Visual — Rev.0

100+AI workflows templated

35+Practice projects supported

75+Personal projects developed

20+Tools evaluated

From experiment to workflow

Curiosity and hands-on testing show where new tools can support architectural and interior design work. Experiments become modular workflows that can be reused, expanded and connected, with lessons from practice projects and independent studies informing each iteration.

The focus is on what a method can reliably achieve, what information it needs and where conventional software or manual work is still the better call. That makes it easier to choose the right approach and explain it to others.

Human-directed by design

Visualisation benefits from continuous judgement: reviewing an image, redirecting the design and responding to unexpected ideas. Repeatable production steps can be automated, while decisions about composition, material, atmosphere and architectural intent stay actively directed.

Autonomy varies with the task. Fixed workflows suit predictable transformations; agentic workflows help interpret context and coordinate stages. Both need clear inputs and review proportionate to the cost of an error.

Working with constraints

The starting point varies. Early concept studies leave room to explore form, material and atmosphere, while work on existing designs needs clear limits around geometry, openings, furniture and finishes.

Model views, reference images and selected materials define what stays fixed and what can change. Outputs are checked against the project information, keeping visual exploration tied to the design and easier to discuss with the wider team.

An adaptable toolset

Figma Weave and Krea are the current preferred environments, combining the latest models and processes within a modular visual workflow. The wider toolset covers image generation, editing, research and specialist outputs. Platforms are judged through practical tests on control, output quality, flexibility, accessibility and cost. Tool choice will shift as capabilities develop; the method and design objectives underneath stay useful.

Logos of AI tools grouped into core, specialist and experimental — **Toolset** | Core · Specialist · Experimental

Approach

One reference, three representations

The aim is controlled, reusable agentic workflows that help teams explore and communicate a design. Each method combines the steps its task needs. These examples use a shared architectural reference to explore three forms of representation.

Pencil sketch of a classical orangery — **Custom sketch style** | Weave tool

Watercolour of the same orangery — **Custom watercolour style** | Weave tool

Photoreal dusk render of the orangery reflected in water — **Realistic visualisation** | Agentic workflow

Agentic system · Independent concept study

Hidden Market

Terracotta Tagine / Amman, Jordan · Agentic workflow | Weave

Terracotta Tagine explores a market in Amman inspired by the form of the traditional tagine lid. A landscape of terracotta cones gives the proposal its identity and introduces a speculative idea for passive cooling.

Weave node graph connecting concept, sketch, sections and results — **Agentic workflow** | Weave node graph

Hero image

An initial hero image sets the visual direction, bringing form, material, scale and atmosphere into one view so the idea is easier to question and develop.

Once selected and refined, it becomes the shared reference for everything that follows, keeping each new view tied to the same concept.

Terracotta cones rising above a busy market — **Hero image** | Starting reference

Man reading beneath terracotta cones — **Taking it slow**

Line sketch of the cone market — **Concept sketch**

Developing the system

The workflow brings together the brief, references and written context to develop an initial concept and a connected set of visual studies.

Each branch has a purpose. A section explores spatial relationships, a night view tests another atmosphere, a sketch communicates the underlying idea. Outputs are reviewed against the original concept, with further iteration wherever form, material or character begins to drift.

Light shaft falling through a cone into the market — **Movement & light**

Cutaway axonometric diagram of the cones — **3D axo diagram**

Aerial night view of the lit market — **Market at night**

Rendered axonometric section through the market — **3D axo section**

What the example shows

The value lies in a coherent set of views from a shared starting point. The system is a reusable structure for exploring a project; its content and direction stay specific to the design.

Control & refinement

Developing existing material around a defined task

Each workflow starts with a different balance of freedom and constraint, from exploring atmosphere to working with fixed geometry and selected furniture.

Story narrative

Develops design-story narratives for architectural visualisations, strengthening visual impact while adding atmosphere, character and a sense of wonder.

Dark rural house at dusk — **Before** | Source visualisation

Same house under a crescent moon with mist and chimney smoke — **After** | Story narrative workflow

Clay to render

Clay-to-render workflows vary in complexity. Some reimagine a project in a style or loose material palette; others, like this one, work within much tighter limits. Existing spaces, fixed geometry and specific furniture and materials in defined locations all have to be respected, which calls for a complex, multi-step workflow.

Collage of selected furniture and timber samples — **Inputs** | Selected furniture and materials

Warm rendered living room with red sofa and timber floor — **Result** | Materialised living room

Empty panelled room with red walls — **Existing space**

Grey clay render of the furnished room — **Clay model**

Render refresh

Refreshing old visualisations is usually tedious. This workflow sets a robust framework that pushes the AI to reinterpret the source image as a realistic visualisation while preserving the original design intent.

Older courtyard render with palms — **Before** | Original visualisation

Refreshed photoreal courtyard render — **After** | Refreshed

Style & representation

Translating a space into another format

Different questions call for different visual formats. These examples translate a space, material palette or design idea into another representation, using references to guide the character of the result.

Interior restyle

Reworks interior images into a specific style taken from a reference image, intelligently keeping, replacing or reshuffling selected elements such as appliances, fixtures and furniture.

Rustic kitchen reference — **Reference**

Kitchen restyled with pale blue cabinets and timber — **After** | Restyled kitchen

3D axo

Creates realistic miniature axonometric visualisations, using reference images for materials and atmosphere and keeping the native model geometry as the base.

Bedroom material reference board — **Material references**

Line axonometric of a bedroom — **Model geometry**

Rendered miniature bedroom axonometric — **Result**

Material mood board

Extracts materials and key elements from reference images to create inspirational material mood boards, part of a wider pipeline for different board types and visual studies.

Collage of interior reference photographs — **Reference images**

Physical-style material mood board with samples — **Result** | Material mood board

Style creator

Distinctive visual content complements architectural visualisation and helps work stand out. This workflow builds those styles.

Photoreal barn house in meadow — **Source**

Grid of printmaking style references — **Style references**

Barn house in a retro collage style with pink sky — **Result** | Custom collage style