AI Prompts for Adaptive Reuse & Renovation Concepts
You're standing in a derelict Victorian pump station. The client wants to turn it into a boutique hotel. They're asking you to "show them what it could look like" -- but you're three months away from having a 3D model, and the budget for speculative CGI is zero.
This is where AI visualization earns its keep in adaptive reuse work. Converting old buildings into new uses is one of the most visually compelling briefs in architecture, and it's perfectly suited to AI image generation. You've got a real existing building (photographs to feed as references), a clear before/after narrative, and clients who need to see the potential before they'll commit the funds.
The prompts below cover common conversion typologies, before/after techniques using img2img, and ChatGPT prompts for the feasibility analysis and written components of adaptive reuse projects. If you want deeper context on the design principles behind building conversion, our guide to adaptive reuse in architecture covers the planning, structural, and sustainability considerations in detail.
Warehouse and Industrial Conversions
Industrial buildings are the bread and butter of adaptive reuse. High ceilings, open spans, raw materials -- they practically beg to be converted.
Warehouse to Residential Loft
"Photorealistic interior of a converted Victorian brick warehouse, now a luxury residential loft apartment. Original features preserved: exposed red brick walls, cast iron columns, heavy timber beams and joinery overhead, original timber floorboards restored. Contemporary insertions: a freestanding white kitchen island with Carrara marble top, a mezzanine sleeping platform accessed by a black steel open-tread staircase. Double-height space with original arched windows. Late afternoon light casting long shadows across the floor. Industrial heritage meets Scandinavian minimalism."
"Photorealistic interior of an industrial warehouse conversion into a co-working space. Preserved: steel roof trusses, concrete floor (polished), brick perimeter walls. New elements: a two-storey glass-enclosed meeting pod structure inserted freestanding within the warehouse volume, green planting walls on the mezzanine level, pendant lighting hanging from the trusses at varying heights. Workbenches and desks arranged in clusters. Atmosphere: productive, light-filled, warm."
"Photorealistic exterior of a converted waterfront warehouse complex, now mixed-use with residential above and a food hall at ground level. Original red brick facades retained and cleaned. New interventions: corten steel-framed balconies inserted into existing window openings on upper floors, a new glass-and-steel entrance canopy at ground level, industrial-style signage. Cobblestone quayside with outdoor seating. Evening, warm light spilling from the food hall."
Factory to Creative Offices
"Photorealistic interior of a former textile factory converted into a creative studio. Triple-height space with original sawtooth roof providing north light through clerestory glazing. Concrete frame and brick infill walls retained. New insertion: a three-level timber CLT structure built inside the factory shell, containing private offices on two levels and a rooftop terrace below the sawtooth roof. Open collaboration space at ground level between the new structure and the existing walls. Industrial pendant lights. Working atmosphere with designers at desks."
"Photorealistic interior of a converted 1960s factory unit, now a tech startup office. Existing: exposed concrete waffle slab ceiling, concrete columns on a 6m grid, strip windows along one wall. New: epoxy-resin floor in warm grey, acoustic felt panels in muted colors on walls between windows, modular furniture system. Phone booths and meeting pods in birch plywood scattered throughout. No false ceiling -- services exposed and painted matte black. Daylight and task lighting combined."
Church and Chapel Conversions
Religious buildings offer drama that few other typologies can match. Vaulted ceilings, stained glass, monumental scale -- the challenge is inserting contemporary life without destroying the atmosphere.
Church to Restaurant
"Photorealistic interior of a converted Gothic revival church functioning as a fine dining restaurant. Original features: pointed stone arches, ribbed vaulting, stained glass windows in east wall, stone column arcade. Contemporary intervention: a long communal dining table running the length of the nave in dark fumed oak, individual tables in the side aisles, a cocktail bar inserted into the former choir stall area in dark green marble with brass fittings. Lighting: warm pendant fixtures hanging between the vault ribs, candlelight on tables, cool colored light filtering through stained glass at dusk. 50 covers. Atmosphere: sacred turned sensual."
"Photorealistic interior of a small chapel converted into a private residence living room. Whitewashed stone walls. Original lancet windows with clear glass replacing stained glass. Timber pew ends repurposed as a room divider. Contemporary furniture: deep linen sofa in natural, a round walnut dining table beneath the rose window. Wood-burning stove in a simple steel surround where the altar was. Exposed timber roof structure. Winter afternoon light. Quiet, warm, devotional quality persisting in domestic use."
Church to Community Hub
"Photorealistic interior of a decommissioned modernist 1960s church converted into a community arts center. Original: raw concrete walls with board-marked texture, triangular clerestory windows, exposed concrete portal frame structure. New program: flexible performance space in the nave (retractable seating for 120), art gallery in the former side chapel, community cafe in the narthex with views to the street through new full-height glazing replacing the original entrance doors. Hanging art installation in the main space. Active, lively atmosphere with people."
Before/After Prompts Using img2img
The most powerful technique for adaptive reuse visualization is img2img -- feeding a photograph of the existing building as a reference and prompting the AI to show the converted version. This works in Stable Diffusion (ControlNet) and to a limited extent in Midjourney (image references).
