1
00:00:08,220 --> 00:00:21,180
Thank you all for coming. This has been ARCHiOx, the Analysis and Recording of Cultural Heritage in Oxford, has been made possible by the 

2
00:00:21,180 --> 00:00:32,370
Helen Hamlyn Trust and really by Helen Hamlyn's extraordinary support of what Factum Arte and Factum Foundation does in Madrid.

3
00:00:32,760 --> 00:00:42,340
So for those of you who don't know, I am Adam Lowe, I set up the Factum Arte about 21 years ago.

4
00:00:42,360 --> 00:00:49,410
in Madrid and we were really building bridges between new technologies and traditional skills for contemporary artists.

5
00:00:49,770 --> 00:00:56,850
But in 2009, we set up Factum Foundation, mainly to work in the Tomb of Tutankhamun.

6
00:00:56,960 --> 00:01:05,560
So it seems a very nice symmetry to see the exhibition outside. Because now we're working all over the world so Ferdi Saumarez.

7
00:01:05,580 --> 00:01:14,810
Smith, who's here, is just back from Benin and Lagos, where we've been training a team to record Benin bronzes,

8
00:01:14,820 --> 00:01:18,450
a local Nigerian team to do high-resolution three dimensional recording.

9
00:01:20,520 --> 00:01:30,720
ARCHiOx started when I was talking to Lady Hamlyn in Venice, where we run our ARCHiVe.

10
00:01:30,750 --> 00:01:39,450
So there's a kind of branding exercise of rather laboured, whatever they call, naming.

11
00:01:39,990 --> 00:01:50,150
So ARCHiVe is the Analysis and Recording of Cultural Heritage in Venice and of course ARCHiOx is its Oxford partner. In Venice.

12
00:01:50,160 --> 00:01:57,330
the idea was that if we brought the technologies that were being developed in Madrid and we

13
00:01:57,330 --> 00:02:03,299
teamed up with École Polytechnique Fédérale de Lausanne, who were doing a lot of machine learning,

14
00:02:03,300 --> 00:02:10,800
a lot of data analysis, and we set up the structure within the Fondazione Giorgio Cini,

15
00:02:10,800 --> 00:02:18,240
which is one of the most significant art historical centres in Italy, then explosive things can happen.

16
00:02:18,750 --> 00:02:23,310
And the truth is over the last four years they have happened.

17
00:02:24,240 --> 00:02:29,910
So there's an enormous diversity of projects that have resulted from that collaboration.

18
00:02:30,900 --> 00:02:38,280
About three years ago, we started talking about what could be achieved if we started doing three dimensional recording of

19
00:02:38,280 --> 00:02:44,190
the surface of books and manuscripts and Lady Hamlyn's been a long time supporter of the Bodleian.

20
00:02:44,760 --> 00:02:56,820
And so the conversations began and John I, don't want to take words out of your mouth or put them into it, but the results I think beyond exciting.

21
00:02:57,390 --> 00:03:07,860
So there's a cascade of things and some of those things will be shown today, will be discussed about by the different speakers today.

22
00:03:08,520 --> 00:03:16,680
And I think there's been for many, many years a sort of lip service paid to the material culture of the book,

23
00:03:17,040 --> 00:03:21,179
the sort of book as a physical object and to me books,

24
00:03:21,180 --> 00:03:30,149
manuscripts, things are completely magical, the way they contain information and the way they release their information.

25
00:03:30,150 --> 00:03:40,560
So normally the digitisation of books is the extraction of the text they contain to be put in a digital form which can be accessed anywhere.

26
00:03:40,830 --> 00:03:44,690
So this is of course, their primary function, the things which are in them.

27
00:03:45,420 --> 00:03:51,209
But actually what you're going to see today is there are many elements about the

28
00:03:51,210 --> 00:03:56,760
material nature of that object that if you can record the three dimensionality,

29
00:03:56,970 --> 00:04:04,440
the surface. So if you think back in paintings when suddenly x rays happened, there were major discoveries all over,

30
00:04:04,830 --> 00:04:09,690
then infrared, major discoveries, then ultraviolet and other things.

31
00:04:09,690 --> 00:04:13,200
So people start understanding the object more.

32
00:04:14,100 --> 00:04:24,000
But it's amazing that this is the first time that the surface of books and manuscripts and maps and objects has been recorded in a major library.

33
00:04:24,900 --> 00:04:28,770
And that's something we're all incredibly proud of.

34
00:04:28,950 --> 00:04:34,829
So what's coming out of it, I think,

35
00:04:34,830 --> 00:04:45,090
is going to be as significant and important as the research ushered in by x rays and infrared and multispectral photography.

36
00:04:45,570 --> 00:04:50,459
And that's just by looking at the surface, looking at what is contained on it.

37
00:04:50,460 --> 00:04:53,010
And today I don't want to pre-empt any of the talks,

38
00:04:53,010 --> 00:05:00,210
but there are some things that the media is getting very excited about at the moment, and there are some big discoveries.

