Cinergie – Il cinema e le altre arti. N.20 (2021), 45–68
ISSN 2280-9481

From Edge to Edge: The Restoration of La battaglia dall’Astico al Piave (1918) and the Search for a Digital Historical-Critical Infrastructure

Simone VenturiniUniversity of Udine (Italy)

Professor at the University of Udine. He is the International Film and Media Studies Conference and Magis Spring School scientific coordinator. He is the Director of the Udine’s MA “Scienze del patrimonio audiovisivo e dell’educazione ai media / International Master in Cinema Studies (IMACS).”  He co-founded the La Camera Ottica Film and Video Restoration Lab. He coordinates several research projects and he deals with history and theory of film archives, film preservation and restoration, media archaeology, technological, cultural and production history of Italian cinema. He is part of the scientific committee of L’Avventura – International Film and Media Studies Journal and of the steering committee of Immagine. He is scientific director of the Plexus Book Series. He publish for Springer, Berghahn, Amsterdam University Press, Carocci, Il Castoro, Marsilio, and in Journals such as Journal of Film Preservation, Cinéma & Cie, Bianco e Nero, Cinergie.

Submitted: 2021-09-11 – Accepted: 2021-11-16 – Published: 2021-12-20


The restoration of La battaglia dall’Astico al Piave (Italy, 1918) has been funded by MiC (the Italian Ministry of Culture) and carried out by the University of Udine in collaboration with several film archives. Starting from a historical-philological and restoration framework, the case study offers some reflections and considerations between the fields of the digital humanities, film preservation, and cinema history. Furthermore, it provides a concrete opportunity to achieve the two goals of raising awareness about the status of digitized film artefacts, framing them as the result of modelling practices, and documenting the film preservation process as well as the surrounding historical and cultural network in a digital historical-critical infrastructure.

Keywords: First World War; Cinema History; Film Restoration; Film Philology; Digital Infrastructure.

1 Introduction

This paper is based on the restoration of La battaglia dall’Astico al Piave (Italy, 1918), funded by MiC (the Italian Ministry of Culture) and carried out by the University of Udine in collaboration with Kinoatelje, La Cineteca del Friuli, Istituto LUCE, Cineteca Milano, Museo Nazionale del Cinema, and Lobster Films.1 Starting from a historical-philological and restoration framework, the case study offers some reflections and considerations between the fields of the digital humanities, film preservation, and cinema history.2

The contribution has two main objectives. On one hand, I will frame the digital representations of film artefacts as the result of modelling practices, in other words, as heuristic processes that construct representations of a specific object or phenomenon (McCarty 2005). Historical and archaeological research on cinema history and film artefacts and the growing field of digital film studies and digital humanities stress the need for specific sub-disciplines. The science of film preservation and film historiography should focus on, and thus become fully aware of, the existence of a kind of film artefact imaging field. This would echo what has happened since the outset in several other disciplinary fields in terms of imaging sciences and the visualization of scientific objects (Daston and Galison 2007; Rheinberger 2010; Olsen et al. 2012; Drucker 2014), namely, constructed and mediated scientific observation for analytical-experimental, archival-documentary, and educational-pedagogical ends. In order to critically frame digital surrogates of film artefacts, their preservation and restoration quality needs to be assessed (Barricelli et al. 2020) along with documentation of the interdisciplinary research framework, methods and choices adopted. What is more, to consciously investigate film history and material culture at a “distance” (Salber Phillips 2013), film and media studies scholars working in digital and data-driven research environments should acquire a critical and educational approach to the modelling concepts and cultural techniques behind the visual representation of archival material artefacts. This approach should also be adopted towards digitized sources and the related digital tools and interfaces.

The second aim is to reload and reframe the longstanding field of digital critical editions of films made for research, education, and presentation purposes. Indeed, our research project will include a second phase in which we will design and create a web-based, historical-critical digital environment. The goal is to document the archival processes, giving a wider account of the restoration process, and of the material, visual, and cultural history of the film as a set of intertwined sources, practices, and discourses scattered among different and often unrelated archives.

Hence, the restoration of La battaglia dell’Astico al Piave provides a concrete opportunity to achieve the two goals of raising awareness about the status of digitized film artefacts and documenting the film preservation process as well as the surrounding historical and cultural network in a digital edition.