Stable Diffusion ControlNet Approach
Upload your photograph of the existing building as the ControlNet input. Use Canny edge detection to preserve the building geometry.
"[Upload existing building photo as ControlNet Canny input, strength 0.6-0.75] Transform this derelict industrial building into a renovated mixed-use development. Keep the existing brick facade, window openings, and roofline exactly as shown. Clean the brickwork. Replace broken windows with new steel-framed double-glazed units. Add a contemporary glass entrance at ground level where the loading bay doors are. New corten steel signage reading 'THE MILL.' Landscaping: new paved forecourt with bench seating and birch trees. People entering. Daytime, bright overcast."
"[Upload interior photo as ControlNet Depth input, strength 0.5-0.7] Convert this abandoned factory interior into a modern restaurant. Preserve the existing concrete columns, ceiling structure, and window positions exactly as shown in the photo. New: polished concrete floor, pendant lighting at 2.5m height between columns, banquette seating along the window wall in dark green leather, open kitchen visible at the far end. Warm evening lighting. Occupied, 60 covers."
Key Settings for Before/After
| Parameter | Recommended Setting | Why |
|---|---|---|
| ControlNet Mode | Canny or Depth | Canny preserves edges (good for facades), Depth preserves spatial relationships (good for interiors) |
| ControlNet Strength | 0.5 - 0.75 | Too low = ignores building geometry. Too high = preserves too much of the derelict state |
| Denoising Strength | 0.4 - 0.65 | Lower = closer to original photo. Higher = more creative freedom for the AI |
| Checkpoint | juggernautXL or RealVisXL | Best photorealism for architectural subjects |
| Negative Prompt | "ruin, derelict, broken, dirty, abandoned, rubble, graffiti" | Pushes the output toward the renovated state |
| CFG Scale | 7 - 9 | Good prompt adherence without oversaturation |
Midjourney Image Reference Approach
Midjourney doesn't have ControlNet, but you can use image references to guide the output.
"/imagine [paste URL of existing building photo] photorealistic renovation of this building. Same structure and facade proportions retained. New: cleaned brickwork, new aluminum windows in all openings, contemporary glass extension on the roof, landscaped entrance with outdoor seating. Converted to a boutique hotel. Daytime, summer. Architectural photography --iw 1.5 --ar 16:9"
The --iw (image weight) parameter controls how much influence the reference photo has. For adaptive reuse, 1.0-1.5 is the sweet spot -- enough to keep the building recognizable, loose enough to show the transformation.
ChatGPT Prompts for Feasibility Analysis
Beyond visualization, ChatGPT is useful for the analytical and written components of adaptive reuse projects.
Initial Feasibility Assessment
"I'm evaluating the adaptive reuse potential of a [building type, e.g., '1920s brick warehouse, 2,000 sqm, three storeys, structural survey indicates good condition, located in a city center conservation area']. The proposed new use is [e.g., 'residential apartments with ground-floor commercial']. Help me draft a preliminary feasibility assessment covering: 1) Structural considerations and likely interventions needed, 2) Planning and heritage consent requirements, 3) Building regulations challenges (fire, access, thermal performance), 4) Approximate cost premium vs. new build, 5) Key risks and red flags. Format as a structured report with sections."
"Compare the feasibility of three potential new uses for a decommissioned [building type] in [location]: Option A: [use], Option B: [use], Option C: [use]. For each option, assess: market demand in the area, compatibility with the existing building form, planning likelihood, estimated conversion cost per sqm, and revenue potential. Present as a comparison table with a recommendation."
Heritage and Planning Statements
"Draft a heritage impact assessment section (500 words) for converting a Grade II listed [building type] into [new use]. The key alterations proposed are: [list alterations]. Structure the assessment as: significance of the heritage asset, how each alteration impacts that significance, mitigation measures, and justification for why the conversion is beneficial. Reference the NPPF approach to heritage assets."
"Write a Design & Access Statement section (400 words) explaining the design approach for an adaptive reuse project converting [original building] into [new use]. Cover: design philosophy (what to preserve, what to insert, how old and new are distinguished), access strategy (how the new use changes circulation and entrance points), and sustainability benefits of reuse vs. demolition. Tone should be professional but persuasive."
Cost and Program Analysis
"Create a preliminary cost comparison between adaptive reuse and new build for a [building type] conversion to [new use] in the UK. Assume: existing building is [describe condition], gross area is [X] sqm, and the site is in a [urban/suburban] location. Include line items for: structural works, facade retention/repair, new services, internal fit-out, heritage specialist costs, planning application costs, and contingency. Present as a table with cost per sqm estimates for both scenarios."
"Help me develop a room-by-room schedule of accommodation for converting a [building type] into [new use]. The existing building has: [describe key spaces, dimensions, floor levels]. For each existing space, suggest the best-fit new use based on ceiling height, natural light, structural grid, and access. Identify spaces that need significant modification and spaces that convert easily."