39
00:05:02,070 --> 00:05:07,530
But after eight months of working on what's been one of the most enjoyable

40
00:05:07,890 --> 00:05:15,450
collaborations Factum Foundation has been involved in and one in which I think the team has really gelled.

41
00:05:16,110 --> 00:05:24,180
There's a few people who need to be thanked I think above and beyond and obviously, Jorge Cano, who's going to be talking at the moment,

42
00:05:24,930 --> 00:05:28,710
has spent the last five years, I mean, Factum's, a mad place.

43
00:05:29,040 --> 00:05:37,840
It's something a bit like a playground full of dysfunctional children in which all sorts of things can happen.

44
00:05:37,860 --> 00:05:49,200
So it's it's sort of the opposite of a professional structure, but it's totally focussed on the love of making things.

45
00:05:49,710 --> 00:05:59,040
And Jorge has been researching, as I say, for five years into photometric stereo, so he's going to talk about what he's doing.

46
00:06:00,510 --> 00:06:12,600
But bit by bit we've managed to produce a system that is able to capture the surface of the amazing things in the Bodleian at about 25 microns.

47
00:06:12,990 --> 00:06:20,190
So suddenly you see things you never saw before. And where I hope this initiative will go

48
00:06:20,650 --> 00:06:27,330
Is that there will be a second year in which we can build on all the things that have started and we

49
00:06:27,330 --> 00:06:31,770
can finalise all the research and development because everything's been experimental,

50
00:06:32,550 --> 00:06:37,590
but we can finalise the research and development into something really very solid,

51
00:06:38,130 --> 00:06:43,830
very stable, and something that could benefit museums and libraries around the world.

52
00:06:44,340 --> 00:06:50,820
So I think what we've achieved is exciting.

53
00:06:51,830 --> 00:06:56,000
But what's still to be achieved is even more full of potential.

54
00:06:56,600 --> 00:07:02,780
And if I think it's the projects that I would love to do, it's not just with books.

55
00:07:03,260 --> 00:07:11,450
So once you can do the recorded books and manuscripts and of course the Bodleian is this amazing resource

56
00:07:11,450 --> 00:07:15,710
which needs to be tapped in many different ways by many different people.

57
00:07:16,040 --> 00:07:27,840
But I really want to look at the collections at Oxford to set up a project to record with a Nigerian team, all the Benin bronzes that are here.

58
00:07:28,400 --> 00:07:37,129
We want to look at recording the facade of Iffley Church, which is one of the great Norman facades, which is in a critical state.

59
00:07:37,130 --> 00:07:43,280
So now it's got a sort of pebble dash on it that is holding it all together.

60
00:07:43,760 --> 00:07:47,989
But the question about how we preserve things, what we're trying to preserve,

61
00:07:47,990 --> 00:07:56,720
how we use the technology to shape the things we value and to ensure they're there for future generations.

62
00:07:57,230 --> 00:08:04,580
And yesterday with Simon Schaffer, who's here and Stephen Johnston,

63
00:08:04,580 --> 00:08:14,960
we were looking at a collection of scrapbooks effectively in the Museum of the History of Science that came from Hartwell House,

64
00:08:15,530 --> 00:08:17,390
and they've never been recorded.

65
00:08:18,020 --> 00:08:26,009
And I think in those scrapbooks, there's a vast amount of information that's going to colour the way we see, for example,

66
00:08:26,010 --> 00:08:36,800
from John Martin's paintings, early celestial observation, early collecting and antiquities and the kind of networks of people who come together.

67
00:08:37,310 --> 00:08:50,840
So what I hope is going to come out of of ARCHiVe is that there's a way of bringing people in to the sheer excitement of what's contained in the Bodleian.

68
00:08:51,320 --> 00:08:59,000
I mean, for me personally, I was a student here at the Ruskin, and the Bodleian was always strangely impenetrable.

69
00:08:59,960 --> 00:09:10,940
So, you know, without having that direct access or studying something specific, the card index is how you found things sort of kept me away.

70
00:09:11,480 --> 00:09:16,500
But if you think now, when you're searching on your computers down the right hand bar,

71
00:09:16,520 --> 00:09:22,280
Google has machine learning algorithms that are pointing you in directions that you didn't know.

72
00:09:22,820 --> 00:09:32,690
The potential for an iconographic way of searching libraries is absolutely enormous, and I hope that's what we're going to see in the talks.

73
00:09:32,990 --> 00:09:38,270
So I more or less everyone is there to introduce themselves. John, I think you're going to begin with the panel here.

74
00:09:38,780 --> 00:09:45,590
But I really hope this morning that the excitement we all feel filters through and can infect you a little.

75
00:09:46,490 --> 00:09:54,860
Thank you. Thank you, Adam.

76
00:09:54,920 --> 00:09:56,040
I'm John Barrett.

77
00:09:56,060 --> 00:10:04,490
I'm the Senior Photographer at the Bodleian Libraries, and I've been using ARCHiOx technology over the last ten months to record our originals in 3D.

78
00:10:05,240 --> 00:10:09,110
So it might not seem obvious how library material and 3D recording go together.

