Use case

Reference Tool for Architects: Image Library Workflow (2026)

By refernLast updated September 202613 min read

Architects accumulate references across the full design process: precedent buildings, material samples, construction details, landscape cues, lighting studies, urban typologies. A typical project touches hundreds of images before a single line is drawn. The right reference tool makes that material findable, composable, and available when you are standing on site with no WiFi.

This guide covers how to build and use an architecture reference library in refern, from first import through material boards on the canvas and offline access in the field.

What architects actually need from a reference tool

Most tools built for creatives assume an artist collecting inspiration. Architects have a more structured demand: precedent images tied to typology, program, climate, and material; detail references linked back to the projects they informed; material boards that need to survive a client presentation and a site visit.

The core requirements:

  • Organized by typology and project, not just by date saved. A flat inspiration feed fails the moment you have references for three concurrent projects and a personal research archive.
  • Rich tagging. A single building can be relevant for its structure, its envelope, its section, its climate response, and its site relationship. Flat single-word tags are insufficient.
  • A canvas for material boards. Arranging references spatially for a presentation or for your own synthesis is a different task from searching the library. Both need to live in the same tool.
  • Full offline access. Construction sites, client offices, and travel do not guarantee a good internet connection.
  • Ownership. References collected over years of practice should not depend on a subscription staying active or a company staying solvent.

Tools architects commonly use (and their real tradeoffs)

Before covering refern's workflow, it is worth being honest about the alternatives architects reach for and what each one does well.

Are.na

Are.na has a genuine architecture-school following and is worth acknowledging on its merits. The platform's no-algorithm, slow-curation ethos appeals to architects who want a thoughtful research tool rather than an engagement-optimized feed. The public channel model means you can follow another architect's research, share a channel with a studio partner, or browse community archives from institutions like the Guggenheim Museum and Chicago Architecture Biennial. [Source: Are.na About page, are.na/about]

The honest constraints: Are.na is web-only with no offline mode and no desktop app. Everything must be uploaded; you cannot point it at a folder of images already on your disk. There is no canvas for spatial arrangement, no color search, no visual similarity search, and the flat channel structure (no nested folders) degrades as a collection grows. [Source: Are.na help docs, help.are.na/docs/getting-started/channels] The free tier caps at 200 blocks total. Premium is $7/month or $70/year as of 2026. [Source: Are.na About page]

Are.na is a good tool for shared research and community discovery. It is not a local precedent library.

Milanote

Milanote is widely used by interior designers and architects for client-facing moodboards. Its interface is polished, collaboration is real-time, and sharing a board with a client requires no account on their end. Templates for moodboards and creative briefs reduce setup time. [Source: Milanote moodboarding page, milanote.com/product/moodboarding]

The constraints: Milanote is cloud-only with very limited offline access. The free plan caps at 100 total items (notes, images, and links combined), which a working architect exhausts in a single project. The individual plan is $9.99/month billed annually as of 2026. [Source: Milanote pricing page, milanote.com/plans] Performance degrades noticeably above 300 to 500 images per board. [Source: CheckThat.ai analysis] There is no image library at scale, no tagging system, no color search, and no way to maintain a persistent library across projects.

Milanote is useful for client-presentation boards. It is not a reference library for thousands of precedent images.

Eagle

Eagle is the closest desktop equivalent to refern for pure library management. It has smart folders, ratings, color labels, broad format coverage, and a refined tagging UX. This page does not publish a numerical large-library benchmark. The price was $34.95 one-time for 2 devices as of August 31, 2026. [Source: Eagle Store, en.eagle.cool/store]

The structural constraints for architects: Eagle's standard import copies media into an Eagle-managed .library folder rather than indexing originals in place. If source originals are retained, their media bytes exist in both locations, but there is no universal storage multiplier. [Source: Eagle support, en.eagle.cool/support/article/does-eagle-import-files-by-copying-or-indexing] There is no official Linux client as of August 31, 2026. There is also no canvas or moodboard view, so composition work needs another tool.

