At this point, we're hoping to open it up. We called this webinar project outcomes in future directions and part of the idea behind that was really kind of. Well, engaging, more broadly beyond the project itself, which is coming to an end as of tomorrow to think about where we've been and where we're going. As Joe mentioned, the project itself was part of the towards National Collection Initiative, which is UK based initiative thinking about ways to move towards a national collection. So I encourage you to put your questions in the chat or in the Q&A. We're happy to promote people to ask their question, live and participate in this fashion with all of our project members who are here today. It doesn't seem like there's any burning questions, so I think probably a good start is to return to Tom's presentation of the manifest editor and at one point he put infrastructure in big red letters right across the top. Perhaps that's a nice place to start. We've been talking about a lot of applications. But what do we need in terms of support for infrastructure in order to make more things possible? Tom, did you want to comment or?

Well, I mean yeah they. The tooling is one thing, the ability to kind of host and and and and and sustain hosting of IIIF Resources is another, and I think that's going to be the biggest obstacle to small institutions because there's an increasing amounts of good tooling, especially if you're building ad hoc manifests. You know if you just if you only have a few things or a few 100 things that you can build by hand, or even if you're kind of, you know you're using things like Mecca or things that can build small chip life. Elections. If there to be, if they're to exist in a national collection. Then you know they could get like completely over. I mean, well, there's a completely separate thing, which is probably not our remit in practical IIIF to to begin talking about. What exactly is a national collection? But assuming that there are IIIF things in it, which is why we're here, you could imagine that that you know the kind of effects of a lot of attention on very small institutions. Assets in national collection, and could be overwhelming and destructive. Even so, you know what?

What infrastructure are companies? These wonderful ideas now? Yeah, just as as as a company. We have lots of things about lots of ideas about that, but. Yeah, it's kind of seems to me an important question to to address like what's the support for small organizations and even medium sized organizations to actually host this stuff. Because I think some of the discussions we've had during the the the project is. I've looked at this as as Tom said, it's kind of out with the scope of what the project was aimed to do, but some of the tests and experiments that people in the consortium have been faking, particularly Glen, that you've been sort of working hours how people can host small numbers of assets directly on services like GitHub and how. Images could be put there and then used from those free services. Now what? Peter and Charlotte presented showed an alternate sort of approach is that if we're looking at the need for. Well managed and structured repositories.

So somewhere where we save our assets and then on top of that you also then get a IIIF service for presentation. But I think Tom's comment on if you get a couple of objects that then get hit by a million people then some kind of load balancing issues at scale. Out of interest, I just like to point out Joe that your your various kind of collection of browsing in in Open Sea Dragon and like putting 50,000 thumbnails into the same deep deep zoom views. That that's exactly the kind of denial of service attack on a small institution that that infrastructure needs to protect against. Yeah, I think I think it's quite easy. I mean most of those big collections, they the API's limit you to 10,000. I mean if you do if you do a search for flowers I think it get 3,000,000 hits from Smithsonian, but you can actually see them all. I think. I think there are sort of things in place to do that. But some people might want to see 3,000,000 and in years to come. There may well be use cases and requirements to to to do those sorts of things.

But yes, having a robust, reliable thing that's not going to kill one small institution would be very important there. Josh and Glenn, I've seen you nodding along. But either of you like Sherman. I may, I guess, yes, this is something that comes up a lot. You know, as I keep thinking about the arc of the IIIF community, it's it's sort of where we are as a community. We spent a bunch of years, just technologists, just trying to figure out the basics. And I think we're a little bit past that first phase and so now we're, you know, from our end from Glen and my perspective, we're really focused on training and documentation and like usable descriptive material that just helps people get started. So there are a lot of open tools. That there there is a lot. There are a lot of options. You know, Glenn, maybe you can talk in a second about you know what we've been looking at in GitHub?

Internet archive. There are a couple of free options, but yeah, it's a very real thing. Partly some of this will be solved by increased vendor adoption, so the more that people don't have to roll their own tools, the more barriers that will lower. And so yeah, it's just it's part of our focus. I think for the next few years for sure. Yeah, we've had a question pop into the chat that. So I think we'll continue the discussion a little bit, asking from an anonymous attendee what support might there be in the future for small museums that want to create IIIF manifests from their own digitized images rather than using assets that are already available and not format. Yeah, I think that's the key consideration, so you know I have been photographing pages of manuscripts or paintings and I've got some high res images. What do I do now? And you know there are. Yeah, there are two you can create.