79
00:10:09,530 --> 00:10:12,830
After all, library metal material is flat.

80
00:10:13,160 --> 00:10:18,890
But for the recordings made for ARCHiOx, it's just the first few millimetres of the surface of an original which is recorded.

81
00:10:19,340 --> 00:10:26,200
In other words, just the texture. These 3D recordings serve to purposes.

82
00:10:26,530 --> 00:10:33,220
Firstly, Discovery - taking an item which is difficult or perhaps impossible to record using conventional photography

83
00:10:33,730 --> 00:10:40,090
and creating an image which shows its three dimensional surface, with its colour and tone completely removed.

84
00:10:42,110 --> 00:10:47,660
 

85
00:10:49,400 --> 00:10:53,420
This is an image of a copper printing plate, possibly from the early 18th century.

86
00:10:54,200 --> 00:10:59,270
Over 300 years, it's become heavily corroded and the design has become difficult to see.

87
00:11:02,890 --> 00:11:09,010
But if we remove the tone and colour and instead look at a render of the 3D surface of the plate, the engraving is revealed.

88
00:11:11,120 --> 00:11:17,850
What's even more exciting is to turn this printing plate over and record the back. Copper plates with valuable and often reused.

89
00:11:18,260 --> 00:11:22,100
And we know very little about the front of this particular plate, let alone the reverse.

90
00:11:22,880 --> 00:11:26,660
From the colour image, we can see some horizontal lines, but not a lot more.

91
00:11:28,920 --> 00:11:38,210
But from the 3D recording, we can begin to see markings which are difficult or perhaps impossible to see from the conventional recording.

92
00:11:39,230 --> 00:11:48,040
Musical notes and words. And if we add some of the original colour back in.

93
00:11:49,700 --> 00:11:53,060
And enhance and mirror the printing plate to make it easier to read.

94
00:11:53,630 --> 00:11:58,960
We are now able to see enough details. That we can annotate over the markings.

95
00:12:00,770 --> 00:12:02,300
Which we can then tidy up.

96
00:12:03,970 --> 00:12:11,710
And finally, with the image removed, we can now see the full design, a piece of music which is likely to have been inspired by Psalm 9.

97
00:12:12,710 --> 00:12:15,650
A brand new discovery made through 3D recording.

98
00:12:19,310 --> 00:12:26,360
But the data which is being captured isn't just being used to make 3D renders, 2D images which show the surface of an original.

99
00:12:26,810 --> 00:12:35,570
The recordings are truly three dimensional. This brings us to the second purpose for making 3D recordings, the possibility to create 3D facsimiles.

100
00:12:37,820 --> 00:12:41,690
This is another copper printing plate, featuring antiquarian Anthony Wood.

101
00:12:42,110 --> 00:12:49,550
Again, it's difficult to see. And here's the render of the surface of the same plate.

102
00:12:49,760 --> 00:12:51,890
Again, the design has been revealed brilliantly.

103
00:12:54,110 --> 00:13:00,020
But by loading a single greyscale image file into software which allows us to view the file in 3D,

104
00:13:00,380 --> 00:13:05,450
it's possible to see the recording in a different way. Now we can move over the surface of the recording.

105
00:13:05,780 --> 00:13:08,840
We can zoom in and explore every engraved line.

106
00:13:11,480 --> 00:13:14,210
And with this recording, it's possible to create a facsimile.

107
00:13:14,870 --> 00:13:20,750
This is a 3D printer in The Factum Arte studios in Madrid, laying down incredibly fine layers of resin.

108
00:13:22,650 --> 00:13:31,620
And slowly building a replica printing plate. And by making a facsimile printing plate, it's possible to use it to produce facsimile prints.

109
00:13:33,260 --> 00:13:38,749
Here is the inked facsimile plate. And here's Adam pulling it through Factum's

110
00:13:38,750 --> 00:13:48,930
Replica replica Goya printing press. And this is the resulting print made from a copy of a printing plate which was engraved 313 years ago.

111
00:13:52,870 --> 00:13:59,710
What makes ARCHiOx so exciting is the hardware and workflow being used to make these 3D recordings is unique.

112
00:14:00,250 --> 00:14:04,270
Although the newest of the recording systems designed and built by the Factum Foundation,

113
00:14:04,570 --> 00:14:09,250
has been developed over several years, the Bodleian Library are the first institution to use it.

114
00:14:10,370 --> 00:14:15,380
But the Factum Foundation have spent over a decade recording with another recording system called the Lucida,

115
00:14:15,860 --> 00:14:19,850
which has also been used for making ARCHiOx recordings. And which Carlos

116
00:14:20,150 --> 00:14:26,420
will now describe. Thank you, John.

117
00:14:27,770 --> 00:14:35,209
Good morning, everybody. Before Jorge will explain in detail the other of the main technologies happening here in ARCHiOx, I'd like to offer

118
00:14:35,210 --> 00:14:43,460
a very quick review of what has been Factum Foundation's continuous research into recording the painted surface.

