Interview : 4 questions to Barthélémy JOBERT, Delacroix Numérique project manager

Date
26 Feb 2025
Authors
Fondation Sorbonne Université
Open link

"The Sorbonne University Foundation warmly thanks Schmidt Sciences for its generous support of the Delacroix Numérique project, one of Sorbonne University's major initiatives at the crossroads of Art History and Artificial Intelligence."

1. You are the initiator of the “Digital Delacroix” project. How did this project come about?

The “Digital Delacroix” project was born out of my desire to revisit the work of Eugène Delacroix using the modern tools of Artificial Intelligence and digital humanities. Gaëlle LAFAGE, a research engineer at the Centre André Chastel, and I have been working on the project for a number of years now.

Thanks to the digitization of Delacroix's writings and the criticism of his work carried out with the Observatoire des textes, des idées et des corpus (ObTIC), we have already been able to carry out an in-depth and pertinent analysis of the texts, which we now need to secure.

At the same time, we have been working with SCAI, Sorbonne Center For Artificial Intelligence, whose collaboration is essential in exploring the use of high-quality digital photographic coverage in two directions. On the one hand, the aim is to produce a virtual reconstruction of Delacroix's major sets, and on the other, to set up an algorithmic analysis to distinguish Delacroix's own elements from those created by his collaborators. The SCAI appointed two of its engineers to assist us: Nicolas Leys, visualization engineer, and Baptiste Gregorutti, data scientist, to coordinate the AI aspect stricto sensu.

At present, the library of the Assemblée Nationale, one of Delacroix's most complex and ambitious decorative works, is being used as a testing ground. This initial analysis will enable us to extend our study to other works of this type (the Senate, the church of Saint-Sulpice, the Louvre) and even to his large-format paintings, for which we have already obtained some promising results.

We are also relying on the skills of LAMS (Laboratoire d'archéologie moléculaire et structurale de Sorbonne Université) which, thanks to a physico-chemical study of the materiality of the works, will be able to help define whether Eugène Delacroix or one of his disciples was the craftsman behind this or that part of the decor of the Assemblée Nationale library.

2. What is the aim of this digital work on Delacroix and his work?

Our main aim is to make Delacroix's work even more accessible to the general public. During the restoration of the decorations in the Saint-Sulpice church, to which I contributed as a member of the scientific committee, one conviction became clear: that of facilitating access to these magnificent decorations, given that visitors don't always have the opportunity to observe them up close. Thus was born the idea of creating virtual reconstructions of the sets painted by Delacroix.

Recently, the decorations in the Bibliothèque de l'Assemblée Nationale have undergone a remarkable restoration. The scaffolding built for the occasion enabled us to collect tens of thousands of photos at low cost, and to design a virtual tour model following a set path. It's not the one we'd like to end up with, as we'd prefer visitors to be able to walk around on their own, but the result is already quite impressive.

 "Through the digital analysis of texts, paintings, drawings and even materials, we hope to improve our understanding of Eugène Delacroix's creative process."

3. In fact, this project, Delacroix Numérique, also includes the idea of a virtual reconstitution of another set: that of the Salon de la Paix at the Hôtel de Ville in Paris. How can this be done, given that the set has disappeared?

The approach is a little different here, since we'll be working with generative AI. Based on the artist's preparatory work, engravings and rare period photographs or existing drawings, the aim is to reconstruct in 3D the décor of the Salon de la Paix at the Hôtel de Ville, which went up in smoke in the fire of 1871. It's impossible at this stage to say whether this will actually be feasible, but we have the skills and tools to work on it.

4. In early February, at the AI Summit in Paris, Sorbonne Université announced SCHMIDT SCIENCES' exceptional support for the project. Why is this sponsorship important?

This sponsorship is crucial because it enables us to pursue the project under the best possible conditions. It's a real boost!

Although its origins lie in art history, this project brings together a wide range of skills. This multidisciplinary approach makes it a unique initiative. The prospects are immense, as they have never been explored before. Our aim is to become the first in the world to establish an innovative working methodology, using AI, to identify the different hands that contributed to the execution of a decor or painting.

Schmidt Sciences, developing a philanthropic focus on Artificial Intelligence and Heritage through the Humanities and AI Virtual Institute (HAVI), confirms our ability to mobilize the best experts in both fields.

This support reinforces the project's legitimacy and enables us to recruit post-doctoral researchers in AI and art history, while securing the resources needed to deepen our research. It will accelerate the validation of our hypotheses and the availability of concrete, innovative results for the public.

 "Through this philanthropic support, the Sorbonne University Foundation is delighted to be able to help accelerate ambitious multi-disciplinary projects at the heart of the university's strategy."