refern vs Are.na: Connected, but Local and Owned (2026)

Are.na is a cloud-based curation tool built around slow, intentional collecting with no algorithm and no ads. refern is a local-first desktop app for organizing images you already own, with an infinite canvas, full-text search, and an explicit relationship graph. If you want community and public sharing, Are.na wins. If you want to own your data, work offline, and build a searchable canvas-ready library, refern is built for that.

Quick verdict

Introduction: two different philosophies about where your references live

Are.na and refern are both used by designers and researchers who want more from their reference workflow than a generic folder system. The similarity ends there.

Are.na is built around a cloud community. Every block (image, link, text, PDF) lives on Are.na's servers. The defining feature is that a single block can live in multiple channels at once, creating implicit cross-referencing without any tagging system. The platform is deliberately slow, anti-algorithmic, and community-oriented. It has been running since around 2011 and has cultivated a specific creative-intellectual audience.

refern is built around a folder you already own. It runs entirely on your computer, indexes your existing files without moving them, and gives you search, canvas, and graph tools that work whether or not you have an internet connection.

The practical question is which problem you are solving. If you need to collaborate with teammates, share research publicly with clients, or tap into a community of other curators, Are.na is the stronger choice today. If you need to organize a large existing library of local images, build spatial mood boards, or find references by color and visual similarity, refern is built for that.

What is refern?

refern is a desktop reference manager for artists that combines Eagle-style organization with a PureRef-style infinite canvas and an Obsidian-style relationship graph. It costs $35 one time, runs on Windows, macOS, and Linux, and indexes existing folders in place.

During normal workspace indexing, refern leaves existing files in place and builds a local SQLite database and thumbnail cache. It reads embedded EXIF/IPTC/XMP metadata and provides full-text search, color search, visual similarity, and 21 user-visible filter operators. Explicit imports and capture actions can create files. An infinite canvas lets you compose references spatially, and a relationship graph shows how folders, images, canvases, and groups connect.

It costs $35 one time. One license covers up to 3 devices, includes commercial use, and comes with a 30-day free trial. No account required. No subscription. Your files stay where they are.

What is Are.na?

Are.na is a web platform for saving and connecting content into collections called channels. Images, links, text snippets, PDFs, and videos are saved as "blocks." The organizing principle is that a single block can belong to multiple channels simultaneously, creating a web of implicit connections across your collections. [1]

The platform is deliberately anti-algorithmic: no likes, follower counts, recommendation engine, or ads. That community model is a genuine advantage for people who want public, connected curation rather than a private desktop archive.

Are.na's community is a genuine strength. Designers, architects, academics, and artists use it as a public research and curation space. That social context is outside refern's current local-library scope.

Organization and search

Are.na organizes content into flat channels. There is no nesting, no folder hierarchy, and no subchannels. A user with hundreds of channels navigates by memory and manual browsing. Search covers channel titles and usernames but does not search the content of blocks. [13] If you have saved a thousand reference images across fifty channels, there is no way to find a specific one by color, content description, or metadata.

refern uses nested folders with unlimited depth, hierarchical tags, color labels, ratings, favorites, source URLs, and custom notes. Search uses SQLite FTS5 across file names, descriptions, notes, tags, source, and creator fields, with 21 user-visible operators (for example type:image, rating:>=3, tag:landscape, color:#ff6b35, is:duplicate). Color and visual-similarity search run locally without a search API.

Verdict: refern provides deeper local-library search; Are.na's connected channel model is stronger for public, collaborative curation. No comparative scale number is published until the benchmark protocol is complete.

Canvas and spatial moodboarding

Are.na does not have a canvas. Channels display blocks in a chronological or manually reordered grid list. There is no way to arrange images freely in space, annotate them, draw over them, or combine them with text and shapes. [12]

refern includes an infinite canvas with layers and groups, text elements, 9 shape types, freehand drawing, color swatches, non-destructive image crop, image filters, and a "find similar" tool on canvas images. The canvas supports pin-window-on-top, window transparency, and mouse clickthrough, which replicates the PureRef overlay use case. Every layer is optionally backgrounded, making it suitable for structured mood boards and open-ended spatial composition. Canvases are saved as .refern-canvas files in your folder alongside your images.

Verdict: refern wins. Are.na does not attempt a canvas. If spatial moodboarding is part of your workflow, Are.na requires a separate tool.

Relationships and graph view

Are.na's connection model is one of its most distinctive features. A block living in multiple channels simultaneously creates an implicit web of cross-referencing: if you put a typeface image in a "Swiss design" channel and a "brutalism" channel, those channels are implicitly connected through that shared block. The "connection sidebar" on any block shows every channel it appears in. This is a genuine form of linked thinking, and it is built into the architecture rather than bolted on. There is no visual graph render, but the co-membership mechanic creates serendipitous discovery that no folder-based tool matches. [2]

refern uses explicit typed entity links. Every link has a kind: cross-reference (you manually linked two images), derived-from (a crop came from an original), placed-in-canvas (an image appears in a canvas file). A "Linked References" sidebar on any image shows everything connected to it. A full-screen relationship graph renders folders, images, canvases, groups, and the links between them as a navigable node-edge graph.