119
00:14:43,460 --> 00:14:55,130
And this is what has become the Lucida project mainly, so I'm Carlos Bayod, I'm in charge of 3D scanning in Factum. So for 21 years

120
00:14:55,400 --> 00:15:02,810
we've been adapting existing recording technology for helping preservation of cultural heritage and works of art in general.

121
00:15:03,260 --> 00:15:09,040
Back in 2001, we were using laser scanning technology with specific adaptations to demonstrate

122
00:15:09,070 --> 00:15:14,209
it was possible and relevant to capture the reliefs of the Tomb of Seti I.

123
00:15:14,210 --> 00:15:22,610
The idea was to obtain information about the relief, so colour and relief were captured with different methods, with different systems.

124
00:15:23,240 --> 00:15:30,410
What we were really trying to prove was that through high resolution recording it was possible to understand the condition of the relief.

125
00:15:31,040 --> 00:15:37,999
So over the years we have been using laser scanning technology, adapting it to different objects.

126
00:15:38,000 --> 00:15:45,260
For example, when recording pages of manuscript of the Beato de Liébana in the Spanish National Library in Madrid,

127
00:15:46,250 --> 00:15:50,450
trying to capture the undulations of pages, but also incisions, etc.,

128
00:15:50,900 --> 00:15:59,059
then trying to apply 3D scanning laser technology onto the surface of sculptures to capture not just the shape but also the skin.

129
00:15:59,060 --> 00:16:02,150
I mean, the information present on the actual surface texture.

130
00:16:02,690 --> 00:16:08,420
And, trying to apply this type of technology to volumetric objects like sculptures is really limited.

131
00:16:08,690 --> 00:16:15,890
So this kind of scanning technology works better into what we could consider two and a half dimension objects.

132
00:16:16,250 --> 00:16:21,680
So things that are really flat in shape, but in which the actual relief information is relevant.

133
00:16:22,370 --> 00:16:28,130
Of course, we've been applying this technology for recording the walls of the funeral chamber of Tutankhamun.

134
00:16:28,760 --> 00:16:39,049
But then when we tried to apply this method, existing recording methods for the surface of paintings, this is where it started to be problematic.

135
00:16:39,050 --> 00:16:46,070
The surface of paintings is one of the most complex things, precisely because of the glossy surface present because of the varnish,

136
00:16:46,070 --> 00:16:50,210
for example, or because of the contrast between clear and dark areas.

137
00:16:50,750 --> 00:16:58,730
This is usually making it very hard or almost impossible for standard 3D scanners to capture useful information out of paintings.

138
00:16:59,750 --> 00:17:07,490
This is why in 2010, working with artist Manuel Franquelo, we decided to create our own 3D scanner from scratch,

139
00:17:07,970 --> 00:17:14,900
specifically aimed at capturing the surface of paintings and other low relief objects.

140
00:17:14,990 --> 00:17:20,690
The Lucida 3D scanner is working by projecting a laser line on the surface of the painting.

141
00:17:21,170 --> 00:17:27,290
And then there are two video cameras that are capturing the distortions of the line as the scanner moves across the surface.

142
00:17:27,650 --> 00:17:37,010
This is providing a depth map. Recording depth information, it's creating topographic data out of paintings that are supposed to be flat objects.

143
00:17:37,010 --> 00:17:44,690
But thanks to this type of technology, it is possible to actually see them as the actual three dimensional objects they are.

144
00:17:45,860 --> 00:17:49,969
The first time that this camera was used in a public museum, was in the Museo del Prado,

145
00:17:49,970 --> 00:17:55,490
for recording this panel painting by Rubens that contains all the difficult things that are usually,

146
00:17:55,790 --> 00:18:01,430
you know, almost impossible for 3D scanners to capture, glossy, surface contrasting areas, etc.

147
00:18:02,210 --> 00:18:05,300
But we wanted to obtain this type of information.

148
00:18:05,300 --> 00:18:14,720
We wanted the scanner to ignore the colour for once and just look at the topographical landscape that a painting is actually formed by.

149
00:18:15,590 --> 00:18:21,530
We are talking about high resolution, 100 microns of resolution for this type of 3D information.

150
00:18:21,740 --> 00:18:27,860
So it is possible to understand the brushstrokes, I mean, the artist's technique, but also the way paintings have been ageing,

151
00:18:28,220 --> 00:18:35,180
have been evolving over time through cracks and other features on the surface. The development

152
00:18:35,180 --> 00:18:42,020
continued, Jorge Cano of was actually part of the software, mainly the software running the Lucida.

153
00:18:42,290 --> 00:18:43,969
And then with this final prototype,

154
00:18:43,970 --> 00:18:49,490
we have been using it for recording hundreds of paintings in the main museums, from the main collections around the world.

155
00:18:49,730 --> 00:18:52,520
This is one of the systems that are part of the ARCHiOx project.

156
00:18:53,120 --> 00:19:00,589
and in combination with the Selene scanner, it's making it possible to capture the objects here in the Bodleian.