IIIF image services by running tools and hosting them on GitHub. Although there are scaling limits with that, how many how many actual? Image tiles you could put there, but ideally you want to kind of you want a nice, easy, easy use service that will just take take those images and and turn them into F image services and then generate manifest for you. That link them all together. Then you could take that manifest into at all like manifest editor and refine it. Mix in messages from other sources and then you know just kind of establish a publishing workflow. That would that would be ideal. I think I think. Workflow is actually one of those things because quite often people we've talked to the the first question pops up is weird race start. There's lots of stuff out there, but where do I start? And I think.

Perhaps a cookbook or workflows might be interesting. I mean, that doesn't come with the manifest as such, but it sort of says these are some of the steps you could do if you start here. You could do this. You could do that. You could then consult this long list of potential commercial support systems to help you. But it is. There's a jump. I mean, if you're an individual researcher and you want to do a few things that are some free services out there. If you're a small institution and you've got 20 or 30 key objects that you would like to surface as you jump one step up. I have several thousand objects I'd like to do, then there's a certain internal institution investment that's going to be required, and that could be a service it could be with a company that's going to host things for you, or it could be an individual you know, internal institutional thing. But I would say just thinking back of what you said, Charlotte earlier is that what the project you've done with Peter allowed you to do is to demonstrate a real example for your own institution of what is possible.

So I think I think that's been very helpful. And it was. It was a big shame that we weren't able to get National Gallery content in there at the same time. It's just quite a a an intense sort of period of work. As Peter indicated that the the work was done quite quickly that we couldn't get that stuff in at that time people wanted to, but we have an ongoing discussion about IPR. It's it's. It's a common issue in institutions. Perhaps I could just elucidate slightly so. We did present accession from. A small collection of images for this pilot. And we used this intermediate Mongo DB based thing to organize the data and make reasonable data.

Site flavoured descriptions of these records, but where we are now with inventing our DM is that if you have an instance of it running. When you upload a record, when you create a record and you upload images associated with that object. It automatically creates a IIIF service when you press publish which is repository speak for makes this live. So we've already moved beyond the step that there's engineering involved. If small organizations simply build. An in venue RDM instance they can immediately create the sort of thing that we presented an hour ago. Probably we need a separate seminar to sort of go through this, because perhaps people haven't understood what we said. But it it's it's that then comes down to you. Still do need a a server on the web and I think I think it's that still a stumbling block for some of the people I suppose. Another question we've had in the chat. Umm from Matthew Westerby asking about what about infrastructure in terms of individuals or researchers that have some technical knowledge or self produced publications as separate days with different questions for persistence.

For example. Tropy has a IIIF. player in that it helps scholars publish collections. Peter, are you able to comment? You were saying that the current version of RDM that you're working on is being pushed into zenodo. So sorry what I've said. Someone has already started reimplementing zenodo using invenio RDM OK, So what I was wondering is what will that mean? Thinking of that question we just had if an individual researcher puts an image into zenodo. What sort of future IIIF functionality are they likely to expect? Well. What I meant to manage to go is that with.

The existing version 9 RDM instance. You you already get. Meridel 3IN Webpack has one possible preview option. And when someone completes reimplementation of zenodo then the main mainstream G7 global catchall repositories, anoda will have the same functionality. OK, so that may be an answer to that person's question. I mean the the other ones from a smaller scale. As we discussed, there are workflows as part of Glenn's training IIIF. If training is how you can put smaller numbers of images on Git help and have it host them in one of the simpler implementations of IIIF, so that is possible to do as an individual person. I don't actually know how to talk my head, how much it costs to buy a small AWS server. Or other similar servers to host a IIIF or an embedding instance oneself, but that then comes down to the scaling issues as Tom was mentioning before. Yep.