How refern handles an architecture reference workflow

CapabilityrefernEagleMilanoteAre.na
Local library storage modelNormal indexing is in place; explicit imports and copy actions can create filesStandard import copies media into managed libraryCloud onlyCloud only
Nested folders and hierarchical tagsYesFolders and flat tagsNo tag systemFlat channels only
Smart folders (saved searches)YesYesNoNo
Infinite canvas for material boardsYesNoFreeform board (no canvas layers)No
Core library workflow offlineYesYesVery limitedNo
Color searchYes, by hexYesNoNo
Visual similarity searchYes, localVia pluginNoNo
Relationship graph viewYesNoNoNo
Linux supportYesNoWeb/PWA onlyWeb only
Price (as of 2026)$35 one-time, 3 devices$34.95 one-time, 2 devices$9.99/month individual$7/month or $70/year
Offline on siteYesYesNoNo

refern is a $35 one-time, local-first desktop reference manager with an infinite canvas and a relationship graph view, for artists and designers. It runs on Windows, macOS, and Linux. Core workflows run locally, and normal workspace setup indexes existing folders in place.

Setting up a precedent library in refern

Step 1: Point refern at your existing folders

If you already have a folder of reference images organized by project or typology, there is nothing to migrate. Point refern at that folder as a workspace. Normal setup indexes the images in place. The SQLite database and generated thumbnails live alongside, or in a configured data location associated with, the workspace.

Architects who have been organizing references in a file system for years can keep that structure and add refern's search and tagging layer on top of it without moving a single file.

If your references are spread across several places, for example a precedent archive on one drive and per-project reference folders on another, a single workspace can index several separate folders (content roots) at once, even across different drives. You browse, search, and tag all of them together in one library, and every file stays exactly where it is. Adding extra content folders uses a separate custom data folder for the index.

Step 1b: Decide which file types the workspace shows

A new workspace shows images and video. If your precedent folders also hold PDFs of competition boards, CAD exports, or spec sheets you want findable in the same place, switch those categories on under Edit workspace, "File types shown." Documents covers PDF, DOC, XLS, and CSV; "3D and CAD" covers DWG, DXF, SKP, and STEP; you can also add your own extensions. Enabled PDFs receive a first-page grid preview and open in a native viewer with continuous scrolling, embedded-text search, selection and copy, print, page controls, view-only rotation, and zoom. Scanned PDFs are not OCRed. CAD and other optional formats remain icon-only generic files that you can name-search, tag, folder, and drag back out. Keep exporting PNG or JPEG sheets for CAD drawings you want to see in the grid or place on a material board. The local PDF reference workflow covers the document-specific setup.

Step 2: Build a typology-based folder hierarchy

The most useful architecture reference hierarchy separates by program or typology at the top level, then by project within each typology. A starting structure:

  • Residential (House, Apartment, Housing Collective)
  • Civic (Library, Museum, Government)
  • Cultural
  • Industrial
  • Landscape
  • Details (Connections, Openings, Facades)
  • Materials

This is not a rigid prescription. The right structure is the one that matches how you think about your references when you are searching for them.

Step 3: Apply hierarchical tags for cross-cutting dimensions

Folders handle the typology axis. Tags handle everything else: structural system, material palette, climate zone, architect or practice, era, and spatial quality.

refern supports hierarchical tags, so you can nest related terms:

  • Material > Concrete > Exposed Concrete
  • Material > Timber > Mass Timber
  • Structure > Load-bearing Wall
  • Structure > Column and Beam
  • Climate > Hot and Arid
  • Climate > Nordic
  • Architect > Sverre Fehn
  • Architect > Peter Zumthor
  • Quality > Light
  • Quality > Mass

Tag macros let you define a shortcut that applies multiple tags at once. A macro named timber-nordic could apply Material > Timber, Climate > Nordic, and Quality > Light in a single keystroke, speeding up batch tagging sessions.