157
00:19:00,590 --> 00:19:02,959
And then with this final prototype,

158
00:19:02,960 --> 00:19:15,110
we've been returning to Seti Tomb where we've been able to capture the entire area of reliefs inside this monumental site and then

159
00:19:15,110 --> 00:19:22,850
providing information of the relief and colour separately that we could use to understand the actual condition and to eventually

160
00:19:23,090 --> 00:19:27,650
create a reproduction that will help effectively the preservation of the monument.

161
00:19:29,330 --> 00:19:29,959
Just to conclude,

162
00:19:29,960 --> 00:19:38,090
I will mention that the Lucida scanner was used a few years ago here at the Bodleian Libraries for recording the surface of the Gough Map.

163
00:19:38,720 --> 00:19:45,680
The colour was provided actually by John Barrett. Then the topographic information was recorded thanks to the Lucida scanner.

164
00:19:46,040 --> 00:19:52,640
So it is thanks to this first exercise that it was possible to see this map under a new light.

165
00:19:52,910 --> 00:20:00,379
Understand that it's not just flat information, but something that, thanks to the three dimensional recording,

166
00:20:00,380 --> 00:20:07,160
it is possible to understand even better, its material nature and its historical trajectory.

167
00:20:07,970 --> 00:20:11,450
The Gough Map will be discussed later by Damien and other speakers,

168
00:20:11,450 --> 00:20:19,279
but just as a quick thing, thanks to the 3D scanning project that was carried out a few years ago,

169
00:20:19,280 --> 00:20:25,490
it was possible to identify and to visualise even better the presence of a series of pouncing marks that would shed,

170
00:20:25,700 --> 00:20:28,070
as I say, new light into its history.

171
00:20:29,420 --> 00:20:36,830
Not far from here in the Ashmolean Museum, we had the opportunity of recording small drawings by Raphael and then again pouncing marks,

172
00:20:37,130 --> 00:20:44,150
which are actually revealing a history of copying the history of translating, transferring information from one original to the next.

173
00:20:44,450 --> 00:20:49,280
It was possible to see it much better, thanks to this act of removing the colour,

174
00:20:49,430 --> 00:20:53,750
virtually speaking, and then looking at the pure relief of the texture.

175
00:20:55,040 --> 00:21:01,790
And finally talking about Rafael as well. But at the much larger scale, at a much more monumental size,

176
00:21:02,090 --> 00:21:05,629
we had the opportunity of applying the Lucida scanner for recording the seven

177
00:21:05,630 --> 00:21:09,230
cartoons by Rafael that are in the Victoria and Albert Museum in London.

178
00:21:09,830 --> 00:21:15,530
This was the first time that this series of artworks were recorded in 3D, of course,

179
00:21:15,530 --> 00:21:24,020
but also as part of a systematic recording with colour and infrared photography, so that details like this can be perceived with colour,

180
00:21:24,290 --> 00:21:25,970
with a combination of colour and relief,

181
00:21:26,300 --> 00:21:33,980
and only with relief revealing these pouncing marks and many other features that are kind of very hard to see in front of the original.

182
00:21:35,390 --> 00:21:39,500
I will now give my screen to Jorge. So thank you very much.

183
00:21:49,670 --> 00:21:53,690
Good morning, everyone. It's a pleasure to be here today in this work.

184
00:21:53,930 --> 00:21:57,870
My name is Jorge Cano and I am Head of Technology at Factum Foundation.

185
00:21:58,350 --> 00:22:03,319
And first of all, I would like to thank all the people who made this project possible,

186
00:22:03,320 --> 00:22:07,040
especially to John Barrett, who has been the perfect companion in this path.

187
00:22:07,610 --> 00:22:10,819
Through his great work and enthusiasm. He has made the methodology.

188
00:22:10,820 --> 00:22:14,390
we brought his own, and has made the best out of it.

189
00:22:16,840 --> 00:22:22,780
Okay. As my colleague Carlos has explained,

190
00:22:22,900 --> 00:22:27,490
the fact of accurately studying the relief of objects that at first sight may seem

191
00:22:27,490 --> 00:22:32,980
flat can lead us to discover things that we had previously overlooked. In this slide

192
00:22:33,010 --> 00:22:39,340
You can see a few technological milestones at Factum. We can see that the processes and technology we are currently using at

193
00:22:39,550 --> 00:22:43,390
ARCHiOx are the results of more than 20 years of work at Factum.

194
00:22:45,890 --> 00:22:56,240
So if we go back to 2021, the question is, did we have the necessary recording system to undertake a project like ARCHiOx?

195
00:22:57,070 --> 00:23:04,120
Not really. Therefore, during the past year we focussed on testing and finding the perfect system for the project.

196
00:23:04,690 --> 00:23:07,840
But the next question arises, is that possible?

197
00:23:08,950 --> 00:23:13,930
Sadly, the perfect 3D recording system does not exist. Each of them has its limitations.