2 A Battle Diary. The Restoration of La Battaglia dall’Astico al Piave (1918)

La battaglia dall’Astico al Piave3 was made by the Italian Royal Army Film Department in 1918.4 TThe film bears testimony to the events between 15 and 24 June 1918 which decided the outcome of the second battle on the Piave (fig. 1). Four versions have been documented to date: the 1918 Italian and French versions premiered during July and August, both made by the military film department; a ‘2000-foot’ Scottish version named The Battle of the Piave, announced in July 1918,5 and a re-edited version (Ta Pum) released in 1927 (probably by Istituto LUCE).

Figure 1. La Battaglia dall’Astico al Piave (1918): Assault on Montello

The restoration has been carried out through the digital intermediate route and aims to reconstruct the Italian edition of 1918. The first public projection of the restored edition, accompanied by live music, was held on 6 October 2021 as part of the Pordenone Silent Film Festival.6

The surviving film materials (witnesses)7 gathered from film archives and private collections during the recensio phase are set out in the diagram below (fig. 2). Witness ‘K’ from Associazione Kinoatelje is a 35mm nitrate, first-generation tinted and toned print. It retains the original Italian intertitles, and handwritten information about editing and the colour palette on the film edge. The ‘G’ family came from Cineteca del Friuli and Lobster collections and witnesses the 1918 French version, whose first public screening in Paris is known to date from 7 August 1918.8 The preservation masters and new positive prints, made several years ago at Haghefilm through a photochemical route, are held by Cineteca del Friuli, while the original nitrate materials are no longer available. Witness ‘RM’ from Istituto LUCE in Rome is a 35mm fourth-generation duplicate negative, in four reels, with flash intertitles. The editing and intertitles witness both the 1918 and the 1927 versions and preserve several segments which are missing in ‘K’ and ‘G’. Witness ‘MI’ from Cineteca Italiana in Milan is a 35mm positive tinted nitrate fragment made close to the 1927 re-release, as attested by the intertitles and the insertion of later documentary materials. Witness ‘TO’, Dio segnò i confini d’Italia, became part of the collection of the Museo Nazionale del Cinema of Turin during the 1960s. It contains just a few scenes of La battaglia dall’Astico al Piave, and has given us a missing intertitle and scene.

Figure 2. Witnesses’ List

Witness K turned out to be the key material for the reconstruction of the 1918 Italian version. Other partial missing sequences were found in RM and G. The final sequence of the film came from witness MI, while we found a further missing scene in witness TO. Specialized scholars gave us fundamental help to identify, locate, and date sequences and find other witnesses and sources, such as the Turin and Library of Congress film materials.9 Indeed, we are currently waiting to analyse materials preserved by the Library of Congress and Cineteca Nazionale in Rome, having spotted a few other missing shots (fig. 3).

Figure 3. La battaglia dall’Astico al Piave (1918): Stemma codicum

As far as the intertitles are concerned, we decided to keep the original ones from the Italian 1918 version as preserved in K (fig. 4). Where missing in K but found in the G family or RM, we opted for a conservative approach,10 reporting the intertitle text, its sequential number and the film number (‘R37’) in square brackets and on a black background in the digital file.

Figure 4. Witness K: intertitle no. 36

As often happens, the film thresholds present specific loci critici: first of all, the four initial missing Italian intertitles which we found in the French 1918 version, but in a different order, probably due to the different role of the Italian monarchy in France, as in Italy the presence of the King’s portrait would force audiences to stand up and pay tribute. Second, the final shot, depicting the awarding of medals to the Arditi assault troops on 29 June 1918, is the result of a hermeneutic and editorial decision, since there are no concrete traces of its original position in the Italian edition.

At present, the reconstruction covers more than 90% of the Italian edition (1170 metres out of the 1255 metres recorded by the censorship visa). This exceptional result has enabled us to indirectly resolve and clarify the relationships between several films and documents of the period.

As far as non-film materials are concerned, they should not be considered secondary or as merely supporting and validating the restoration and reconstruction tasks. Indeed, studying these materials is a core activity in framing the film artefacts as historical objects and inserting them in an entangled intermedial and intertextual network of sources and practices. Hence, by using them, we were able to date and locate most of the sequences in the film and indirectly confirm the chronological coherence of the film editing.

