How We Transformed CNRS Graphoskop into a Scalable, Browser-Based Web Application

Transforming a fragmented, legacy desktop plugin into a modern, highly scalable academic research platform capable of processing hundreds of manuscript images natively in the browser.

Read Case Study

Project Overview

Product

Graphoskop Web Platform

Industry

Academic Research

Services Provided

Technologies

React

Tailwind CSS

shadcn/ui

Docker

HTML Canvas

Nginx

Summary

IRHT – Institut de recherche et d’histoire des textes needed to modernize Graphoskop, a legacy Java plugin originally built as an extension of ImageJ for analyzing medieval manuscript images. What began as a fragile, desktop-bound tool used for manual geometric and text measurements has now evolved into a robust web application. Integrated seamlessly into the ARCA MSS visualization platform, the modernized Graphoskop allows distributed research teams to seamlessly analyze up to 200 manuscript images per session directly from their web browsers.

The Challenge

The legacy Graphoskop plugin was highly functional but operationally fragmented. Researchers relied on it to measure critical historical metrics such as line spacing, margins, letter angles, and writing surface density. However, utilizing the tool created major technical bottlenecks.

To conduct an analysis, researchers were forced to:

Install complex Java dependencies manually.
Configure local ImageJ and OpenOffice environments.
Upload and manage large manuscript images locally.
Perform measurements entirely by hand.
Manually export all results into offline spreadsheets.

This localized, fragmented workflow resulted in severe pain points:

Operational Inefficiency: Researchers spent more time setting up IT environments than conducting actual scientific analysis.
Scalability Limitations: The plugin frequently crashed and was not built to handle large image batches (up to 200 images per session).
Accessibility Constraints: As a desktop-bound tool, it was incredibly difficult for distributed, non-technical academic teams to collaborate.
Maintainability Risks: Developed by a former student, the original codebase lacked documentation and long-term architectural scalability.

The legacy Graphoskop plugin was functional but highly fragmented. Researchers spent significant time setting up tools instead of conducting analysis. Desktop interactions didn’t translate well to distributed academic teams, and the tool simply wasn't built for handling large image batches.

IRHT – Institut de recherche et d'histoire des textes (CNRS)

The Objectives

To eliminate these barriers, we targeted four primary goals:

Modernize Legacy Software: Convert the outdated Java/ImageJ desktop plugin into a seamless browser-based application.
Improve Accessibility: Remove the need for local installations so researchers can universally access the platform.
Maintain Scientific Accuracy: Ensure all measurement logic, algorithms, and statistical outputs remain precisely faithful to legacy benchmarks.
Improve Scalability: Build an architecture capable of supporting batch processing natively within the browser.

The Solution: A 3-Phase Transformation

We approached the modernization systematically to ensure scientific precision, eliminate risks, and drastically improve usability.

Phase 1: Product Discovery & Scope Definition

With no formal specifications available for the legacy code, our initial focus was deep discovery to prevent scope creep.

Activities: We audited the legacy Java plugin, reverse-engineered its hidden workflows, mathematically mapped its measurement capabilities, and conducted requirement workshops with CNRS researchers.
Deliverables: We produced a comprehensive functional specification document, a prioritized feature roadmap, budget estimation, and sprint planning structures to remove all ambiguity before development began.

Phase 2: Web Product Development

We wholly translated clunky legacy desktop interactions into fluid, browser-native functionalities:

Image Upload System: Engineered to handle bulk uploads and massive, high-resolution manuscript image files without crashing.
Calibration Tool: Developed a precise feature allowing researchers to calibrate the image scale based on real-world manuscript dimensions.
10 Specialized Measurement Tools: Successfully replicated and modernized legacy capabilities, including line spacing, margin measurement, word spacing, writing angle, surface density, generic measurements, curvature calculations, and automated statistical analysis.

Phase 3: Modern Architecture Implementation

To guarantee high performance and future scalability, we implemented a robust modern technology stack:

Frontend Ecosystem: Built with React 19, TypeScript, Vite, Tailwind CSS, and shadcn/ui.
State Management: React Context API for streamlined cross-component data handling.
Image Rendering: Deployed HTML Canvas integration to overcome browser lag, drastically improving rendering speed for large image batches.
Infrastructure & Export: Set up a secure deployment architecture via Docker and Nginx, integrated with automated Excel export support to eliminate manual spreadsheet data entry.

Check Out the Website

Overcoming Key Challenges

Legacy Code Understanding: Because the original plugin lacked documentation, we meticulously reverse-engineered the workflows, mapped out the measurement logic, and created robust internal documentation from scratch.
Preserving Scientific Accuracy: Minor algorithmic deviations could invalidate real-world historical research. We carefully rebuilt the algorithms, conducted repeated validation tests, and compared outputs directly against the legacy plugin to ensure 100% precision.
Large Image Processing: Rendering hundreds of high-res images natively in a browser typically causes severe lag. By optimizing frontend rendering logic and leveraging HTML Canvas, we successfully achieved smooth batch processing.
Web UX Modernization: We redesigned complex desktop user interactions into intuitive web workflows, making onboarding radically easier for non-technical researchers.

Impact & Results

The Graphoskop web transformation successfully balanced accuracy, scalability, and usability, vastly improving the day-to-day workflow for academic researchers.

Metrics of Success:

User Onboarding Time:
↓ 80% reduction (frictionless, zero-setup onboarding)
Average Processing Time:
↓ 50% faster turnaround due to automated exports and streamlined workflows
Measurement Accuracy:
≥ 100% match to legacy benchmarks
Export Completion Rate:
> 95%
Platform Uptime:
> 99%
Adoption Rate:
Consistent Month-over-Month (MoM) growth across CNRS teams.

Key Learnings

Discovery Matters Most in Legacy Modernization: Skipping the discovery phase would have triggered major scope risks given the lack of documentation.
Precision Over Speed: Academic software requires rigorous validation—speed to market is secondary to scientific accuracy.
UX is a Major Value Driver: The biggest transformation wasn’t merely technical; drastically improving the user experience was the defining win for researchers.
Incremental Migration Reduces Risk: Rebuilding the platform feature-by-feature isolated issues and minimized deployment failures.

Future Opportunities

With a future-proof React architecture now firmly in place, the platform is ready to scale with powerful new features:

AI-Assisted Manuscript Detection: Automatically detecting manuscript boundaries and outlining critical measurement areas using AI.
OCR Integration: Extracting deep historical text automatically for richer analysis.
Collaboration Layer: Allowing multiple global researchers to annotate the same manuscripts simultaneously.
Cloud Storage & Analytics: Centrally storing historical datasets to run complex trend analysis across massive manuscript collections.

Conclusion

This project serves as a masterclass in legacy modernization and scientific SaaS transformation. What began as an isolated, fragile desktop tool has successfully evolved into a highly accessible, scalable research platform for CNRS. By meticulously balancing scientific accuracy, scalability, and usability, this project proves that legacy academic software can be modernized into premium digital products without ever losing their domain-specific precision.

Ready to Modernize Your Legacy Systems?

Is legacy software slowing down your workflow? Riseup Labs creates custom solutions that scale operations, improve accessibility, and modernize UX, all while preserving your data and precision.

Talk to Our Experts