Verdict: these are genuinely different models. Are.na's implicit connections emerge naturally from how you organize. refern's explicit typed links are more deliberate and offer a clearer map of intent. An artist who wants serendipitous discovery across a community benefits from Are.na's model. An artist who wants to trace provenance (this crop came from that scan), composition (this image appears in three canvases), and cross-reference (these two images inform each other) will find refern's explicit graph more useful. Both are honest answers to "how do I understand how my references relate?"

Data ownership and offline access

Are.na is a hosted service built around uploaded or linked blocks. It does not index an arbitrary local folder as a desktop library. Are.na documents channel exports as PDF, ZIP, and HTML, which provide a migration and backup path; test whether the chosen format preserves the fields you need.

refern keeps its core library, index, thumbnails, and canvases local. Normal indexing leaves existing media in place, while explicit imports can create files. The core library works without an account; activation, trusted-time checks, updates, optional model downloads, feedback, and web capture use the network. After 30 days, the app is gated by activation while normal files and data remain accessible on disk.

Verdict: refern wins for users who prioritize ownership. Are.na wins for users who want cloud access from any device.

Pricing

Are.na raised prices in January 2023 (from $5/month and $45/year to $7/month and $70/year). [3] The 200-block free tier is easy to exhaust for active reference collectors. A student or educator rate of $3.50/month or $35/year is available on request. [3]

For long-term use, the math favors refern at $35 one-time against $70/year recurring. At 13 months Are.na Premium costs more than refern in perpetuity. For users who need Are.na's collaboration and community features, the subscription is the cost of access to something refern does not currently offer.

Full feature comparison

Who should choose refern

Choose refern if you already have a folder of reference images on your computer and want to search, organize, and compose with them without re-uploading everything to a cloud service. It is the right tool if you:

• Work with a large local library (hundreds to tens of thousands of images) and need fast search by text, color, or visual similarity.
• Want to build spatial mood boards on an infinite canvas without switching to a separate tool like PureRef.
• Value data ownership: your files stay in your folder, the app costs $35 once, and there is no ongoing subscription.
• Work offline regularly, travel frequently, or operate in environments without reliable internet.
• Import from Eagle and want to carry over folders, tags, ratings, and sources.
• Want to understand how your references relate through explicit typed links and a visual graph.
• Use PureRef or Eagle today and want more: better search, graph view, and browser extension capture in one tool.

Who should choose Are.na

Are.na is the right choice if:

• Collaboration is core to your workflow. You need teammates to add to shared collections, or you run a research project with multiple contributors. Are.na's multi-user channels and group workspaces have no current equivalent in refern. [12]
• You want to share collections publicly with clients or collaborators without requiring them to create an account. Are.na's public shareable channel URLs cover this cleanly.
• You are saving links and web content rather than managing local files. Are.na handles YouTube embeds, text snippets, and links as first-class blocks alongside images.
• You value the Are.na community itself. Browsing how designers, architects, and researchers curate publicly is a genuine source of inspiration that refern, as a local tool, cannot replicate.
• You need iOS or Android access to capture and review references on mobile devices. refern is desktop-only.
• You prefer a web tool with no installation. Are.na requires no download.
• You are philosophically drawn to Are.na's independent, member-supported, no-algorithm ethos. The platform's culture is a real product feature, not a marketing line.

Switching from Are.na to refern

Are.na allows channel downloads as PDF, ZIP, and HTML from channel settings, and channels can also be retrieved through its public REST API. [13][14]

refern does not have a direct Are.na import path. The typical migration looks like this:

1. Download your Are.na channels (via the export option or API).
2. Organize the downloaded files into a folder structure that reflects your channels.
3. Point refern at that folder. refern indexes the files in place, reads any embedded EXIF/IPTC/XMP metadata, and builds thumbnails.
4. Create tags and tag hierarchies in refern to recreate the organizational layer you built in Are.na.

During normal workspace indexing, refern leaves the exported files in place. Explicit import actions can copy, so confirm which workflow you use and keep the Are.na export until you verify the result.

If you currently use both Are.na and another local tool (Eagle, PureRef), the refern Eagle importer reads folders, tags, ratings, source URLs, and notes directly.

The verdict

Are.na and refern are not direct substitutes. Are.na is a community and a philosophy as much as it is a tool. It has been built carefully by a small team for over a decade, it has a track record with institutions and universities, and the implicit-connection model genuinely surfaces unexpected relationships across a curated collection. For collaborative and public-facing research, Are.na is hard to beat.

refern is for the artist or designer who has a folder of thousands of reference images and wants to actually find things in it. It is for the person who builds mood boards and wants spatial composition, not just a grid list. It is for the person who has paid for three software subscriptions this year and would rather pay $35 once and own their tool.

Many serious creative researchers use both: Are.na for public and collaborative research, refern for the private local library they actually compose from. If you are deciding between them based on one use case, the table above shows clearly which tool wins on each dimension.