1. Introduction

Today, many institutions provide digital image data for their collections. Easy access to high-quality images not only improves efficiency in art history research but also changes how research is conducted. Our approach to a style-comparative study makes use of this trend with a web-based tool called the “IIIF Curation Viewer,” built using IIIF (International Image Interoperability Framework), to change the input and output of research.

We studied pictorial manuscripts called “Emaki,” “Eiribon,” or “Nara Ehon” (illustrated scrolls and books with calligraphy) from the Edo period in Japan through the IIIF Curation Viewer, then discussed the efficiency and shareability of this approach.

2. Tools and materials

Composing lists of notable elements from target materials is a fundamental step in style comparison in art history research. The IIIF Curation viewer, developed by the Center for Open Data in the Humanities (CODH), is a useful tool for IIIF-compliant image resources. It has a function called “curation” that creates a list of interesting canvases with metadata. It reduces the effort of using cut and paste for the target material [Figure 1]. The result of cutting and pasting can easily be saved and shared in a JSON format.

[Figure 1] Selecting element by mouse drag operation

The “selected thumbnails” function shows a list of 20 curated elements at a time. This function is useful for comparing small details [Figure 2].

[Figure 2] Example of the “Selected thumbnail” mode and list of facial expressions

3. Analysis with the IIIF Curation Viewer

We picked up all facial expressions from four Eiribon and compared lists of the facial expressions using the IIIF Curation Viewer. Comparison suggests that pictures in each Eiribon were painted by different painters, but the same calligrapher wrote the texts. It also suggests that these Eiribon were created by a workgroup of artists.

We further analyzed using the IIIF Curation Viewer by comparing the above-mentioned curation with other Eiribon created by anonymous painters and calligraphers. We found that two anonymous works have the same drawing style as pictures in Asakura’s Eiribon [Figure 3].　

[Figure 3] Comparing facial expressions in Eiribon: Asakura’s text (above) and an anonymous work (below)

We found that our approach was useful for both the style comparing process and the sharing process. It is helpful to share pictures as evidence that supports the conclusion of a paper, but many journals did not allow it because of space limitations. Sharing the curation and citing it from the paper can solve this problem. For example, the evidence used in this paper is accessible at the CODH web site, as shown in the reference, so that other researchers can easily verify the results. Curated data can be increasingly promoted, due to its shareablity and reusability, by publishing them in repositories with persistent identifiers.

4. Conclusion

The IIIF Curation Viewer is important not only for making the entrance process easy through its cut and paste function, but also for making the output process useful through its sharing function. Both insertion and output is useful to art history research, in particular, in Eiribon research. There are many remaining unexamined Eiribon ; each Eiribon has many facial expressions. An easy cut, paste and share tool has been long awaited, and we hope it will enable the creation of a comprehensive facial expression database of Eiribon and Emaki.

We focused on Japanese art in this paper, but we can use this tool for any artwork as long as the images are served in IIIF. For example, we picked up facial expressions from portraits in the Yale Center for British Art and grouped them by century. The increased reusability of research extends possibilities for art historians terms of education and machine learning. Curated data can be reused as training data for machine learning.

Two issues remain for futures study. First, we need to increase IIIF-compliant image services. Especially in Japan, few institutions provide digital images in IIIF. Second, we need an ecosystem for sharing the results of curation, such as correcting metadata, identifier, and a repository for sharing and editing.