EPISODE 5: Martina Hoffmann, Swiss National Library

In this episode of Wider Lens, the first podcast dedicated to cultural heritage digitization, Digital Transitions sits down with Martina Hoffmann, Head of Digitization Services at the Swiss National Library, for a wide-ranging conversation recorded live in front of an international audience. Martina took an unconventional path into the field, arriving not through fine arts or photography but through project administration on one of the Netherlands' first mass digitization initiatives. Fifteen years later, she leads digitization services for a national library and still runs her own independent consultancy on the side.

The conversation covers why digitization standards matter, how the ISO 19264-1 standard differs from guidelines like FADGI and Metamorfoze, what a real quality assurance process looks like in practice, the most common mistakes labs make when building workflows, how to train new digitization staff with no formal education path to draw on, the differences in cross-institutional collaboration across Europe versus the U.S., and a candid take on where AI genuinely helps (and where it doesn't) in preservation-grade imaging. The episode closes with a lightning round and live audience Q&A.

About Martina Hoffmann
Martina Hoffmann is Head of Digitization Services at the **Swiss National Library**, a role she has held for nearly six years, and works internationally as an independent consultant on cultural heritage digitization and quality assurance workflows. Before Switzerland, she served as Senior Production Manager for Digitization at the **National Library of the Netherlands** (Metamorfoze) and as Operations Manager of QA for Digitized Products at the **National Archives of the Netherlands**. She also hosts a monthly community meetup now marking its sixth anniversary.

Conversation Highlights

• Why "everybody wants everything online yesterday," and how the field's understanding of production and process has matured
• Why standards remove subjectivity: comparing ISO 19264-1 to guidelines like FADGI and Metamorfoze
• The real cost conversation: why meeting a standard isn't necessarily more expensive — it depends on what you're actually trying to produce
• What a genuine quality assurance process covers, and the most common mistake labs make (trusting target-reading software instead of your own eyes)
• Training digitization staff with no formal education path, and the real priority: handling fragile originals safely
• Where AI actually helps in cultural heritage work — and where Martina firmly draws the line
• Audience Q&A

#CulturalHeritage #Digitization #DigitalPreservation #Archives #Libraries #Museums #GLAM #ISOStandards #FADGI #Metamorfoze #QualityAssurance #SwissNationalLibrary #WiderLensPodcast #DigitalTransitions

Transcript

SPEAKER_00 0:01

Welcome to Wider Lands, a podcast from digital transitions. This series, we sit down with leaders from across the cultural heritage community to talk about the real work of digitization, what's happening today, what's coming next, and what it means for the people behind the projects. Each conversation explores heritage imaging in practice, the technical challenges, evolving standards, and complex collections, but we also focus our lens on the human side, the stories, lessons learned, and moments that don't always make it into the conference papers or imaging specifications. Whether you're working in a museum, library, archive, corporation, university, or research facility, WiderLens is a space to share experience, spark discussion, and think critically about the future of cultural heritage digitization. Let's get started. So I want to introduce our esteemed guest, Martina Hoffman. She's the head of digitization services at the Swiss National Library. Works internationally as an independent consultant as well in cultural heritage digitization and QA workflows. She was a senior production manager for digitization at the National Library for the Netherlands, uh, archival section of Metamorphos, operation manager of QA of Digital Digitized Products in the National Archives of the Netherlands. Um very well known, very, very knowledgeable. We're super excited to have her here. Um I'm gonna stop talking about her and pass it over to her. Martina, welcome. Thank you for joining us. Um before we really dive into the meat of the conversation, uh, would you mind giving just a little bit of background about yourself and how you came to be here?

SPEAKER_01 1:34

Hi, everybody. Thanks for uh inviting me to this talk, uh Patrick and uh DT, of course. Um happy to be here, happy to talk about uh the work that we all love, uh that we all do. And um how I ended up here in this podcast or how I ended up here in this uh field. Let's uh let's put a decisive um point here. Because I ended up here because um you and I we talked, and that was a very nice conversation, and here we are. But how I ended up in the field is a little bit of a stranger story. Um I'm I'm not your typical uh cultural heritage employee, let's put it this way. I do not come from an arts background, I do not come from a fine arts photography background, I come from a completely different background, but I ended up by luck, chance, coincidence in this field very uh many years ago by now. I always keep telling it's 15, but I'm can tell that for several years now. So it's probably a different number now, but I still keep telling everybody it's like 15 years, and because it's like 15 years, this can be everything more than that. Um came into the whole field with um completely different tasks. We had a huge first-time mass digitization project in the Netherlands that I was uh able to uh support with administrative work, how to upscale processes and administration for huge amounts of um well, images at the time. It was a photographic collection and um normally it was a little bit smaller, like a few hundred, a few couple of images, and this was millions of images. And my background was um more in the organizational part. So I ended up uh helping upscale the whole project, and that's how I rolled into the whole field of quality management, how to digitize, how to do project management for digitization and cultural heritage. And with everything that you said, Patrick, I um ended up from uh working in the Netherlands, now in Switzerland, the last uh six years almost as uh head of digitization service for the Swiss National Library. And in between the times, I still have my little consultancy where I'm very happy to help people um do stuff with their collections and upscale their projects.

