Computational Study of Light in Paintings

José Ballesteros Zapata

Example of the methodology applied to the second part of the project, using contours to analyze the lighting signature of the Candlelight Master. On the right, contours are highlighted in different colors. In yellow, the probability distribution of each light direction estimation. The heat map illustrates the Maximum Likelihood Estimation for the light source location. We use probability areas as lighting signatures for artists. The red cross indicates the estimated light source location, near the depicted candlelight. Painting: Vanitas, Candlelight Master (1620–1635), Palazzo Barberini, Rome. Source: Bibliotheca Hertziana

Example of the methodology applied to the second part of the project, using contours to analyze the lighting signature of the Candlelight Master. On the right, contours are highlighted in different colors. In yellow, the probability distribution of each light direction estimation. The heat map illustrates the Maximum Likelihood Estimation for the light source location. We use probability areas as lighting signatures for artists. The red cross indicates the estimated light source location, near the depicted candlelight. Painting: Vanitas, Candlelight Master (1620–1635), Palazzo Barberini, Rome. Source: Bibliotheca Hertziana

Art historians have preferred to emphasize perspective as the major Renaissance achievement rather than light because perspective is more easily defined. Due to its abstract nature, and unlike perspective, accident and intention are not easy to distinguish when describing light with language.
The first part of the project uses computer-graphics inspired inverse rendering pipelines to extract light environments (e.g., direction, intensity, diffuseness) from paintings. This model is used to understand the light rendering conventions in time, for example, by comparing the diffuseness in Rembrandt’s oeuvre with the rest of the 17th-century Netherlandish school. By rendering 3D models of Leonardo’s light theories, we can compare his theoretical understanding of light with the light environments he painted. This approach highlights conflicts between theory and practice.
The second part of the project uses lighting estimation techniques, heavily inspired by image tampering methods, to understand the lighting congruence in candlelight paintings. The primary hypothesis is that we can use lighting information to model intrinsic features that reveal unique artists’ techniques. Ideally, this method serves as another source of information to distinguish one candlelight artist from another. By examining the spatial arrangement of contours – the edges that define shapes within the artwork – we can infer probability maps for the light source location. This work uses such probability maps as the lighting signatures of artists.
The study of candlelight paintings is motivated by an attribution of approximately 60 unsigned Caravaggesque paintings to the enigmatic artist that the art historian Benedict Nicolson called The Candlelight Master due to the recurring presence of candlelight in the artworks. The Candlelight Master seems to be an uncanny character without a clear identity, believed to have been active in Rome between 1620 and 1634. These paintings are now displayed in museums around the world under the attribution of different names, like Honthorst, Trophime Bigot, Maestro Jacomo, alongside the Candlelight Master.
To this day, the attribution of this oeuvre is still unclear. The absence of historical records and documentary evidence has compelled art historians to use visual evidence on carefully selected paintings to establish relationships between artists. However, human perception is inherently limited in discerning subtle nuances in lighting environments – such as cast shadows, lighting direction, or contrast. Interestingly, candlelight emerges as the unique visual feature consistently present across these works, serving as a distinct artistic tool employed by these painters. This scenario presents itself as the perfect candidate to test the potential of computational methods to augment the analysis of these artworks by uncovering hidden information imperceptible to the naked eye. The consistent presence of candlelight as a central element across the oeuvre suggests a deliberate artistic choice that can be quantitatively analyzed. The proposed method in this research aims to contribute to a deeper understanding of the artist’s workshop techniques and influences.