Seedlabs

Recurring-Sitter Finder for Digitized Manuscripts

An image tool that finds the same person painted multiple times across different pages of a digital manuscript.

An image-analysis tool that scans a digitized illuminated manuscript, detects the human heads painted in its decorated initials, and clusters them so scholars can see when the same person appears again and again — the pattern that revealed the poet John Gower ten times hidden in a duke's prayer book.

Arts and HumanitiesMedieval Literature and History
Cultural-heritage digitization and manuscript cataloging (digital humanities research infrastructure)

Concept

A hosted image-analysis service for digitized illuminated manuscripts. A library or scholar points it at a manuscript (an IIIF manifest or a set of high-resolution page scans); the tool detects the human heads painted in decorated initials and margins, clusters them by visual similarity, and reports every head that recurs across the book, with side-by-side crops for confirmation. When an expert labels one instance — for example, an inscribed portrait that names its sitter — the label propagates across the cluster, producing a sitter index for the whole manuscript: who appears, how often, and at which text divisions. The single wedge is recurrence detection: one book, many appearances of the same face, surfaced automatically. The tool proposes clusters; naming the sitters stays with the scholar.

Why now

The paper shows why recurrence is the decisive clue: in the Duke of Bedford's Psalter-Hours, 290 of the 300 minor text divisions carry portrait heads, many of them repeated, and a single inscribed portrait of John Gower served as the key that let the scholar recognize him ten further times — as prophet, preacher, and penitent — revealing a concealed 'national portrait gallery' of Lancastrian friends and foes inside a prayer book. That discovery was made by hand, in one manuscript; the demonstrated payoff (repeated heads carrying a hidden political subtext) is exactly the pattern that clustering can now search for at scale across the tens of thousands of manuscripts libraries have already digitized.

Written by GLM · 10 Oct 2026

AI assessment

Backed by 1 paper89

A highly focused, evidence-backed tool for digital humanities that transforms a manual scholarly discovery process into a scalable image-clustering service.

Evidence strength
5/5
The idea directly operationalizes the specific finding from the provided text regarding the recurrence of portrait heads in the Bedford Psalter-Hours.
Market pull
3/5
The target users are prestigious academic institutions with clear needs, but the total addressable market is small and budget-constrained.
Novelty & moat
5/5
Unlike the listed catalog items which focus on text or provenance, this introduces a specific visual-clustering mechanism for art-historical analysis.
Feasibility
4/5
Modern face detection and clustering (e.g., using Siamese networks or CLIP) are mature technologies that can be applied to these high-res scans.
Wedge clarity
5/5
The 'recurrence detection' in a single book is a sharp, well-defined entry point that provides immediate value to a researcher.
Simplicity / focus
5/5
The product is a single-purpose tool with a clear input (IIIF manifest) and a clear output (clustered crops).

Scored by Gemma 4 · 10 Oct 2026. Rated against a fixed rubric (evidence, market, novelty, feasibility, wedge, simplicity), relative to the closest ideas already in the catalog. A prior estimate to compare ideas before real-world signal arrives.

Persona discussion

AI personas trained on real people's expertise debate this idea as it evolves.

View the discussion →

Act on this idea

Ideas only matter if someone runs with them. Your message goes straight to the founder's inbox — nothing is stored on our servers.

Who benefits

  • British Libraryorganization

    It holds Add. MS 42131 and has mass-digitized its manuscripts; a sitter-recurrence index would enrich its catalog records and give scholars a new reason to work with its digitized collections.

  • Index of Medieval Art, Princeton Universityorganization

    Its database indexes the subject matter of medieval art; clustered, labeled portrait heads from digitized manuscripts are exactly the structured iconographic records it collects and serves to researchers.

  • Fitzwilliam Museum, Cambridgeorganization

    Its curators manage a large collection of illuminated manuscripts and could run the tool over their own holdings to find repeated sitters and hidden portrait programs.

  • The Warburg Instituteorganization

    Its photographic collection and iconographic database support attribution and iconography research; recurrence clusters of portrait heads would feed directly into its research tools.

  • e-codices (Virtual Manuscript Library of Switzerland)organization

    It hosts thousands of digitized manuscripts; offering a recurring-sitter analysis layer would differentiate the platform for the medievalists who use it.

Research it builds on

  1. The author portraits in the Bedford Psalter-Hours: Gower, Chaucer and Hoccleve
    Wright, Sylvia · 2026 · 115 citations
    All ideas from this paper →
Leave feedback
feasibility