SPEAKER_00 4:04

Um nice. Yeah. Um you've been involved in so much. And I do have to say, like, I think every other guest we've had either came from some sort of fine arts background or photography background. I think you're the first one to come through sort of this administration sort of uh path here. I think that's really interesting. Um with you know, you work for libraries, archives, standards, organization, you have your own consultancy, and you've been doing this uh, let's say 15 years. I'll stick with with that number for you. Um what do you think has changed the most through that whole journey? Everything you've dealt with. I mean, obviously technology and processes, like what has been your biggest biggest takeaway about how things have evolved and what's changed over that time period?

SPEAKER_01 4:41

Well, let's put it this way the maturity of the field has evolved a lot. Um if I say what has not changed in the past years is like everybody wants everything yesterday, like everybody wants all the cultural heritage to be online and um preferably like yesterday. And the knowledge about it's not possible is not there, but the knowledge in the field about what you have to do to actually get more production done, to get upscale programs, upscale processes to how to um make more of uh the few things that we have in cultural heritage, um, that has definitely evolved. That um besides all the technology, yes, we do have now 250 megapixel cameras. We have whatever now, you know, but that's that's a different part of the thing that we cannot so much influence because that depends on manufacturers and all that kind of thing. But how to how to deal with processes and how to make it work, and also how to ask for help that has evolved a lot. I think a lot of people um now, or I at least I I try to believe that that a lot of people now know that there is a whole community that is very willing to share, and that's why people also ask more questions to the community, and so we learn from each other and we evolve and mature in this whole thing.

SPEAKER_00 6:18

I I love that you know, you talk about community, that's something I'm saying, DT. We we've tried to kind of participate in that and foster that as much as possible. Um, have you how have you seen that evolve? I mean, you know, with the rise of social media and the internet and all these things to kind of make connecting better. How have you seen that community evolve and take place? Where where is that community right now? Is it at specific websites? Is it in person? I know you participate in the Grabica thing as well. Um, talk more about the community side and how that goes.

SPEAKER_01 6:45

I'm not I'm not participating, I'm hosting it.

SPEAKER_00 6:48

Well, I was gonna let you you say that, but yeah, you you host the grabica, obviously. I mean that I I gave you a demotion there. Sorry about that. But yeah, you you host grab a cup.

SPEAKER_01 6:57

That that that's that's completely fine. Just for people that don't know about it, they should be aware that I I'm basically all the time there because I'm the host. So if you don't want to deal with me, I'm always telling people you shouldn't come to the grabica because you will meet me and you will meet uh my green here, and you will have to deal with uh what I'm talking about because uh um I'm I'm basically on on the side that you are now. I'm introducing my guests and uh the and the topics. But we have um we are we are actually next next month, uh no, this month, it's already July. In a week, we will have our sixth anniversary in the Grabber Cup, which means uh once a once a month there is an open get-together for the community, and the community takes still advantage of it. It of course came out of um our darker times uh in pandemia, where we were not able to meet in person. And meeting up in person is also harder because we are spread across the whole globe and not all have the budget to travel for conferences, which is the other big thing that I'm I'm super happy about is the conferences, you know, to go to the conferences where we meet up uh wherever they are, and that we spend time together, that we are also able to talk to each other, that you you go to on behind-the-scenes tours, you know. Also, very important to just take a peek look when you go somewhere and then like what are you doing about it? And sometimes on the grabbicup, we are happy and and have somebody share what they are doing. So that's that's a cool thing. The other thing that um that of course is um also very cool about this community is wherever you are or wherever you go on vacation, maybe you can just send an email ahead and ask, um, are you available? Can I just pop in when I'm there uh by chance? You know, and everybody has a very open door policy. Like I've never had anybody say no. And also, I try to do the same thing in the Swiss National Library. We have an open door. If anybody is uh nearby, just uh drop me an email and we will see what makes uh what what is possible with all this. So that's a strong sign that the community is really engaged with each other for me, and I'm I'm really happy to be part of that.

SPEAKER_00 9:21

Yeah, absolutely. I think um that's fantastic that that you're you know create this open, welcoming environment. And I think um, you know, kind of leading into another topic uh we're gonna get into here, but you know, one of the things that that we discovered at DT, you know, many years ago, a decade ago, maybe more, when we started doing our roundtable is that you have so many people in different digitization departments and different organizations around the world, and they're all trying to solve problems within their own department, but there wasn't that place you could go to get a question answered. You had to kind of figure out on your own. And I think this community helps that people share information and share best practices, the best way of doing things, which is why, you know, from my point of view, you know, we've seen things like digitization standards have become more and more um necessary and more popular and more well-known. We still get questions about them all the time. But obviously, I know you have strong opinions like this about this, but talk to me about why digitization standards matter in our cultural heritage industry. Let's start with the big question and hear what you have to say about that.

SPEAKER_01 10:22

Well, uh the biggest argument for me is that it takes away the subjectivity of things. You know, it does not matter if you take the shot or I take the shot, it does not necessarily to a certain uh degree um depend on what kind of technology we use to take this image, to make this digitized product that we are producing here. It will come out the same as if we settle to the same parameters and to the same quality. And that's for me the biggest thing that I can compare something that Kurt Humiller at the NGA is doing, because if we are using the same method and the same standards and we adhere to the same parameters and the same quality levels, it will come out the same no matter if he did it or I did it. It's it spans across a globe and we can share with each other. And the worst thing that you can have, for example, with um some things that are done in digital humanities, which is super cool that you have a part of a collection here and part of a collection there, and you want to bring it together for your research, and then the one looks green and the other one looks red. I mean, you know, not the best thing for your research, and also something that we as a professional probably do not want our customers, whoever wherever they are and whatever they want to do with it, to experience that it's completely different from one another, you know? Yeah, that's that's the main thing for me.