198
00:23:14,680 --> 00:23:21,160
For example, in this slide we can see the first three systems we tested the Lucida scanner,

199
00:23:21,490 --> 00:23:25,360
close-range photogrammetry and a very expensive, confocal profilometer.

200
00:23:26,080 --> 00:23:31,120
Neither of the first two gave us the expected results in terms of spatial resolution.

201
00:23:31,570 --> 00:23:36,550
And the third one provided great resolution, but was too slow and didn't provide colour information.

202
00:23:37,240 --> 00:23:42,850
So at this point, we began to consider other alternatives that we had experimented with in recent years.

203
00:23:43,480 --> 00:23:47,890
RTI and photometric stereo. Before going on.

204
00:23:48,220 --> 00:23:54,970
One should ask oneself, is the difference between 120 microns resolution so important that

205
00:23:54,970 --> 00:23:59,620
we have to complicate our lives by switching to another technology or develop a new system?

206
00:24:00,730 --> 00:24:05,080
The short answer is yes, and the long one can be found in the following slides.

207
00:24:07,560 --> 00:24:11,340
Here we can see the render of the Lucida scanner at 254dpi.

208
00:24:12,000 --> 00:24:20,430
The object is a copper plate and the image on the screen represents a small section, more or less four by four centimetres.

209
00:24:22,280 --> 00:24:28,640
On the next one we have a render of a photometric stereo scan, it was done in 2021.

210
00:24:29,390 --> 00:24:32,170
while preparing the system to be installed at the Bodleian.

211
00:24:33,080 --> 00:24:43,010
It's a 750dpi capture and it shows perfectly how increasing the spacial resolution helps to see details that otherwise would be hidden.

212
00:24:45,800 --> 00:24:54,150
This is a side by side comparison. Lucida left, Selene, or first prototype of the Selene on the right.

213
00:24:55,920 --> 00:24:58,170
Coming back to RTI photometric stereo,

214
00:24:58,230 --> 00:25:06,030
we can say that both techniques are great if the aim is to virtually and interactively reveal the characteristic of an image surface.

215
00:25:06,780 --> 00:25:11,610
But for us, the goal is not to end up producing a virtual object on a screen.

216
00:25:11,910 --> 00:25:22,200
The goal is to recreate the real characteristics of the object most of the time, building up a real solid replica of it. By making some assumptions,

217
00:25:22,290 --> 00:25:26,460
like considering that the shape of the source object is almost flat.

218
00:25:27,090 --> 00:25:32,130
We can use the information given by both techniques to calculate the real geometry of an object.

219
00:25:34,350 --> 00:25:38,880
At the end, we decided to use photometric stereo.

220
00:25:39,270 --> 00:25:47,700
And we made this decision because after several tests we obtained better 3D reconstructions using fewer lights compared with RTI.

221
00:25:51,070 --> 00:25:54,430
Therefore in the end, why did we choose photometric stereo?

222
00:25:55,420 --> 00:26:01,870
There are three reasons. One, we need very high spatial resolution in the order of 25 microns per pixel.

223
00:26:03,550 --> 00:26:10,000
Two, colour scanning is fundamental, so we need to scan both relief and colour, ideally at the same time.

224
00:26:10,630 --> 00:26:15,760
And third and last, if we were to be successful, it was necessary to record large

225
00:26:17,070 --> 00:26:22,050
amounts of data. So the system had to be very fast in capturing and post-processing.

226
00:26:23,280 --> 00:26:27,450
So photometric stereo was clearly the technique that could fit our needs.

227
00:26:27,750 --> 00:26:33,240
And the Selene scanner is the physical outcome with which we have implemented this technology.

228
00:26:33,600 --> 00:26:41,340
We will see it in detail a bit later. But before, I would like to give you some technical requirements to better understand

229
00:26:42,680 --> 00:26:48,440
The photometric stereo principle and the workflow. First, there is one important choice we had to make.

230
00:26:48,980 --> 00:26:52,880
Are we going to store the 3D data into 3D meshes or inside images?

231
00:26:53,600 --> 00:26:58,790
We prefer raster maps. Second, how are we going to store information inside images?

232
00:27:00,280 --> 00:27:05,170
I'm going to explain that with two examples depth maps are normal maps.

233
00:27:06,040 --> 00:27:10,630
And third, what is frequency separation and why is so important in this project?

234
00:27:12,040 --> 00:27:23,620
One of the first and most important decisions to be made when designing a 3D capture system is how the data will be stored.

235
00:27:24,280 --> 00:27:31,210
If we are talking about 3D data, we are very used to seeing images like the one on the left, a wireframe view of the model.

236
00:27:31,900 --> 00:27:37,750
However, we have decided to store the data in the way you can see on the right using what is called raster maps.

237
00:27:38,290 --> 00:27:42,670
And this is a fancy name for a single image. In this case, a greyscale one.

238
00:27:46,890 --> 00:27:54,120
And why have we chosen that? There are several reasons, but the most important one is that we can achieve faster simulations.

239
00:27:54,450 --> 00:27:59,460
For example, calculating, a shaded version of a 3D model is very fast.

