Finding a place: advancing digital methods to unlock the use of digitized book illustrations in cultural institutions

在旧书的封面之间隐藏着一个关于世界及其居民的非凡的视觉记录。书籍插图跨越几个世纪，代表了从威尔士到怀俄明州的所有地方。插图讲述了那些跨越时空创造和阅读它们的人的想法和信仰的故事。它们也是实物，在地球仪中流通，促成了跨国世界观，既有共同的，也有争议的。这些图像赋予了自然和建筑环境新的意义，并在描绘当地人和外国人、旅行者、土著人、殖民者和流离失所者时突出了人与地之间复杂的交叉点。这些例证对文化机构尤其重要，因为这些机构往往在各地点和社区进行大量投资。地点的概念与收藏品的语境化以及机构自我定义、体验和参与的方式错综复杂地交织在一起。然而，历史书籍插图很少在文化机构中使用或分析，即使插图的数字化收藏存在于图书馆中，也可以在网上访问。有几个因素导致了这种忽视：人们对插图的重要性认识有限，特别是因为许多插图已经消失了很长时间，被视为仅次于独特的艺术品;数字化插图的庞大规模和多样性使得找到相关图像并相互分析具有挑战性。简单地说，很难确定“在那里”有什么插图，即使找到了，也很难对它们做些什么。在插图书籍数字化之前，理论家热拉尔·热内特决定不研究插图，因为这个领域是一个“广阔的大陆”。正是这个“广阔的大陆”，该项目试图通过汇集插图、博物馆研究、在英国和美国的文化机构中，有着不同的需求、受众和收藏：特别收藏和档案馆、卡迪夫大学、美国遗产中心、怀俄明州大学、西部布法罗比尔中心、兰贝斯宫图书馆和国家信托。合作模式为我们的实践提供了信息，策展人提出的研究问题塑造了数字方法的发展，这些方法将揭示一个被遗忘的图像世界并揭示边缘化的历史。计算工具（如众包标签、标题捕获、地理解析）将用于从大英图书馆的核心数字化插图书籍收藏中确定一个重要的地方和人物书籍插图子集，从而丰富其搜索能力和相关性。为了回应文化机构的研究问题，这些插图将被放置在彼此之间以及与其他图像和藏品的对话中，挑战传统的分类，并使隐藏的历史得以揭示。计算机视觉，很少在这种情况下使用，将使我们能够询问图像的相似性和差异，有效地绘制视觉“刻板印象”（一个本身与插图印刷有关的术语）的构建，转换和颠覆。我们将第一次能够识别建筑物或景观插图的变化，或者追踪人物表现的相似性和差异性，这是一种比较分析，这在本书的物质形式中是不可能的。这个项目所揭示的历史本身以及用于揭示它们的数字方法都很重要。我们的目标是创造一个环境和基础设施的遗产，将产生一个更深入的了解的要求和数字方法的可能性，理解插图，并找到一个地方，这些图像在文化机构。

Hidden between the covers of old books lies a remarkable visual record of the world and its inhabitants. Book illustrations span centuries and represent everywhere from Wales to Wyoming. Illustrations tell stories about the ideas and beliefs of those who created and read them across time and space. They are also material objects, which circulated around the globe and contributed to transnational world views, both shared and contested. These images give new meanings to natural and built environments and foreground complex intersections between people and place in their depiction of the local and foreign, the traveller, the indigenous, colonized, and displaced. Illustrations are especially significant for cultural institutions, which are often heavily invested in locations and communities. A notion of place is intricately interwoven in the contextualization of collections and the ways in which institutions define themselves and are experienced and engaged with.Yet historical book illustrations are rarely used or analyzed in cultural institutions, even though digitized collections of illustrations exist in libraries and are accessible online. Several factors underlie this neglect: there is a limited awareness of the importance of illustrations, especially as many have been out of sight for so long and viewed as secondary to unique artistic objects; and the sheer scale and variety of digitized illustrations makes it challenging to find relevant images and to analyze them alongside each other. To put it simply, it is difficult to identify what illustrations are 'out there' and what can be done with them even if they are found. Prior to the digitization of illustrated books, the theorist, Gérard Genette, decided against research into illustrations because the field was an 'immense continent'.It is this 'immense continent' that the project seeks to explore by bringing together scholars in illustration, museum studies, and computer science in UK and US cultural institutions with different requirements, audiences, and collections: Special Collections and Archives, Cardiff University, the American Heritage Center, University of Wyoming, Buffalo Bill Center of the West, Lambeth Palace Library, and the National Trust. A collaborative model informs our practice, with the research questions posed by curators moulding the development of digital methods that will reveal a forgotten world of images and uncover marginalized histories. Computational tools (e.g. crowdsourced tags, caption capture, geoparsing) will be used to identify a significant subset of book illustrations of places and people from a core digitized collection of illustrated books from the British Library, thus enriching its searchability and relevance. In response to the research questions of the cultural institutions, these illustrations will then be placed in dialogue with each other and with other images and objects in collections, challenging conventional categorizations and enabling hidden histories to be revealed. Computer vision, which has rarely been used in this context, will allow us to interrogate iconographic similarities and divergences, effectively charting the construction, transformation, and subversion of visual 'stereotypes' (a term that is itself bound up in the printing of illustrations). We will for the first time be able to identify changes in illustrations of buildings or landscapes, or trace affinities and differences across representations of people, a comparative analysis that is impossible in the material form of the book.The histories uncovered on this project are significant in themselves and in the digital methods used to reveal them. Our objective is to create an environment and infrastructural legacy that will generate a deeper knowledge of the requirements and possibilities of digital methods for understanding illustrations, and to find a place for these images within cultural institutions.