SPEAKER_00 11:56

Yeah, and and how have you seen standards evolve? I mean, Fadgy 4, which is sort of like in North America, kind of the default, only came out 10 years ago, I think, something like that. So, how have standards evolved and and where do where do you see the difference between the different standards? And and is there a difference? You know, talk through the different ones and how you've seen them evolve over the years.

SPEAKER_01 12:15

Okay, let's start with the really important thing here. And uh, as you said, I have opinions, but it's also facts. There is only one standard. It's the ISO standard 19264-1 at this moment in time for reflective material. Dash two will come out as soon as they're ready, hopefully soon. I'm bugging them basically every month to publish uh anything because we really, really need it, but it's not there yet. But that's an ISO standard, which means it is a standard that is local, uh globally valid for everybody, and then you have guidelines, and fatty is a guideline and that helps people to make decisions on a different level. So, for example, the ISO standard doesn't tell you which uh PPI is the correct PPI for your object, and FATGI gives you a helping line in well, we suggest that this is a great idea. If you want to adhere to FATU4 star as a highest quality, for example, then you use XYZ PPI. The same thing is true for metamorphos, the thing that people are um or has developed in Europe, let's put it this way. So FATI is uh going back to 2010 already. I know people do not like to be remembered about that one because it so was so hard to read, but 16 was not the first version, uh, it was just the one that everybody could read, actually. And um Metamorphosa in the very, very first beginnings started 2007. So um long history, but they they do not exactly have the status of a standard. Going back to my own practice, before we had the ISO standard, so before 2017 and then the reversion of uh 2020 uh 21, we had only guidelines. And um, if you are on on my side of the table, like I I give you a contract and I want you to perform to this well, best practice guideline and quality level that I ask you to do. Um it made it hard to communicate or harder to communicate with vendors and suppliers and service uh suppliers to actually get what I wanted because there was more room for discussion. Let's put it this way. And the ISO standard is um for me um the perfect thing because uh there is no discussion. It's it says and states everything very clearly. So big fan of the ISO standard, of course, and also saying if you want to read the ISO standard, it's a great uh bedtime um thing because you can fall asleep very easily if you are not so into all those technical things, but it also gives you a lot of flexibility. The ISO standard wants you to know a little bit more about what it says and what it doesn't say and how to actually use it. So you have to dig a little bit deeper, but that's also a lot of fun. Um, and you can use a lot of flexibility, and then the guidelines are there for the for the practical use that is valid for a lot of institutions. There is nothing wrong with uh following the FatG4 star, the FatG3 star. It's a decision you have to make for your own institution, for your own program, or for your own uh project that you have.

SPEAKER_00 15:40

Yeah, I I I appreciate the distinction there. I think that's really important to understand uh what a standard is and what a guideline is, and um and and where that's important. Um I want to talk about something else. You know, we we run into this a lot where we start talking about standards or guidelines and and trying to meet them. And I think a lot of organizations have the perception that it's either really expensive or they don't have time or they can't afford some specific level of equipment, like a 250 megapixel camera or something to reach that. Well, how do you respond to those people when they say they don't have the budget to for to match standards?

SPEAKER_01 16:13

There is not one good answer for that because it depends on what do you actually want as a as a product, as an as an end product, let's put it this way. If your goal is um and JPEG on a web, well, don't bother. You don't need any standard, you know. It's like, okay, why would you do the effort? I can completely understand that. Just take your smartphone, be my guest, you know. But if if your if your aim is something completely different, like um for the Swiss National Library, the aim would be uh long-term preservation, because the motto behind all of this is do it once, do it good. You do not want to do the same thing over and over again. Besides that, that means that the object is well, more often under digitization than you actually want it and is good for the object, it also costs a lot of things. So, from a from an economic point of view, doing the same thing over and over again is not very clever, and we have so many other things that we need to attend to. So let's not do it twice, let's do it once, and then it should be done. So the level of quality that I want or that we want from a product is higher, it's not a chapic for the web. And then that warrants also more investment in equipment, for example, if I have to do it myself. The easy and short answer is it doesn't cost me necessarily more to do a high quality image. It just is about um how do I do it, which maths do I do, like if I do like one picture, and for what? What is the purpose of my digitization? But if you have that clear and the decision is you want high quality, it doesn't cost more than well, low quality, and the equipment is not that expensive. You can go a long way. Well, please do not come with your smartphone to me. I know there are people that like that because everybody can use smartphones, and that's something I do not agree with.

SPEAKER_00 18:21

Yeah, absolutely. And but I think the point, you know, there are there's a wide range of price ranges for equipment you can buy, and you don't necessarily need to get the the most expensive to achieve these standards. There's there's plenty of alternatives, and there's a there's a big gap between you know the best, most expensive, and what can still get you there. Um I think this this kind of leads you into an another topic here, which is you know, we talk about standards, and this is sort of conceptual what you want the end product to be. But the step from what you want and actually getting the digital asset, that's where workflows come in. So, how do you build a workflow that that produces consistent images that that meet standards every day? And you've done a lot of work in this. Um, so how do you begin that process and what do you tell people when they ask you about that?

