Methodology

This page explains the working process that led us to publish the digital catalogue raisonné: the software we chose, the recovery of the data, the digitization of the images, the design of the information structure, and the launch of the public layer. It is not meant as a technical manual, but as a record of the path we followed.

Cataloguing software

We based the whole project on CollectiveAccess, an open-source platform designed for cultural inventories. We work with two faces of the same system: Providence, the management environment where we catalogue, organize metadata, media and relationships; and Pawtucket, the layer through which the public consults the catalogue on the web.

We discarded generic publishing solutions (an expanded spreadsheet or a conventional content management system) because we needed a model able to support diverse work types, stable identifiers, high-resolution images, dates, locations and links between records — and to grow over time without losing coherence.

Database

We did not invent the starting data from scratch: they come from the database Montserrat Farré built for the 1993 book, recovered from the original floppy disks. The story of that recovery — obsolete drives, a shipment to New Zealand and the return of a work-by-work spreadsheet — can be found in About this catalogue. Here we only note that that Excel file was the starting point for feeding CollectiveAccess: fields had to be cleaned, normalized and completed before rows could become catalogue records.

Digitization

In parallel, we digitized more than two thousand colour transparencies and negatives — the photographic record of the known work made between 1991 and 1992, numbered and catalogued. Capture was done with digital macro photography: a Fujifilm X-T2 camera with a macro lens, mounted on an arm over a light table, to obtain faithful, high-resolution files without losing the detail of the analogue original.

Each image has been identified and linked to the corresponding record, so that the catalogue entry and the visual representation move forward together.

We must say, however, that colour management of the works may not be entirely accurate relative to the physical pieces. The entire digitization process was carried out with non-professional means, and that can introduce differences in tone, saturation or contrast compared with the original. The catalogue images are therefore as faithful an approximation to the original as possible, not a laboratory-calibrated substitute.

Metadata structure

In Providence we designed the catalogue data structure: object types, custom fields (titles, dating, location, identifiers), related entities and the documentation that surrounds the work. This model defines what is required, what is optional and how relationships between records are represented — work and context, work and source, architectural space, and so on — before we even thought about the public presentation.

A central axis has been the thematic hierarchy of the works. We did not want an extensive or overly academic taxonomy, but a functional tree, closely centred on the subject of the works, so that filtering and exploring remain clear. Alongside it we structured other search dimensions — technique, support, decade and work type, among others — so the catalogue can be refined from different criteria without useless complexity.

Metadata work is not cosmetic: it is what makes it possible to search, filter, relate and keep the catalogue coherent as we add new material.

Ingest and linking

Once the model was stable, we loaded the records into CollectiveAccess, associated each work with its digitized images, checked for duplicates and filled obvious gaps. It is the least visible part of the project and, at the same time, what turns a tabular inventory and an image archive into a consultable catalogue.

Web publication

All catalogue and collection management runs through Providence. On the public side, we carried out an extensive custom adaptation of Pawtucket: the theme, the navigation and, above all, the catalogue exploration pages and work detail pages differ significantly from the default implementation, so that they match how we want to present Farré’s work.

We devoted much attention to adjustments that ease navigation: a thumbnail viewer designed for fluid browsing of the catalogue; related-works blocks on the record pages; and memory of view and position when going back or forward in the browser, so the thread of the consultation is not lost. The aim has been for inventory and public experience to move together, without turning the website into a shop window detached from cataloguing.

In developing this layer we were helped by code-assistance tools — among them Cursor and other artificial intelligences — always as support for human work in design, implementation and review.

Access, publication and continuity

Not everything in Providence is published automatically in Pawtucket. We have defined access criteria so that only the records and media that should be published appear on the web, respecting rights and each item’s review status. In addition, we foresee that academics and scholars may access Providence — upon prior request and approval — to consult unpublished data and to collaborate with technical contributions on the work, the frescoes and other aspects of the catalogue.

The catalogue is a living space: we correct, re-digitize when needed, expand the model and incorporate works or documentation that were not there yet. The methodology described on this page is therefore also a commitment to maintenance.

This entire process has been carried out and funded entirely privately, without the involvement of any institution. It is a project undertaken for the pleasure of making Miquel Farré’s work available to everyone. We began in April 2024 and put it online in September 2026: almost two and a half years of work.

If you have come this far, you already know a little of the workshop behind the catalogue. We now invite you to open it and let the works — and the path you choose to explore them — tell you the rest.