We have a further question in the chat from Beth Fisher. Stating that she's particularly interested in workflows that have connected manifest editing for outside users, students, and faculty, in their case with controlled vocabularies is anyone aware of project that's looked at or work being done in this space? So we did discuss the notion of controlled vocabularies in a relation to the manifest editor. I think that was one of the things that ended up in the that would be really nice to have Q rather than the immediate requirement. I know a number of projects that are looking at the concept of vocabularies in general there are. In existence, several vocabularies out there that can be used for referencing. For heritage materials, the Gettys vocabularies and obviously things like WIKI data can be used. Were you best thinking of how you connect them into the metadata fields within manifests? Or I'm not quite sure. How far down the rabbit hole you're wanting to go with that? Diego, I know you were involved in lots of our early discussions about sort of vocabulary around the manifest editor.

Whether we call things canvases and slide. Do anything you'd like to add. We we didn't have much time to explore that in terms of ideas in terms of solutions, you know that was kind of pushing the backlog to the very bottom. So yeah, no conceptual work has been done to date. Can I just clarify Beth's question there because it's about the connected manifest editing for outside users? Is that not like editing a Google doc? Multiple users editing at the same manifest at the same time? Is that what that means? Beth, we can promote you to actually ask the question directly if you'd like. She said oh put one in the chat. She clarifies we are trying to make it easier for external exports in our circuit to expand our data in ways directly connected to our images, so they'll be developing presentations, but we'd like to capture that for later in a useful format.

Experts not experts, so external experts in our circuit. Yeah, sorry if this also kind of brings in, uh, crowdsourcing as an option to start screen mentioned to me. Yeah, so and there are various crowdsourcing projects which are looking at kind of getting volunteers to kind of assign controls, vocabulary either kind of private ones or I think Tom and the folks have worked with Maddock and kind of linking it to Wikidata entities. So there's definitely kind of a history of using home control. Vocabulary is kind of like a crowdsourcing project. Is that what you can say something? Yeah, I know so just just to emphasize that on the very last trip life community call there was the Zooniverse presentation. So Zooniverse now, which is, you know, the world's biggest crowdsourcing platform. I think it's fair to say has IIIF support in it. So there are kind of. Yeah, there are.

There are lots of IIIF projects that would allow multiple people to contribute. You know tags or annotations that user control vocabulary on the same material at the same time. If if if that was. If that was the intent, I think yeah, I mean there are quite a few examples already of like using recoveries like Library, Congress subject headings or whatever to to tag things in live manifests. So I mean I don't have any links to hand, but there are quite a few examples of those kinds of projects. I think julianne's been contributing some examples on the chat. So, for anyone who's interested in that, definitely follow those links. Glenn, were you wanting to say how often you do those training sessions, which would sort of relate to people starting up and doing those small instances? So we're running them monthly at the moment and and I'm sure Josh put the link in the agenda, but I will try and dig it afterwards so we do go through some and free IIIF options and and some kind of paid for options. So we use the Internet archive as as a good way of hosting life images for free. They're always going to be downsides and to kind of free options, but the Internet archive is one of the options we use.

We've mentioned a few times this GitHub approach for tiles, which is great for kind of small projects, especially for demos. And again, there are some limitations to that. And then we've also talked about Cantaloupe, which is an open source image server and gives some instructions about, uh, installing that and so we focus first on the image API and then we move on to the presentation API and we use the body and manifest editor at the moment. Although we're looking at the Digitality manifest editor going forward and then we use kind of some free annotation tools that you can use to try out annotations and they're about. There are five day course and with. An hour long session each day, and then recordings, and to be able to catch you up with the rest of the stuff and and then on Fridays we usually have a project session where people demo kind of what they've learned during the week. No, I'll take out a link. Oh, I think yeah Josh put one in the chat. Along with a link from to Maddock from Julian, thank you. Yeah, OK, so are there any other questions people would like to air areas of discussion. People would like to look at.