SPEAKER_01 19:04

Well, uh I give lectures about that, you know that at the university actually in Zurich. But the the the main part is to find out what actually somebody wants, you know? The the the the main problem in a lot of institutions is that somebody comes and says, I want a high resolution picture. I'm like, yeah, all right. What's a high resolution picture for you? You know, like can we can we have a conversation about it? And they there is on the other side of the table sometimes like it's high resolution. I mean, yeah, it's high resolution, but what does high resolution mean? Or what is a large format for you? I mean, is it they they probably know it's a TIF or it's a JPEG or it's a PDF. And if they come with a PDF, I'm like, oh, I don't know, I don't think that you want a PDF from me. But um, so finding out what what my counterpart actually wants from me, you know, and then telling them, okay, so high resolution doesn't mean that you get a good quality. I mean, I can give you any resolution with not quality, let's put it this way. And then seeing where where we where we go with what kind of object do they give me, because the workflow or the requirements for the workflow, but also for the quality, depend very much on the object that they give me or the objects that they give me. Like you can imagine newspaper is not the same thing than um photo from uh 1890s, uh 1920s. I don't know. It's one photograph and a lot of newspapers. So I have to build a workflow for millions of pages of newspapers, for example, which is a completely different beast than having this one small collection with priceless letters from I don't know. Um means completely different things for my for my requirements. What I try to do is to find a workflow that works for an institution. So one workflow where you have high-level steps and they're always the same. Like it doesn't matter if like you are behind the machine or if I give it to an outside service supplier, um, the the thing that has to be done is it has to be prepped, it has to be transported, it has to be done with the requirements. And because I'm I chose at the Swiss National Library for that ISO standard, I can say, okay, I want ISO standard quality level B from you, Patrick, because you are my internal employee, but I also want exactly the same quality if I send a million pages to a to a service supplier, you know? And I'm writing all the requirements for the quality. Like I want you to do a test target, like every session, and I want the test results from your test target because I want to see that you actually did. Your own quality control. And then I want to see a portion of the production. Like I'm doing a sampling and I'm giving all this information out before I'm sending a project out, basically, internally or externally, it doesn't matter. And then doing the quality control, you know, the quality control of the whole thing is a really big step. I'm always telling people, you can ask for everything that you want. First of all, it has a price and a time impliance. I mean, it costs time to produce things. But if you do not check if you get what you asked for, well, don't ask for anything. You don't need to ask anything.

SPEAKER_00 22:39

Yeah, absolutely. What does a good quality assurance process look like?

SPEAKER_01 22:43

Well, a good quality assurance process um covers at least all the things that you asked for, you know? Like if you asked for a hundred images, then you should check if you actually got a hundred images. If you sent out originals, I don't know, across country, for example, and it comes back, you you hopefully said that you want it back in the same state as you sent it out. So you should check when it comes back. Did I get everything back? Is it in the same state that I that my originals went out? Um if you have um a transport, and you probably always have a transport from data from A to B, is there uh a transport insurance in the sense are all the pixels and all the series and ones that you created somewhere in New York the same as arrives with me in Switzerland? No, all those kinds of things because there can tons of things can happen, you know. Even if you just put it on a USB stick, it can happen that uh file gets corrupt or so. So you don't want that in your long-term repository, for example. But then again, if I want an ISO level B, I will check for an ISO level B because there we have a workflow for targets, and I can check with the target if you actually produced it. And if I said I do not want to see any hands of Patrick in my images, then I will take a sample and look if there's any hand of Patrick in my image, you know? So everything that I specified, I will try to find also a step where I can actually control that.

SPEAKER_00 24:15

Yeah, I mean, it sounds like it's so much of, and this makes sense with your background, but so much of what you're discussing is is it really about um process, as much as just like actually clicking a button to capture an image, like really defining what you want and checking that you're getting that and really doing a lot of work up front to make sure that what you're getting at the end is meets all the expectations you're setting for itself. Um before I get into the next topic, uh just on this in terms of like workflows and quality assurance, do you have one piece of advice that you would share with the industry? Like what's what's the most common mistake or shortcoming you see in digitization facilities when they're trying to do these things?

SPEAKER_01 24:47

Uh uh when they're trying to do what exactly?

SPEAKER_00 24:51

When the when they're building a workflow and they're trying to meet you know the standards that they've set for themselves. Do you see common mistakes people are making or maybe just shortcomings?

SPEAKER_01 24:59

Uh try not to trust the software.

SPEAKER_00 25:02

Yeah.

SPEAKER_01 25:04

There's there's one common thing that people um very often do is when they are checking targets, for example, they all the targets we are not measuring them ourselves. I mean, we could, but it takes way too much time, not efficient. So we all have software that we run the targets through, and then it says either green or red, probably. Most of the software says either green or red or a variation of both. Um and then if it's green, everybody says, like, it's great. And that does not exactly mean that it's great. So try to actually look at the image of the target, even though the software says it's green. Every now and then just open the image and trust your eyes, and then look at that. Also, the other thing that I always kind of suggest as a tip is don't do the quality control on the same software and PC that you actually created the image with.