Typology-Specific Prompts
School to Housing
"Photorealistic interior of a converted Victorian primary school, now a residential apartment. Former classroom with 3.5m ceilings and large sash windows on two walls. Preserved: original cornicing, high skirting boards, timber window frames, one section of original blackboard as a feature. New: open-plan kitchen-living room, engineered oak flooring, contemporary furniture in warm neutrals. The scale of the room makes the apartment feel grand despite modest square footage. Morning light flooding in."
Hospital to Hotel
"Photorealistic hotel bedroom in a converted Art Deco hospital building. Original: terrazzo floor (restored), curved plaster ceiling with geometric relief pattern, steel Crittall windows. New: boutique hotel fit-out with upholstered bed frame in midnight blue velvet, wall-mounted reading lights in brushed brass, freestanding bath positioned near the window. The clinical geometry of the original building given warmth through rich textiles and soft lighting. Evening atmosphere."
Cinema to Performance Venue
"Photorealistic interior of a 1930s Art Deco cinema converted into a live music and performance venue. Preserved: the original proscenium arch and decorative plaster surround, stepped ceiling with geometric coffers, curved balcony structure. Modified: original fixed seating removed, replaced with flexible standing/seated configuration, new acoustic panels in sympathetic dark fabric behind preserved decorative elements. Stage lighting rig integrated into the existing ceiling structure. Sound check in progress with a band on stage."
Post Office to Food Market
"Photorealistic interior of a grand Edwardian post office hall converted into an indoor food market. Preserved: barrel-vaulted ceiling with decorative plasterwork, marble floor (repaired and polished), original brass-grilled counter positions. New use: individual food vendor stalls inserted into the original counter bays with custom timber and steel counters, central communal dining tables, atmospheric pendant lighting. Busy lunchtime scene with customers and vendors. Natural light from the clerestory windows mixing with warm artificial light."
Prompts for Structural Intervention Visualization
One of the hardest things to communicate in adaptive reuse is how new structure meets old. These prompts focus on that junction.
"Photorealistic architectural detail showing a new steel mezzanine structure inserted into a historic masonry building. The new steel beams are fixed to the existing brick wall using visible bolted connection plates. Clear gap between new steelwork and old masonry -- no direct bonding. Steel finished in matte dark grey paint. Original brick left exposed and unpainted. Close-up showing the tectonic honesty of the intervention: you can read what's old and what's new."
"Photorealistic interior showing a new glass floor panel inserted into a historic building, revealing original stone foundations below. Structural glass floor with steel frame, flush with the surrounding original timber floorboards. Lighting beneath illuminates medieval stonework. A contemporary reading nook is arranged around the glass panel. The archaeological layer is visible and celebrated, not hidden."
"Photorealistic architectural photograph of a contemporary rooftop extension on a Victorian terrace building. The extension is set back from the parapet, clad in dark zinc standing seam, with full-height glazing facing the rear garden. The original slate roof and chimney stacks remain on either side. The junction between old and new is a shadow gap -- the extension floats above the original building without touching the party walls. Street-level view showing the extension barely visible above the parapet."
For quick before/after concept testing, ArchGee's AI design tools let you upload a photo of an existing building or interior and generate redesigned versions -- useful for client presentations before committing to detailed proposals.
FAQ
How accurate are AI-generated adaptive reuse visualizations?
They're conceptually useful but not dimensionally accurate. AI doesn't understand structural capacity, building regulations, or actual room dimensions. An AI-generated image of a warehouse conversion shows what the atmosphere and material quality could be, not what the finished building will actually look like. Use them for client buy-in, mood communication, and design direction -- not for construction documentation or planning applications. Always label them as "concept visualization, for illustrative purposes only."
What's the best ControlNet mode for before/after building conversion images?
Canny edge detection works best for exteriors because it preserves the outline, window positions, and roofline of the existing building. Depth mode works better for interiors because it preserves the spatial relationships (foreground furniture, midground columns, background walls) rather than every edge. For facade renovations specifically, try MLSD (line segment detection) -- it picks up straight architectural lines while ignoring surface texture, which helps when the existing building is dirty or overgrown.
Can I use AI to visualize the construction process of an adaptive reuse project?
To a limited extent. You can create "during construction" visualizations showing partial demolition, exposed structure, and scaffolding by prompting for those elements specifically. However, AI doesn't understand construction sequencing, so the images may show structurally impossible intermediate states. For phased renovation visuals, create separate prompts for each phase rather than asking for a sequence, and have a structural engineer sanity-check the interim states you're showing.
How do heritage officers react to AI-generated images in planning applications?
Mixed reactions as of 2026. Some conservation officers appreciate AI visualizations as supplementary material that helps them understand the proposal's intent. Others dismiss them as misleading because AI doesn't accurately represent the existing building fabric. Best practice: use AI renders as discussion aids in pre-application meetings, but submit verified CGI or photomontages for the formal application. Always pair AI images with measured survey drawings and condition photographs.
What's the ideal ControlNet strength for keeping a building recognizable?
Between 0.55 and 0.70 for most adaptive reuse work. Below 0.5, the AI ignores too much of the original building geometry and produces something that doesn't look like the actual building. Above 0.75, it preserves too much detail -- including the derelict state, broken windows, and grime -- making it hard to show the renovated condition. Start at 0.65 and adjust based on how much of the original character you want to preserve versus transform.