240
00:28:05,420 --> 00:28:13,520
Also imagine that we have a model with several million 3D points and we want to store it, keeping the same number of points.

241
00:28:14,270 --> 00:28:20,390
The file saved as an image can be 10 to 20 times smaller than the one stored as a 3D mesh.

242
00:28:21,470 --> 00:28:27,800
Finally, the fact of working in a raster format allows us a lot of versatility in terms of the tools to use.

243
00:28:28,330 --> 00:28:35,780
If you come from the world of photography, you can use imaging software like GIMP or Photoshop.  Or in the case of coming from another fields,

244
00:28:35,780 --> 00:28:42,920
such as archaeology or cartography, we can use geographic information systems such as GRASS or ARCGis.

245
00:28:44,850 --> 00:28:48,240
We have seen the advantages of working with images to store data.

246
00:28:49,950 --> 00:28:56,670
We can go into more detail. I think everyone knows how about how an image is formed.

247
00:28:57,690 --> 00:29:10,860
We store numbers inside an array of pixels, in this case it's in 8-Bit resolution, in greyscale so we can store values from 0 to 255.

248
00:29:13,050 --> 00:29:17,220
Basically, we usually store the brightness of the image.

249
00:29:17,280 --> 00:29:23,790
But what if instead of brightness, we want to represent another property of the object?

250
00:29:27,200 --> 00:29:35,180
So, for example, if the depth of an object is known for each pixel, we might say that zero represents the deepest.

251
00:29:35,300 --> 00:29:45,990
and 255 represents the closest. And in this case, black means a deeper area, equal to zero millimetres and white.

252
00:29:46,100 --> 00:29:49,490
The peak of the surface equal to nine millimetres.

253
00:29:52,490 --> 00:30:03,110
The last one was a depth map. And now we can see that raster maps are not only for height information, it could be used for any variable.

254
00:30:03,620 --> 00:30:14,260
Another interesting raster map is the gradient map. To simplify, let's focus on a simple shape, like a sphere.

255
00:30:14,890 --> 00:30:18,520
What does a gradient map look like in that case?

256
00:30:20,010 --> 00:30:24,870
In that case, the horizontal gradient map indicates for each pixel,

257
00:30:24,880 --> 00:30:32,560
whether the surface is pointing to the right or to the left, that is, if it points to the right, we draw a white pixel.

258
00:30:32,590 --> 00:30:41,590
If it points to the left, we draw black. Having seen the above, it is now easier to understand what the normal map is.

259
00:30:42,340 --> 00:30:46,360
Normal maps use a colour image to store the gradients of the surface,

260
00:30:46,510 --> 00:30:51,550
the red channel has the horizontal gradient data and the green channel has the vertical gradient.

261
00:30:52,870 --> 00:30:59,020
In other words, the normal map stores the orientation of the surface inside an RGB image.

262
00:31:02,490 --> 00:31:08,220
Among other things. It allows us to re-light an object virtually.

263
00:31:09,130 --> 00:31:18,810
Later, I think we can see great examples. And the third and final important technical concept is frequency separation.

264
00:31:20,270 --> 00:31:27,680
We all know that we can add waves. We can start with a sine wave like the one shown on the screen.

265
00:31:31,490 --> 00:31:38,110
Continue with another higher frequency wave. And add the two frequencies together getting what we see on the screen.

266
00:31:39,950 --> 00:31:45,380
But the most interesting thing about this is that it also works the other way around.

267
00:31:45,950 --> 00:31:55,730
Moreover, instead of abstract waves, we can work on the shape of objects as if they were waves, or, in other words, work in the frequency domain.

268
00:31:57,280 --> 00:32:02,470
Starting with a surface like the one we see on the screen. We can decompose the shape into two.

269
00:32:02,950 --> 00:32:07,180
general topography, or low frequency, and fine relief, or high frequency.

270
00:32:10,940 --> 00:32:21,740
The advantages of working this way are that we can clean noise, merge data from different systems and focus on important details.

271
00:32:22,640 --> 00:32:32,030
Just as an example, I would like to show you the Gough Map and how we used frequency separation to merge the data coming from two techniques.

272
00:32:32,210 --> 00:32:43,670
On one hand Lucida. On the other hand, Selene. Lucida provides precise low frequency relief data, and Selene the high frequency ones.

273
00:32:44,360 --> 00:32:49,430
The final model has more than 800 million pixels.

274
00:32:52,060 --> 00:33:03,680
I'm going to move to explain a little bit more about photometric stereo. It is based on Shape From Shading, an article written by Horn in the seventies.

275
00:33:05,150 --> 00:33:12,980
Basically in that article the question was can we recover the shape of an object based on its appearance, how it looks?

276
00:33:13,490 --> 00:33:16,640
The answer is yes. And the basis is Lambert's Law.

277
00:33:20,300 --> 00:33:26,330
But what is photometric stereo? It is a technique to create a model of a scene.

278
00:33:26,810 --> 00:33:34,580
What is a model? This is a topic that I find extremely interesting and could be talked about for hours.