SPEAKER_00 26:06

Yeah, yeah, that's that's one I've heard many times. But yeah, that's a really, really great point. Because you talk about like you're you're you're training for the test, like you, you're you know, you're not getting different verification. Um, that's really great advice. Um, so through all this, like, you know, doing all you're building on the process and uh you know defining what you want out of it. At the end of the day, someone's got to sit down at a machine and take a picture of something. Um and right now there's there's no formal education path to get into heritage digitization. Um, if somebody wants to go down this path, where do they start? Where do you how does someone become excellent at digitization? You said before you can't learn this entirely online. Talk through what what you want people to do or what you recommend they should do if they want to get into this.

SPEAKER_01 26:50

That's a really hard one because um it would be so much easier and and really, really great to have a study or uh, you know, classes, where whatever, to just send people to and they come back and they are, oh my god, I know everything about cultural heritage digitization in mass production. Because there is a lot of people that come from the fine arts or photography sector that um know a lot, and actually for the mass production that we are we are aiming for to get out of the subjectivity, like I I like this picture, or this picture is a pretty picture, which is something that they learn, of course, because that's more the focus of this study. It's like, no, it doesn't look too great, but it's exactly like it should look for a preservation image. And to learn that is actually harder for people that have this sense of artists, uh, you know, that's what I what I always find very intriguing and also very common. I'm always saying to everybody, you don't have to be a photographer to work for me. I would like to have uh an education for people, but actually I'm trading since the 15 years that I'm working in this field, I'm training all my stuff myself, which is it's there are some helping lines like the DT classes that you can do, or certificates that you can do, or videos that are shared online. But in the end, it's I think the most that I do is actually I have workflows in place, I've things set in in motion that they can actually start running with the production, and then they get the practice. And then the more the practice comes, the more we need to know is why is this actually like this? And then we we we go from there, you know, it's like a different approach, but also from the necessity that you need somebody to be available very, very fast for production.

SPEAKER_00 28:57

There's this there's the first step is like learning how to use the equipment. Like here's what a target is, here's your image capture system, here's the lights, here's how you you know color calibrate, and you learn kind of the nuts and bolts on how to operate the equipment on a technical level. But to be really great at producing preservation grade work, there's another, I think, step you need to go in in the education. To your point, like that's where doing it a lot helps. What are those things that you learn beyond just the technical operating the machine? What are the things you want to see people learn after that?

SPEAKER_01 29:27

Well, I I actually in my operation, people don't even learn how to do all this uh technical stuff. They learn how to actually operate the machine and like push the button. And the main challenge that we have is to train them how to handle the originals, because we are not working with trained professionals in that sense. We are working with um a thing that is not common in America, for example, but people that are not going to the army, that have no education at all, they are young uh guys that have to do civil duty, and they come to us from all kinds of backgrounds. Some are just out of high school, and we have to train them actually what cultural heritage means, how to handle cultural heritage, to be careful not to make any damage to the original. Like we are using a lot of glass with our digitization for mass digitization. So um, how not to press the glass table down on the object and break the object, you know? All those kinds of things are way more important and cost way more time for us. Um, if we have, of course, trained professionals that know how to handle colour heritage, that's the easier stuff. But um, there are very few of those around. Uh and they um they have more the the challenges that they actually then want to know more about the image or how to how to make the the image presentable on the on the web, for example. Like, oh, I'm doing a book. Uh does it show like up like a book? Because the connection would be, oh, I have a book now, I have a book online, but most of our platforms only have one-page viewers. So you have to, why do we cut the pages? All those kinds of things, you know. It's easy for people to understand, push this button, push this button, push this button. That's also why our workflows are able to host students, um, short-term uh employees, whatever not. But it depends a lot on on who you are and what you want to know besides that. It works also without any knowledge. Let's put it this way.

SPEAKER_00 31:47

Yeah. Um, that's a great distinction. Um, and obviously that the handling of material and and how delicate some of the stuff is is is really critical. Um, I'm curious, uh, I'm gonna we're gonna change subjects in a minute. I want to I have some some thoughts on technology and questions around that. Before we get into that, I do want to ask one other thing. Working in Europe, I think it's very different than in the United States, it's it's just the United States. We have, you know, the things don't vary that much from coast to coast. But you've worked across multiple national institutions, across, you know, different countries and borders. What is cross-institutional collaboration and standards look like in Europe? And and do you see differences over there in terms of how um institutions align with their standards, how they work on these things? What what what has your experience been there?