The primary sources include bulletins (Gazzetta Ufficiale del Regno d’Italia) and journals (e.g., The Bioscope 1918, Kinema, Il mattino illustrato, Giornale del Mattino, Il Resto del Carlino) which attest to the circulation of the film between 1918 and 1933 in several countries and cities.11 Furthermore, we gathered many other primary sources, such as books, military reports, diaries (such as the diaries of Captain Maurizio Rava and Lieutenant Luigi Marzocchi),12 and photographs of the same events taken by the Italian Royal Army photographers.13 Images and related captions on the Battle of Piave circulated through the press of the time, such as in issue no. 14 of La Guerra published by Fratelli Treves, made using military department photographs. By comparing diaries, moving images, and pictures we were able to observe key moments from a prismatic historical point of view as the battle was told visually and in writing from different complementary angles and voices (fig. 5). Military archives (USSMI) and cinematic sources accessible through the European Film Gateway (EFG) and other digital repositories provided further crucial support.14

Figure 5. Witness K, frames 1181 and 2486 (on the left) and issue no. 14 of La Guerra. La battaglia dall’Astico al Piave, published by Fratelli Treves (1918), images 27 and 53 (on the right)

3 What Remains

Every restoration project implies an interdisciplinary approach and the support of different devices and infrastructures. With the multitude of sources and documentation available in the La battaglia dall’Astico al Piave project, the research results harbour great potential. However, since the results will be filtered into a basic digital package file (the restored version on DCP, an archival master) accompanied by essays and articles, this potential will remain unexpressed. Besides, the field of film and media studies has to deal with the fact that Italy at least is behind in designing and implementing digital research infrastructure that can support innovative forms of dissemination and different scales of investigation. A digital (media) infrastructure should

(i) enable distant reading […], that is, [identify] patterns or new research questions in and across aggregated collections, (ii) facilitate close reading: the detailed examination of individual items (e.g., videos) in a collection or specific sections of these items […] and (iii) make sure that the “scholarly primitives” […], basic activities such as “discovering”, “annotating”, “comparing” and “storing” […] are well supported (Ordelman et al. 2019: 133-4).

Even though this kind of infrastructure is being set up in various ongoing projects and research activities,15 in the Italian context many film and media scholars still struggle to place digital methodologies and applications within their heuristic outlook or to see archive practices as an equal field of knowledge production (Hanley and Heftberger, 2012; Op den Kamp 2017; Venturini 2019). This is in spite of the potential unleashed by radical technological changes intertwined with film preservation and research and disciplinary areas such as the digital humanities and media archaeology which, over the last 20 years, have helped to put medial agency in the limelight and therefore rewrite the conditions of archival research.

In this picture, our case study offers grounds for reflection and useful cues for discussion. As a recent publication devoted to the Italian WWI film heritage states,

a large part of the Italian films made during the Great War have been lost. The impact when we see the surviving images, often scattered in new film carriers that have little to do with the “originals”, is the same as when we look at a building or late medieval monument built with stones from ancient times: moved, reused, rematched. We can sense where they come from but struggle to imagine what the ancient building they came from might have looked like (Faccioli 2020: 9, own translation).

Despite the impressive work to recover and digitize the European WWI heritage in recent years, a large part of the surviving cinematic documents, especially in the Italian area, lack adequate datasets and critical descriptions establishing their exact intertextual relationship, historical entanglements, modes of production, formal structures, archival history, and material status. As is often the case, little or nothing is known about the archival backgrounds of the surviving copies (Flueckiger et al. 2016). We can only fully trace the provenance (Bernardi et al. 2021) and cultural history of witness K thanks to the careful documentation at the moment of excavation and recovery (Humar 2001). In some ways it is a longstanding and transversal question:

the availability of comprehensive, reliable datasets is still rather limited, especially when it comes to the earlier periods of media history. Another problem is that even datasets on the same phenomenon […] are not always based on the same data model, complicating their combined use (Noordegraaf 2016: 55).

A scholar wishing to combine archaeological, historical, and formal research on Italian WWI film archival artefacts and documents can encounter both a lack of entangled digital environments (able to connect different sources by linking data, annotate, and compare materials, etc.) and information (un)available on the tangible, technical, archival, and historical status of the materials:

scholars therefore struggle with an abundance of sources […] At the same time access to primary sources is in many cases still lacking, and the databases currently employed by film archives and other collecting institutions as well as online sources tend to differ, at times substantially, in terms of metadata quality and the available search functions (Heftberger 2019).