Maybe yeah, sorry, just maybe while folks are typing in other questions I can address. There was a question in the Q&A that asked kind of a general question about accessibility. You know, we could obviously spend the rest of today talking about that topic, but really, at a high level I guess what I'll say is, well, one. We're working on a document that sort of is gonna echo some of what I'm about to say, but at a high level, basically AAA F. People have worked very well with good accessibility, right? So I'll put another way, like IIIF doesn't make or it doesn't help or hinder. Kind of the accessibility of a web application. The major viewers have been reviewed in terms of like Aria labels and other keyboard accessibility. So like universal viewer and Mirador have accessibility statements and they've been reviewed by places like national libraries which have often more. Engine requirements for those kinds of integrations. Yeah, and at a higher level like good web practices in terms of text based alternatives and alt text and all these different things work very easily with Triad in different contexts, and particularly in terms of AV materials that actually works really nicely in terms of having transcripts embedded alongside the audio that's native to treif.

And that's obviously best practice for accessibility. The one thing I'll say you know, like. Play I can't solve it for you if if you have other major accessibility concerns related to web pages, IIIF isn't gonna fix that for you, so it has to go hand in hand with a broader accessibility strategy and you know, paid development and attention to implementing those things, but hopefully at a high level that kind of answers the question. And then you know each project and each institution makes decisions about how these things get integrated and you know work with screen readers and all those sorts of things. OK, hopefully that answered that question. Thank you very much, Josh. Umm? One of the things I might ask a question. I don't think anyone's asked it. Tom, do you have an idea of when? The current development of the manifest editor will create a release.

As such, I mean, obviously you've indicated there's there's the plan is for it to be available for long term community development and such, but at a certain point. Do you have a sense that still fluid well it's? I would say we we're we're certainly aiming to have a a kind of coherent set of functionality that's pretty stable by at the time of the IIIF conference. Which is actually not very far away now, so that's the kind of beginning of June. But obviously you know, basically, you know we did a lot of research and a lot of analysis, and we've got a great big GitHub issue backlog and we want to kind of arrive at. You know, we don't want to do all that backlog, it's just there's just too much stuff in there, and it's all fantastic stuff. So we wanted to lose it, but it's kind of we have to kind of think, OK, where's the kind of where's the boundary? And so I think we'll yeah with that, we'll be aiming for. A kind of launchable thing by the IIIF conference. In yeah beginning of June. Which is not very far away now.

It's gonna worry me. It sounds like well beginning of June. It's like that's that's in the summer. That's months away, but it's not. At the moment only a month. OK. There were a few more comments in the chat, but I think they were information and links rather than additional questions. Do any of the other shout out to Glenn? I think people like your training Glen. I think you doing well though. Kind, thank you.

Denny, the panelists have any questions or any other issues you'd like to add? No. Sorry, just gone, I was just gonna put a link since you mentioned the conference I I probably should have mentioned it. But if folks are interested, this is exactly, you know. These are the kinds of conversations that move the community forward and so we're doing that June. Yeah, first week of June or June 7th and 8th is the formal context. There's a link in the chat there. Tom, I was gonna say it's kind of the kind of position of this work in the wider sort of tank activities, because what really interests me is just is that like? What exactly is a national collection and where does the discussion around that happen? So from the point of view of have you seen that here, here on a plate is an obviously useful set of technologies to make whatever a national collection might be. If it's vaguely digital.

Help that become a thing, but what is that thing that seems to me to be actually a much bigger question and much more important question than than the kind of technical particulars about it seems to be kind of social and political question, but I'm not sure where that discussion happens. Of the as I said at the start, we'll be putting together a final report for our project, and part of that final report will include recommendations and comments towards. Some of those issues the towards the national collection program, as I said, is continuing with five discovery projects, but there is also ongoing work from the program office itself to do smaller scoping projects, surveys gathering some of that information about might be needed, but also the larger funding infrastructure in use. The word twice landscape, shall we say in the UK, is looking at infrastructure funding. Of what might be needed to create a National Gallery, the the the tank project is towards so they were never actually sort of required for that particular program to create one, but they're effectively that the main point is to look at some of the issues involved and trying to bring people together to do more collaborative research. And I think we've quite happily showed that. IIIF you're gonna have to have images in a national collection, and you're very likely to have video and audio these days and very soon. Lots of 3D as well, so the option is to make use of this wonderful, supported community led robust technology or do exactly the same development for the next 15 years to do another one. I think it's kind of a no brainer really, but that's what we'll have to express in the final report. Is is that this technology works? It's developing.