SPEAKER_01 32:30

Super different. Each country is um completely different. Uh, not all want to uh improve their quality by standards or um using standards in their projects. We do have, at least in the library and archive field, people that actually know it, that have heard of it, that try to implement it, but because they are limited in um their resources in-house, for example, they they cannot buy new equipment or they don't have uh um people to actually do the digitization. There's a lot of institutional thoughts why they did not yet do it, or why they want to do it, but it's only for this kind of object. We also tend to send out a lot of stuff to service suppliers, and the service suppliers are happy if they can do whatever they do best, and that is not necessarily what we want them to do. So uh you need to be a very um strong advocate for your own requirements. That uh that is something that I definitely learned. Um, it helps definitely if you are funded. Well, we I mean, that's an international thing, of course, but because we are not one big nation in Europe, we are very few um institutions that really are funded by their own governments, and some have less funding and some have no funding, and there's a lot of cultural heritage where we have to preserve. Everybody struggles to find money. So we are also very aware that we need to move fast, very fast. And having a program like in the Netherlands that is actually a governmental program funding the digitization of cultural heritage helps a huge amount uh for also the development and getting the technology and the knowledge and standards involved, and others struggle because they are more alone. And what helps here a lot, and I've seen it in the past couple of years happening all over Europe, people are getting very, very interested because the sense that quality and do it once and good means less cost in the end. This settles in because we are more money aware, let's put it this way, more budget aware. We we cannot afford people now start to see that oh, this image from like 20 years ago, I can't even open it anymore, you know, because the file format is gone. And and that that makes a huge shift. We are in a in a huge shifting, changing world in Europe.

SPEAKER_00 35:18

Yeah, it's it's you know, going back to the conversation we talk about um, you know, do it once, do it well sort of mindset. And you know, it it can seem more expensive up front to spend the extra time to do it to upgrade your equipment. But yeah, not having to do it again every three years or five years or whatever. I mean, uh, we've all had that moment where we've opened up a JPEG that we took 20 years ago, and it's like, oh my god, look at this. And it's it's this tiny little thing, and the color's horrible, and the you know, the quality, the compression on it, whatever it is. Um, and so not having to do that again, I think it's communicating that that is saving you money long term. It may it may be some short-term pain, but long term it's the healthier thing to do financially. Um, so speaking of technology, I'm just gonna throw a loaded question your way, and I just want to hear your thoughts. AI is obviously everywhere right now. How do you view it? Uh, and specifically, how do you view it in the cultural heritage world?

SPEAKER_01 36:08

Yes, it's everywhere, unfortunately. Um, everybody uses it for everything, and it's um not good um development. Let's put it this way. Um, AI is um overly used for a lot of things that should not be done by AI. AI is great for tasks, yes. If you do know how to actually find the errors in the code that the AI gives out, for example, if you have repetitive tasks that are saving time, you you have a script from the AI and it helps you. I did it myself with uh Doug last year. So um there are a few very narrow scenarios where where I see AI as a very big helper, but in the end, um being sustainable is I think more important. AI does come with a cost that nobody sees. On one side, we are we're telling people we cannot save everything that we are doing on our servers because the files are too grow, too, too big, and we have too much cost, and the environment is bad, and the energy that and then everybody's using AI. I mean, there is a problem here for me, like ethically, you cannot say one and the other is the same uh truth. So be careful if you use AI um and responsible with what you are using AI. And if somebody tells me AI can do my quality control, I would say yes to a certain extent. But um the AI cannot decide for me what the actual picture should look like because it doesn't know the original, because AI cannot look, it has no eyes, and until we are there, basically the human person is still the most valuable asset that I have for those things. And also don't forget that there are implications for people if you use AI for everything and if AI is your savior for everything. Um, and maybe also it's more sustainable to actually hire a programmer that actually can code uh instead of doing all the AI things yourself, which costs also a lot of very expensive money that you cost.

SPEAKER_00 38:36

Yeah, we're seeing we're seeing this sort of backpedal now from a lot of companies that they fired a bunch of people at the end of last year, beginning of this year, and said, Yeah, AI is gonna do everything. And now they're going back and saying, please come work for us again, because the AI, it's more expensive, it's not as good. Um, we need human beings to actually know what they're doing back in these seats. So I think the world is still trying to figure out where AI fits and what it can be trusted to do and what it can't. Um, what are your thoughts? I mean, you kind of mentioned things like like quality control, maybe you know, coding, maybe do you see other places like metadata or um making maybe helping define a workflow or you know, standards, what we're trying to get out of a collections uh you know, final assets. Do you see places where AI has a use and and where would you not use it?

SPEAKER_01 39:20

Well, I I I would not use it to define workflows for me, um, first of all, because I'm convinced that I'm still way faster and better to oversee all the consequences until I've written all my story to AI. I already have made the workflow. So um, no sense in doing that. The other thing is metadata, um difficult if it's like a lot of times people would like AI to actually see what's on the object and then make at least a suggestion what's what the content is. And I absolutely see that as a very useful thing because you can have a lot more data process in the same time, but and here comes my big but that only works if you do input good quality images, because if the AI does not see or cannot determine what it is on it because it's a bad quality image, well, basically your metadata will be bad, and then nobody has anything from it. So makes actually this use case actually makes my point to make good images, do the quality.

SPEAKER_00 40:37

Yeah, you can get much more out of it. And and you know, just full discussion, DT has done a lot of projects in the past year or two with with that kind of thing, like basically pulling like uncovering data within the image. So to your point, like recognizing what's in there and using that to tag images or search through images based on that. But yeah, you have to have a good asset to start with, or it doesn't make uh make any difference. Um, you know, kind of tying this back into just imaging standards. Do you see a world or a day where we start to define standards for using AI within cultural heritage, like defining best practices or some sort of standards around it? Or do you see that happening? And do you think that's a good idea? Should we do that?