Step 4: Record source metadata

For each precedent, recording the source is as important as the image itself. refern has a source URL field, creator field, and free-text notes field on every image. A useful notes entry for a precedent image might include the building name, architect, location, year, and the aspect of the project you are referencing it for.

When you use the browser extension to save an image from an architecture publication or archival database, refern captures the source URL automatically.

Step 5: Use smart folders for project-specific research

Smart folders are saved search queries that auto-populate based on any combination of tags, ratings, metadata fields, and text. For a project involving a timber civic building in a cold climate, a smart folder with the query tag:Civic tag:"Mass Timber" tag:"Cold Climate" surfaces every relevant precedent in one click, without duplicating images into a project folder.

Smart folders update automatically as you add new images. Add a new precedent and tag it correctly; it appears in the relevant smart folder without any further action.

Building material boards on the canvas

Once a library is organized, the next task is synthesis: composing selected references into a spatial arrangement for your own thinking or for a client presentation.

refern's infinite canvas supports this workflow without leaving the app.

Drag from the library onto the canvas. Select images from any folder, search result, or smart folder and drag them directly onto a canvas file. No export, no copy-paste, no intermediate step.

Arrange by material family or spatial sequence. The canvas supports layers and groups, so you can create separate layers for structural references, envelope references, and interior material references and toggle between them. Group backgrounds let you box each layer visually, similar to a Photoshop frame.

Add text annotations. Label each reference with a note about what you are drawing from it. Text elements on the canvas are separate from the image metadata, so the canvas annotation can be specific to the board without affecting the library record.

Add color swatches. refern's canvas includes color swatch elements. Pull dominant colors from your material references and place them alongside the images to build a palette study.

Non-destructive crop. If you want to isolate a facade detail or a section from a larger image, the canvas includes non-destructive crop. The original image in the library is untouched.

Pin the canvas on top. For drawing sessions where you want the material board visible while working in another application, refern supports a pin-window-on-top mode with adjustable transparency and mouse clickthrough. This is the same overlay use case that PureRef serves, built into the same tool as the library.

Using the reference library on site

Construction sites and client meetings are where the library needs to be most reliable. Previously indexed media, canvases, search, and tags remain available without a connection. Network-dependent actions such as license activation, updates, optional model downloads, feedback, and user-requested web downloads are separate.

For architects who work across a desktop and a laptop, one refern license covers up to 3 devices, but the app does not currently include cross-device sync. Do not treat simultaneous access to one live workspace database on a shared drive, network location, or cloud folder as collaboration. Use copied exports, backups, or selected source files between machines, or choose a purpose-built sync or collaboration workflow.

Cloud sync across devices is planned without a published date. It is not available today.

Searching the library when a project starts

One of the highest-value moments for a reference library is the beginning of a new project, when you are looking for precedents across a collection built over years of practice.

refern's search accepts 21 user-visible search operator keywords, so you can construct precise queries:

  • tag:Civic tag:"Exposed Concrete" rating:>=4 surfaces highly rated civic precedents in exposed concrete
  • color:#8B7355 finds images dominated by that warm brown tone, useful for a material palette study
  • type:image tag:Detail narrows to detail photographs specifically
  • tag:"Nordic Climate" sort:added orders your Nordic-climate references chronologically

The color search by hex is particularly useful for material research: pick a color from a material sample or a paint chip and find every reference image in your library that shares that dominant hue.

Visual similarity search lets you start from one good precedent and find others in the library that look like it, catching images you tagged inconsistently or added before your taxonomy was developed.

The relationship graph for practices with deep archives

For architecture practices with large archives, the relationship graph view adds a layer of navigation that search alone does not provide.

refern's graph view renders all folders, images, canvas files, and the links between them as a navigable map. Over time, as you link related precedents (cross-reference links), track which images appear in which presentation canvases (placed-in-canvas links), and note which images were cropped from a source (derived-from links), the graph becomes a spatial representation of your research over the life of a practice.