Despite the extraordinary and generous collaboration of several archives and scholars, when surveying and examining the witnesses we had to deal with limited research tools and functions. We experienced this paradoxical condition while performing the critical examination of the single film materials described above and the parallel analysis of other films belonging to the WWI film heritage. It was only by mainly using traditional methods that we were able to rebuild the network of intertextual relations expressed by the tradition (in the philological sense) of the film. At the current stage of the research, the witnesses of La battaglia dall’Astico al Piave contain or conflate with L’altro esercito (1917), Resistere (1918), L’ingresso degli Italiani a Trento e Rovereto (1918), Da Capodistria a Fiume italiana (1918), and The Battle of Arras (1918). Images from the film are reused in many other post-war films, among which Dio segnò i confini d’Italia (1918), Guerra Nostra (1927 and 1929), Il Piave mormorò (1934), and Gloria — La Grande Guerra (1934).

Considering the general picture briefly described here, ‘to imagine […] the ancient building they came from’ (Faccioli 2020), we should on one hand go beyond the analogue-digital divide (Heftberger 2018; Burghardt et al. 2020) and on the other foster the entanglement between historical-philological and archaeologically driven research and the digital humanities. In other words, my aim is to ‘think about how the digital is modulated within various materialities’ (Berry and Fagerjord 2017: 16). As such, here I will adopt a post-digital framework and critical approach to computational knowledge (Dobson 2019). This slant provides evident added value both for the restoration and reconstruction of the film and the large-scale organization of an important slice of the Italian WWI film heritage tradition.

4 Non-isomorphic and Isomorphic Modellizations and Knowledge Representations

I will take the digital surrogates and infrastructures used during the La battaglia dall’Astico al Piave restoration project to be forms and tools emerging from modelling processes which in turn shape the “choices we make in representing and analyzing the materials we study” (Flanders and Jannidis 2019: 4). As represented in the digital humanities:

by “modelling” I mean the heuristic process of constructing and manipulating models; a “model” I take to be either a representation of something for purposes of study, or a design for realizing something new. These two senses follow Clifford Geertz’s analytic distinction between a denotative “model of”, such as a grammar describing the features of a language, and an exemplary “model for”, such as an architectural plan (McCarty 2005: 24).

Here I use the subdivision introduced by McCarty (in the wake of Geertz, Goodman, and many others), to reflect on two different types of modelling used in our project: isomorphic (such as ‘edge-to-edge’ scanning) and/or non-isomorphic (such as visual and/or textual scene-by-scene annotation) representation, visualization, and annotations of film artefacts (the ‘model of’) and the design of a critical edition as a ‘design for realising something new’ (the ‘model for’). The model of/model for subdivision must not be taken as absolute since there is a thin boundary between visualization as representation and as a knowledge generator: ‘A basic distinction can be made between visualizations that are representations of information already known and those that are knowledge generators capable of creating new information through their use’ (Drucker 2014: 65). On these bases, in the following paragraphs I will reflect on the non-isomorphic and isomorphic representations of the film artefacts involved, and the current function and usefulness of a film-centred critical media environment and infrastructure.

In other words, I propose subdividing the modellizations into non-isomorphic (or hybrid) and isomorphic. In the first case, the film is seen as a witness and the focus is on textual reconstruction so it requires annotation and comparison tools. Instead, in the second case, the film is seen as a material object, so attention is paid to critically framing its digital representations and annotating and documenting its archaeological history (Marconi 2002: 48).

It is clear that the two types overlap a great deal: hiding behind the difference between isomorphic and non-isomorphic is the traditional distinction between restoration and reconstruction, which, ever since the configuration given to it by the so-called Bologna School,16 has exposed the deep entanglement between editorial and formal order and material order. This entanglement has been widely discussed in the literature, from different angles and at different levels of framing the film dimension, starting from the film-as-object (film come oggetto), film-as-spectacle (film come spettacolo performativo), and film-as-work (film come opera) originating in the Bologna School literature. In the last twenty years, several other similar tensions and layers have been added: the distinction between ‘mystical body’ and ‘mechanical body’ (Toffetti in Vignaux 2003, Catanese 2014);17 the film artefact as a conceptual, logical, and material record (Kirschenbaum 2007: 3); film as a material and conceptual artefact and the related frameworks (film-as-dispositive and film-as-performance, among others, see Fossati 2009 and 2018); and film-as-work, film-as-score, and film-as-performance (Hediger, 2011). Furthermore, there has been reflection on the material properties of film in the digital (Flueckiger, 2012), experimental cinema restoration practices (Venturini et al., 2013), and the archaeology of handmade films and artist-run lab practices (Catanese and Parikka 2018). With this knowledge, the distinction between isomorphic and non-isomorphic was used to create a practical frame not so much of the film as the forms of description and visualization adopted during a restoration and reconstruction process.