SPEAKER_01 41:14

I'm not exactly sure if we will have standards for that, but I'm definitely sure that we will have best practices and that we as a community are really thinking of how and what we can do and what we can expect from it, and what what we should not ethically do, because the community itself is a very ethic community, you know. We are thinking a lot about can we actually do that? Can we put this image online? Is it do we need to frame, make a PR announcement uh with it, make um another website with it to give context to the user and not just throw the data at the user, you know? How can we explain to people that this is a math standard and where they actually find all the maths definitions so they can actually read our metadata in this specific format? So we are we are maybe even more caring that we than we would need to, but I think it's a very ethical point of view that we take. Uh, and that's why I think we will, as a community, end up with at least best practices uh on how to use AI in our cult of heritage field. And what we think is an absolute no-go.

SPEAKER_00 42:31

Yeah, I I I I I expect that's gonna happen too. I think things like best practices as well as things like um, you know, some sort of watermark to show that things are not AI, that you know it's the actual authentic, you know, image, those kind of things are definitely coming in and heard talk about that. We've got about 10 minutes left, so I want to leave some time for our audience to give us some questions. So I'm just gonna ask a couple of sort of lightning round questions at the end here. Just throw some questions your way. Is there any technology you see coming out soon or that's come out recently that you're excited about? What's what's the digitization sort of tool technology, software, practice knowledge that you're most excited about coming out?

SPEAKER_01 43:08

I know everybody says 3D. Um that's that's not something that I would now classify as new technology or advanced technology. I think that the most interesting thing to follow is actually how we use AI and where AI actually goes, and if if this is really the path that we want to go down as a society, um given the implications uh economically and um also um with uh with prices, because um everything costs money and we don't have that uh in the cultural heritage field. So that's actually the technology I'm I'm most curious about to follow. And the rest is well, bring it on.

SPEAKER_00 43:56

Um okay, last couple of questions. Do you have one piece of tech? Technology that you just can't live without, whether at work or personally. What is your go-to tech tool that you have to have every day?

SPEAKER_01 44:07

Actually, it's my computer.

SPEAKER_00 44:10

Yeah. I think like all of us. Yeah.

SPEAKER_01 44:14

I I can't I can't do anything without a computer in my line of work, let's put it this way. So um yeah, that's that's the one weak thing. If people if if people know how often I actually am behind a camera, it um they would be surprised.

SPEAKER_00 44:29

Yeah, I think 90% of my waking life is is staring at a laptop drinking coffee. I think I think that's mostly what I'm doing throughout the day. Um here this is a fun one I like to ask people. So when you meet somebody for the first time, you're at a party and they ask you, what do you do? How do you describe your job to that?

SPEAKER_01 44:47

Oh, I just did that like an hour ago, which was kind of a super funny thing. Um so what I tell them is that I'm the hand of the digitization service at the Swiss National Library, and I'm responsible to get all the data as truth as possible to you that is actually in our magazines and that you can't see, but you want to see it online. And it's my job to make sure that it's actually the real data that's also on the object.

SPEAKER_00 45:12

That's that's really good, very succinct and very uh good, accurate technical description too. I know sometimes it's you know you tell people like, well, we take pictures of old books and things, and and it can be kind of hard to communicate. I think that's a really good answer. Um, okay, with our last few minutes here, I'm gonna turn it over to our audience. We have a couple of questions here I'm gonna give to you. And if anybody else uh uh that's joined us has questions from Martina, please use the QA function at the bottom of Zoom and submit it there, and and we'll pass it on. So um let's see, we've got a few here. Um beyond capturing the image itself, how do you organize and build the database in a way that's accessible to researchers? I mean, just very high level. We don't have to get too in the weeds, but like how do you approach that to make sure people can access the image once it's been captured?

SPEAKER_01 45:57

Well, it depends very much on the institution, of course, but uh most institutions have one or more platforms where they actually give the images plus all the metadata that surround the image, like the cataloging data or the uh full text from the OCR, if there is any uh full text, um, available on their platform in the viewers basically. If you are talking about research researchers, like um they want the real big massive amount of data that you actually create. Like, for example, the Swiss National Library has over 16 million newspaper pages online. If they want like even a portion of that, they don't want to download that article by article because it's all full text and OCR and it all has a lot of metadata. So they can actually ask for a data transfer. And then it it really depends on your technology in in the institution and what you want, APIs, um, or even sending it out on USB stacks or hard drives. I mean, everything's possible. Yeah.

SPEAKER_00 46:59

Yeah, actually that ties into another question here. Somebody's asking about when you when we're capturing documents, like things, things that have value beyond just the aesthetic image. What's the best way to save those? And and the person asking, like, do you save them as PDFs or do you save them as an image? I think you just touched on that, like extracting the information out of that. Um, when do you make that decision? Like, what type of documents do you make the decision on? And how do you save the document? Ultimately, what are we saving out of the document? What are the assets we get out of that?

SPEAKER_01 47:26

Well, for me, the the the biggest thing is always the picture because I can I can recreate all the data out of the picture. If the the picture is wrong or doesn't have all the information from the object, there is a problem. So so the picture is an image file um that does not per se need all the information, but within itself it has all the all the information. And then, of course, you have some medical and technical metadata, but that's a few things. So the cataloging you can actually um export and uh the cataloging with it, or you can connect it within your DAMS or with your the file naming signatures, barcodes, whatever not. And then what we do, for example, with newspapers is we are have we are sending them out to have them processed with uh zoning and OCR, and where we create derivatives like uh METs files, auto files, uh that's XML files, so machine readable, even we create um nice web pictures because the preservation images that we are creating are a little bit too big for the for the web. So we are creating nice pictures that can be viewed on the web. We are doing even PDFs to uh facilitate downloads, um, but all this can be done basically also on the fly, as long as I have a good image.