This is similar in concept to what Obsidian offers for written notes, applied to visual references instead.

Honest gaps to consider

refern does not yet have cloud sync or a web or mobile app. If you need to share a reference board with a client online, or access your library from a phone on site, those features are not available today. Cloud sync and shareable canvases are planned without published dates.

refern does not have real-time collaboration. If your studio needs multiple architects to add to and search the same library simultaneously from different machines, that is a limitation today. Eagle has the same limitation; Are.na and Milanote handle collaboration better.

PDFs and CAD files such as .dwg, .dxf, .skp, and .step can be indexed by switching Documents and "3D and CAD" on under Edit workspace, "File types shown." PDFs receive a first-page preview and open in the native viewer. CAD files remain icon-only generic entries that are searchable by name, taggable, folderable, and draggable back out into your CAD app. Keep exporting PNG or JPEG sheets for CAD drawings you want to browse visually or put on a canvas. .rvt is not on the built-in list, but you can add it as a custom extension.

Common problems and fixes

Images are hard to find after tagging sessions. This usually means tags were applied inconsistently. Use a smart folder with a tag: operator to audit images by tag and spot gaps. Tag macros help prevent inconsistency going forward by bundling related tags into a single shortcut.

Canvas performance feels slow with large image counts. refern uses thumbnail proxies for canvas display, so file size of originals does not directly drive canvas performance. If the canvas is sluggish, check whether you have images from very high-resolution scans. The non-destructive crop can reduce the displayed region without touching the original file.

Library on site has outdated images. Before leaving the studio, copy the canvas exports or selected source files you need onto the laptop. Within one local workspace, refern's reconcile feature (Settings, Sync with disk) picks up images added or changed on that machine's disk; it is not cross-device sync.

Next steps

Frequently asked questions

What is the best reference image tool for architects?

refern works well for architects who need a local precedent library with search, tagging, and a canvas for material boards. For collaborative online boards shared with clients, Milanote is also worth considering. For research-community access, Are.na has a strong architecture following.

Can architects use refern offline on a construction site?

Yes. Core library, search, viewing, canvas, and study workflows run locally without uploading your existing library. License activation, update checks, optional model downloads, feedback, user-requested web downloads, and an unlicensed trusted-time check can use the network.

Does refern copy my reference images into its own folder?

Normal refern workspace setup indexes existing source files in place, so that setup does not relocate them. Explicit imports and file actions can create or change files.

How do architects organize precedent images in refern?

Use nested folders for typology (residential, civic, cultural), then apply hierarchical tags for material, structure type, climate, and architect. Smart folders surface images matching multiple criteria without duplicating files.

Is Are.na good for architectural research?

Are.na has a genuine architecture-school following and works well for online collaborative research and sharing with peers. It is web-only with no canvas, no local files, and no offline access, so it complements rather than replaces a local reference library.
  • $35 one-time, no subscription
  • Windows, macOS, Linux
  • Local-first and private
  • 10,000+ creatives
  • Community on Discord
“Organization and search like Eagle cool, canvas from PureRef.”
An early refern user

Try it yourself

One library for your references, with a canvas built in.

refern keeps your images organized and searchable, with an infinite canvas in the same app. Normal workspaces read existing files in place. $35 one-time, updates included.

No account required for the 30-day trial. Your normal files remain accessible on disk if the trial ends.

Sources

  1. 1.Eagle homepage, pricing, platform
  2. 2.Eagle: no Linux client
  3. 3.Eagle $34.95 one-time, 2 devices
  4. 4.Milanote pricing, $9.99/mo individual
  5. 5.Are.na: member count, pricing, team
  6. 6.Are.na channels: flat, no nested folders