In these terms, a film restoration project implies immersion in a hybrid research and production environment, in which traditional analytical methods and tools combine and are aided by several different digital environments and analysis tools.

Nevertheless, even for the most up-to-date researchers accustomed to dealing with digital preservation practices, ‘introducing computational research requires [them] to review and complement the methods they traditionally work with’ (Noordegraaf 2016: 51-52). An additional difficulty is finding one’s way in the vast field of solutions and tools on offer. It is not rare for them to be unadaptable or obsolete, and so it has to be decided whether to use them anyway or to develop new ones without, however, the right skills or resources (Melgar Estrada et al. 2017: 44; Karsdorp et al. 2021).

In our case study, except for the research performed in digital archives, we mainly used digital research environments and tools as organizational support to deal with the different data gathered in the initial phases of the research. As will become clear in the next paragraphs, our attention was mainly focused on the digital assessment, visualization, and analysis of the film artefacts.

4.1 The Non-isomorphic or Mixed Model

The non-isomorphic (or mixed) approach aims to document the material and editorial film layout and at the same time lay the foundations for its reconstruction. The non-isomorphic description and annotation of a film artefact is grounded in ‘its segmentary nature’ (Heftberger 2018: 29) and therefore historical tools such as so-called découpage, in the sense of a prismatic theoretical and analytical concept for understanding film structure. Straddling avant-garde and classical European film theory (Barnard et al. 2020), it was designed as an entomological and anatomical approach to the formal and material artefact, in order to dissect it. Historically, the so-called ‘sceneggiatura desunta’ (usually translated as ‘technical screenplay’) has played a particular documentary and pedagogical function since it was used in the 1920s and 30s during the institutionalization of film culture, in an age lacking in study resources (Pitassio and Venturini 2014). At the time Renato May wrote:

it is in essence more of a documentation [emphasis added] than a script […] It is like examining the structure of a building and discovering the static laws that allow two thin columns to support a massive arch, or how the stones were arranged so that a wall would acquire a certain solidity. (May 1939: 21-22, own translation).

Like in the case of the allegory of WWI film heritage as a ‘late medieval monument built with stones from ancient times’, here I deliberately take the metaphor of documenting a composition’s rules in a much more literal and materialistic way than May intended in his time.18

During our project, which adopted a philological method, the examinatio (the description and analysis of the single materials/witnesses) and collatio (their comparison) phases were conducted using basic spreadsheets, pivot tables, and markers and annotation tools which supported us in segmenting and comparing the material, formal, temporal, and intertextual characteristics of each witness (fig. 6). Even though here it is not one of my goals to assess methods and uses, it is evident that a digital film philology has to evolve towards specific tools and applications that mix collation, multimedia annotation, and automatic recognition software (such as the CLARIAH app suite or VIAN annotation software).

Figure 6. Examinatio and Collatio

In the example shown in the figure (fig. 7), we imported three different versions of the film incipit into CollateX, each of which is witnessed in the surviving materials. We normalized the different intertitles so they could be used as references to apply a collation algorithm, realign and reorder the witnesses, and provide a visual account of the variants. This tool enabled us to check and support our critical hypothesis and so reconstruct the incipit of the 1918 Italian version.19

Figure 7. Incipit: collation and variants (French and Italian versions)

In the reconstruction it was not possible to operate without non-linear editing systems, such as Resolve, to annotate, align, and visually compare the main witnesses with each other. These systems are increasingly used as learning and sharing interfaces for post-production restoration lab analysis because of the verbal, graphic, and visual annotation tools, and the multiscreen and multi-trace visualizations (fig. 8) which support both comparative and macro readings of the whole film tradition and the examination of variants, contaminations,20 and loci critici. This measurement and comparison post-production practice did not come about in the digital era but is rooted in analogue film editing techniques which used comparison moviola and film synchronizers, practices still used in restoration projects involving direct, side-by-side comparison with original film materials.