SPEAKER_00 48:46

Yeah, that's such a great point because I think you know, just make the point that the the the data you're gonna share is in the image. So make sure that again goes back. Image quality is the top priority always. And then everything else just kind of falls out of that. I think that's really well made point. Um we got, let's see, another question here. Image Okay, image metadata and original object metadata are not the same. How do you handle that difference with uh with AI or even manual workflow? Yeah, I guess we'll just I'll I'll leave that for you to to kind of fuss up. Do you do you change the workflow based on that? Do you use when using AI, do you take that into account?

SPEAKER_01 49:24

Um no, because I don't trust the AI to handle that.

SPEAKER_00 49:28

Yeah. I kind of kind of assume that. Um I guess we're we're kind of at the end here. Um I guess before we I close out, Martin, is there anything you want to share with our audience? Anything you just want people to know or or anything coming up that you're working on that you want to highlight?

SPEAKER_01 49:46

Well, um what what I want people to know um is basically the one thing that I started with. Um if you if you are already part of the community or if you are not part of the uh community yet and you want to be part of it, uh don't be shy. The community is extremely open, welcoming, sharing, and um welcoming for everybody. Uh if you are 20 years in the field or one day in the field, it doesn't really matter. Um, there are no stupid questions. And uh use the things that are there, like the gravel clubs, like the webinars, like the uh lists, go to conferences, build a great network because uh we are all in it together. And maybe we have already solved your problem or we have an idea how to solve it, or we know somebody who we can point you to, which is something that I I really, really feel very strong about that we are uh happy in this community to do that, and also um don't be afraid of the technical part of the whole things because I know targets and getting all the parameters right and everything like that. It can be daunting, but it's actually not that daunting if you find a good workflow for your production. Um you can start with very little knowledge, actually, in order to be able to produce great quality images, and then you can always go down and he rabbit hole in there. And there are many of those.

SPEAKER_00 51:25

Yeah, absolutely. I I really, really uh appreciate that message. That definitely resonates. I think again with DT, we we've also tried to build the sense of community and connect. I mean, that's why we're on this podcast right now, because we want to help share information. You know, to your point, a lot of times somebody's working on their own digitization lab trying to solve a problem, but that's been solved by other people in a different organization. So if there's a way to all we all connect, share information, um, and save everyone, you know, time, trouble, effort by by connecting and building that sense of community. So definitely echo that sentiment. Um, well, that's it. We're at time here. So, Martina, I just want to thank you so much for joining us. I really, really enjoyed the conversation, the experience of getting to talk to you and hearing your knowledge and thoughts. Um, I want to thank everyone that joined us live today for this podcast. Um, appreciate your questions, the comments, again, community, just giving people a place to connect and chat is great. For everyone that's uh signed up for this uh this webinar event, you will be receiving a video of this. I know sometimes with internet speeds, the video can you know kind of jump in and out. We will be sharing a full resolution video of this entire thing. So um don't worry about that, as well as the podcast, of course. Um we have other events coming up, um, so be sure to visit our website, uh, digitaltransitions.com. We have um August 6th, we're doing a live demo of DT's new Nexus software, which is really exciting. It's got some really, really kind of incredible game-changing features uh for digitization workflows. We're excited to show people, you know, in real time how that works. August 27th, we have another podcast recording with Sierra Orazzo, the curator of collection for the NASCAR Hall of Fame. That's really uh gonna be a really great conversation. I mentioned we did some AI sort of metadata image search stuff that was actually with NASCAR. Uh, we're really excited to talk more about that. Uh, we have another AMA Ask Me Anything session scheduled with Doug Peterson, DT's ahead of RD on November 19th. All of these are on our website under the events tab, so be sure to check them out. All the events are free. They're all gonna be online so anyone can join us from anywhere in the world. Uh, big thanks to DT coworkers here that are helping out, Wayne Casolino and Alana Ursia, as always. We very, very much appreciate your help. You you don't see them, but they're working hard behind the scenes and making sure that I have the information so that I sound uh like I know what I'm talking about. Um, and of course, look out for the official podcast release of this episode should come out in about a week. Uh, we'll be sharing out through all the normal channels and also be on our website and social media. So thank you again, Martina. And until next time, everybody, thank you for helping us focus a wider lens on the work that preserves our past and shapes our future. Thanks again.

SPEAKER_01 53:49

Thank you.

SPEAKER_00 53:51

Thanks for listening to Wider Lens, a podcast from digital transitions. We hope today's conversation offered practical insight into the evolving world of cultural heritage digitization, both the technical realities and the people behind the projects. If you have thoughts, questions, feedback, or ideas for future guests and topics, we'd love to hear from you. Reach out anytime at widerlens at digital transitions.com. And if this episode was valuable, please share with your cultural heritage friends, co-workers, and peers. Until next time, thanks for helping us focus a wider lens on the work that preserves our past and shapes its future.