobblestones, overcast sky, street-level perspective,
architectural rendering --ar 16:9 --v 6.1

Prompt Elements That Matter Most

Element	Effect	Example
Material specification	Controls surface appearance	”Corten steel cladding”, “exposed aggregate concrete”, “charred timber”
Photography style	Controls composition and realism	”architectural photography”, “magazine editorial”, “drone aerial view”
Lighting condition	Sets mood and atmosphere	”golden hour”, “overcast diffused light”, “blue hour dusk”
Lens/perspective	Controls framing	”wide angle lens”, “eye-level perspective”, “birds eye view”
Aspect ratio (—ar)	Output dimensions	--ar 16:9 for presentations, --ar 3:2 for portfolio, --ar 1:1 for social
Version (—v)	Model version	--v 6.1 (latest) produces most photorealistic results
Stylize (—s)	AI creativity level	--s 100 (default), --s 250 (more artistic), --s 50 (more literal)

---

What Midjourney Is Good At (Use It For This)

1. Early Concept Exploration

Before you’ve drawn a single line, Midjourney can help you explore the design space. Generate 20 variations of “a timber community centre in a woodland clearing” and you’ll see approaches you wouldn’t have considered.

Workflow: Generate 50-100 images exploring different massing, material, and context combinations. Print the best 10-15 and pin them up. Use them as a springboard for design development - not as the design itself.

2. Client Mood References

Instead of spending hours assembling Pinterest boards, generate custom mood imagery that matches your specific project context. “Show me what a warm, modern house extension in a Victorian terrace could feel like” is faster and more targeted than searching reference libraries.

3. Competition Atmosphere Images

For competition entries where the design is developed but you need evocative imagery quickly, Midjourney can produce atmospheric visuals that complement your technical drawings and diagrams.

4. Design Brief Visualisation

When briefing a client or presenting to a planning committee, AI-generated context images can communicate spatial qualities (light, scale, materiality) faster than words.

---

What Midjourney Is Bad At (Don’t Use It For This)

Architectural Accuracy

Midjourney doesn’t understand construction. It will generate:

• Columns that don’t align with structure above
• Windows floating in walls without proper reveals or frames
• Stairs with inconsistent riser heights
• Impossible structural spans
• Buildings that couldn’t physically stand up

Never present Midjourney output as a buildable design. It’s inspiration and communication, not technical output.

Consistency Across Views

You can’t generate a north elevation and then ask for “the same building from the south.” Midjourney doesn’t maintain a 3D model internally - each image is a new generation. If you need multiple consistent views of a design, use your actual BIM/3D modelling tools.

Interior Space Planning

Midjourney produces beautiful interior images but can’t maintain accurate room proportions, furniture scale, or spatial relationships. An AI-generated “open plan kitchen-living room” may look great but have a 1.5m wide dining table and a 4m tall ceiling.

Detail Design

Don’t try to use Midjourney for junction details, construction sections, or material specifications. The AI approximates appearances but doesn’t understand how buildings are actually put together.

---

Integrating Midjourney Into Your Design Workflow

Phase-Appropriate Use

Design Phase	Midjourney Use	Appropriate?
Brief/feasibility	Mood imagery, context exploration	Yes - ideal
Concept design	Massing exploration, atmosphere references	Yes - with caveats
Developed design	Presentation imagery alongside real drawings	Carefully
Technical design	None	No
Construction	None	No

Combining With Other Tools

Midjourney + SketchUp/Rhino: Model your design in 3D, export a screenshot, then use Midjourney’s image-to-image mode to add atmosphere, materials, and context. This gives you architectural accuracy from your model plus the visual quality of AI generation.

Midjourney + Photoshop: Use AI-generated images as backgrounds, sky replacements, or vegetation layers in your rendered compositions. This is faster than finding stock photography and gives you more control over the aesthetic.

Midjourney + InDesign: Generate custom page backgrounds, divider images, and concept boards for presentations directly from prompts describing your project.

---

Practical Tips From Regular Use

Be specific about materials. “Modern house” produces generic results. “House with blackened timber cladding, zinc standing seam roof, and recessed aluminium window frames” produces something you can actually discuss with a client.

Control the perspective. Add “eye-level perspective,” “drone aerial shot,” or “worm’s eye view” to your prompts. Without this, Midjourney defaults to whatever angle it finds most visually interesting, which may not be useful.

Use negative prompts. Add --no people, cars, text to remove distracting elements that Midjourney adds by default.

Save your best prompts. Build a library of prompt templates for common situations (exterior residential, interior commercial, urban streetscape, landscape). Modify the material and context details for each project.

Generate in batches. Don’t refine one image endlessly. Generate 20-30 images, select the 3-5 best, then iterate on those. The variation between generations is part of the value.

---

Ethical Considerations

Client transparency: If you use AI-generated images in a presentation, clients should know they’re AI-generated, not photographs or accurate renders of the proposed design. Misrepresenting AI images as technical visualisations is misleading.

Copyright: The legal status of AI-generated images varies by jurisdiction. In most cases, you can use Midjourney output commercially (paid plans include commercial rights), but the images can’t be copyrighted. If your competition entry relies heavily on AI imagery, be aware of this.

Credit: In competition entries or publications, disclosing AI tool use is becoming standard practice. This is honest and avoids the impression that the imagery represents a higher level of design resolution than it does.

---

Want to explore AI and computational design tools? The Archgyan Academy offers courses that help architects integrate new technology into practical design workflows.