Lots of options available from the storage point, right that we're down to an individual researcher outputting or just playing with existing technologies or outputting their own work. So it it has to be involved. Some of the discovery projects I think mentioned AAA F, but none of them are IIIF research, so I think some of them might make use of IIIF, but it wasn't a core flavor. I don't think in any of them there are a number of other projects that are mentioning F as as a key tool. And I think some of the work that Peter and Charlotte demonstrated. There's a big push on the sort of European and global level to try and make research information more accessible, so the ability to allow access to image outputs as a result of research. And this could be gigapixel images, it could be thousands of thumbnails or something. That requirement for the reuse of this content is becoming more and more required based on the funding streams. So if all of the research in the UK inheritage science requires the output to be available and reusable. Then part of the recommendation may well be that the funders need to sort of look at certain technologies and say. It will be more cost effective for you to make use of this valuable technology in your funding application and we will look favourably, but I can't put words into them.

The both of you. OK I so, but that's the sort of recommendation that we'll be sort of putting forward, is that? There are countries around the world which nod towards guided technologies or recommended technologies rather than required ones, and I think. The momentums growing more and more people, I mean, as as Josh said it's IIIF hasn't been here for that many years in comparison to the the time span of museums and galleries and archives, but it's expanded an awful lot in that time frame. And if we had a nice little graph we would be going upwards quite quickly. So the next set of five to 10 years of what can be done if UK are UK are infrastructure funding comes through. There's going to be a lot of interesting things to do and IIIF it's going to have to be a key component of it. There's that kind of focus on research infrastructure that kind of makes an assumption, or I'm not sure if it does make an assumption that the audience of a national collection. Comprises researchers that's that's kind of. Yeah, that's that's that's a fair point. But it's kind of more along the lines of.

If we think of Zenodo as well, CERN can offer this functionality of people uploading data there because they create so much data themselves, but everyone else's data. Is is manageable, and I think that's The thing is that if we start looking at. Outputting huge 3D environments, 3D spaces and more audio and video, and these sorts of things. If the research outputs need to be managed in this way, managing the other resources of smaller institutions may well. Be absorbed shall we say? But I don't know, I don't know. It's I think there were some discussions about programs like Art UK and whether they would offer certain options for smaller institutions. The difficulty is we use this term smaller institutions and I'm speaking from the National Gallery and I know how large our actual collection is and there are many small institutions that have got just as big or if not much bigger image collections than we might. So yes, a smaller institution may well have need of quite a bit of infrastructure. OK. I think there's a few more comments coming up, but there are comments rather than questions in the chat.

Umm? OK, well if we're coming towards the end then I would like to say in our webinars. This is our last webinar, but it's been really, really interesting and exciting and sort of transformative in certain aspects to to to do this research and this work with the people on the project. And I'd like to thank those who are here and also thank the the wider IIIF community because a lot of the things we've done have been on the back of conversations, and obviously Peter's not in the project. And we were able to do that work with Peter at the time in a collaboration, because the project was going along. So a lot of these things happen and do happen through conversation. So if people are still interested in trouble, if I would thoroughly recommend taking part of some of the discussion forums that the community offers and just ask silly questions. I mean, some of the work that we did came off the back. As I said of people say, oh, can you do X? And if people are working in that sphere and they're doing something similar, then it is possible solutions. Will appear as they're going along, so I think it is.

It's it would be really good for people to join in. Just ask questions and see you. There are very very nice group of people. All right? OK, there are some more comments popping up. Yeah, but thanks a few more comments, but. I think with that, uh, it's probably time to thank you all for participating. Thanks to our audience for their comments and their questions, as well as thank you to our panelists as well as all the project partners that we've had the privilege of working with. Over the past, well, technically, 22 months, but really much more like a year. Mobile COVID delays, but it has really been a pleasure. All of these recordings will be sent out to her and who registered for the conference.

You'll get a notification when they're ready as well as they'll be placed on to no, no as well as on our project website, so it should be there for you to access or to refer your friends to or you know, should you have enjoyed this just so much that you want to watch it all again, that's an option as well. With that told you good evening, good morning, good afternoon, wherever it may be. And thank you very much for attending all right.

Thank you.