Systematizing Generative Design: A Framework for Controlled AI Spatial Prototyping

AI Productivity | AI in Architecture | Midjourney | Parametric Design | Architectural Visualisation | AI Workflows | AI For Business | AI Design Tools | Interior Design AI

If you've ever generated a beautiful Midjourney interior and then watched it completely change when you swapped one material, you know the lottery problem.

The Problem

You've used Midjourney. You know it can produce beautiful architectural imagery. But the moment you need consistency — six material options for the same space, or a client deck where every image reads as a direct comparison — the tool falls apart.

Ask for the same living room twice with different floors. You get two different rooms. Different scale. Different camera angle. Different ceiling height. Different light. You cannot present this to a client. You cannot use it for a material decision. Standard prompting has no mechanism for spatial continuity between generations.

This is the lottery problem. And it makes Midjourney useless for real architectural work — until now.

The Solution

Systematizing Generative Design gives you the complete Controlled Iteration Framework: a four-step parametric workflow that transforms Midjourney into a precision instrument for generating spatially consistent, client-ready imagery.

The framework uses four advanced Midjourney parameters — **--sref**, **--cref**, **--seed**, and ::weighting — applied in a strict sequence. You set the spatial DNA once, in a hero image. Every subsequent generation inherits that DNA exactly. Only your specified variable changes. Six images of the same space. Same light. Same perspective. Different material. That is what this framework produces.

Tested with real architectural practitioners. Every parameter value is documented. Every workflow step is reversible.

AI generates. The architect authorises. The client decides.

What's Inside

The Guide (~28 pages):

• The Controlled Iteration Framework — four steps, fully documented
• Step 1: Hero Image Establishment — how to generate a spatially accurate baseline
• Step 2: DNA Lock — applying --sref, --cref, and --seed to anchor consistency
• Step 3: Controlled Variable Iteration — changing one variable at a time, correctly
• Step 4: Orthographic Consistency Check — ::weighting syntax and consistency thresholds
• Two complete worked examples: residential material deck and hotel lobby programme deck
• Full parameter reference tables for --sref, --cref, --sw, --cw, --seed, and ::weighting
• Common errors and how to avoid them — seven documented failure modes
• Legal and professional accountability guidance, including mandatory AI imagery disclosure
• Tool Stack Reference for Midjourney v6.1 and v7

The Prompt Pack (~9 pages, 25 prompts):

• 4 Hero Image Prompts (residential, commercial, exterior, material detail)
• 2 DNA Lock Prompts (validation + first material iteration)
• 7 Materiality Iteration Prompts (concrete, oak, walnut, marble, microcement, brick, terrazzo)
• 4 Programme Iteration Prompts (residential, hotel, gallery, co-working)
• 3 Lighting Iteration Prompts (morning, midday, golden hour)
• Orthographic and Variation prompts for extended applications

All prompts are copy-paste ready. Replace the [Insert X] placeholders. Submit. Done.

Who It's For

• Mid-to-senior architects who have hit the consistency ceiling with Midjourney
• Interior designers who need material decks that clients can actually read as comparisons
• Architectural visualisers who need spatially consistent concept imagery for client meetings
• Property developers who need quick, quality early-phase spatial visuals for stakeholder communication

• Not for you if: You have never used Midjourney (start with the free tool first). You need a design tool, not a workflow guide. You are looking for guaranteed outcomes — this is a tested system, not a magic button.

Tools Used

| Tool | Version |

|------|---------|

| Midjourney | v6.1 and v7 |

| Discord | Web / Desktop (for image generation and seed extraction) |

Tested with Midjourney v6.1 and v7. Workflow is stable across both versions.

Quality Assurance

Every workflow step is documented with tested parameter values. Every prompt structure is explained. The guide includes two complete worked examples with full prompt sequences, consistency threshold tables, and validation checklists. You are never guessing — you are following a system.

Practical workflows, professionally structured.

---

© 2026 Theo Hartley | WIZGUIDES

⚖️ Practical AI for Professionals ⚖️