279
00:33:35,180 --> 00:33:40,280
Cartographers know a lot about this. The history of maps is very rich in this sense.

280
00:33:40,790 --> 00:33:47,090
They know how difficult it is to represent reality and the need for condensation underneath for clarity.

281
00:33:48,100 --> 00:33:54,790
In this sense, I believe that the key lies in the necessary balance between the infinitely complex reality and the conscious and

282
00:33:54,790 --> 00:34:01,390
prior decision that must be taken to decide on which properties you want to focus the model you are building.

283
00:34:06,380 --> 00:34:12,260
In our case, we want to create a model of an object through several pictures.

284
00:34:15,220 --> 00:34:18,340
Moving the light. We end up with a group of pictures.

285
00:34:19,500 --> 00:34:22,530
And through a reconstruction, we produce a model.

286
00:34:23,130 --> 00:34:28,380
The goal is to end up with a model of the object as close as possible to the real one.

287
00:34:34,150 --> 00:34:42,280
In the photometric stereo case, the model extracts two properties of the object or scene. The shape and the albedo. The albedo

288
00:34:42,610 --> 00:34:47,710
is the the lambertian reflectance or the colour without shadows or highlights.

289
00:34:50,070 --> 00:34:55,040
It's an active elimination method. On controlling the lighting of the scene.

290
00:34:55,040 --> 00:35:01,700
We can recover properties of the scene with more accuracy than using the general ambient lighting of the scene itself.

291
00:35:03,930 --> 00:35:07,350
In summary, these are the main aspects of this technique.

292
00:35:10,070 --> 00:35:18,860
And now we can see how it works from scanning the different illuminations on the left.

293
00:35:19,220 --> 00:35:26,920
We reconstruct a normal map, an albedo, and through integration

294
00:35:27,480 --> 00:35:34,210
We end up with the depth map. We are talking about different illuminations.

295
00:35:34,390 --> 00:35:46,630
But how many? I would like to show you our first prototype developed in Factum, in 2017 with only four lights.

296
00:35:47,290 --> 00:35:53,710
And if you realise we achieved 254dpi and captured both colour and shape.

297
00:35:54,010 --> 00:35:57,130
But there were problems with low frequency relief.

298
00:35:58,180 --> 00:36:02,750
Then, we move to

299
00:36:04,950 --> 00:36:07,620
the results from the first prototype.

300
00:36:08,910 --> 00:36:19,470
And then we moved to the second prototype. The problems with low frequency relief was partially resolved, adding a photogrammetry workflow,

301
00:36:19,830 --> 00:36:24,780
But there were problems like the difficulty in use and the transportation of the system.

302
00:36:25,440 --> 00:36:33,510
This scanning and processing time was larger than Lucida and huge space on disck were needed to store the data.

303
00:36:37,110 --> 00:36:47,670
So at the end we decided to go with the Selene system, which is a high resolution 3D recording system using the photometric stereo principle.

304
00:36:48,480 --> 00:36:51,990
Here, I would like to show you a video of the machine working.

305
00:36:55,200 --> 00:36:58,410
This is the unit installed here at the Bodleian.

306
00:37:10,450 --> 00:37:15,280
And do we need a special sensor? No.

307
00:37:17,220 --> 00:37:23,620
It uses a regular camera... and some glue!

308
00:37:23,620 --> 00:37:27,099
Here I would like to show you a few implementations

309
00:37:27,100 --> 00:37:35,110
For example, the most simple one could be something like, holding a smartphone and having a moving LED light.

310
00:37:38,530 --> 00:37:44,560
More closer, versions like black boxes from commercial brands.

311
00:37:46,780 --> 00:37:51,100
But here we are coming to talk about our implementation.

312
00:37:51,850 --> 00:37:55,839
We tried to be cost effective, easy to use,

313
00:37:55,840 --> 00:38:10,600
and try to get the fast capure. Use no batteries, use different lenses, and try to avoid limits on size.

314
00:38:13,240 --> 00:38:21,520
And what's the Selene? Basically, it's a Canon R5 camera with a custom lighting system with four lights.

315
00:38:22,030 --> 00:38:27,920
Built in-house, in Factum. Some electronics.

316
00:38:28,860 --> 00:38:33,120
An X and Y CNC system, and the computer.

317
00:38:35,070 --> 00:38:38,850
Inside the computer, there is a software developed by us.

318
00:38:39,930 --> 00:38:58,450
We are finishing it and that's it. I think, as is obvious, Jorge is a genius and he's been an amazing and very patient teacher.

319
00:38:58,450 --> 00:39:03,429
And my thanks to Jorge and Carlos for all of that help

320
00:39:03,430 --> 00:39:09,870
Over the last ten months, it's been amazing! And the Selene is an amazing machine and the workflow designs by Jorge

321
00:39:10,000 --> 00:39:15,730
And the Factum Foundation has been used to make hundreds of incredible recordings of the Bodleian over the last ten months.

322
00:39:18,520 --> 00:39:22,179
This is another example of an early 18th century copper printing plate from the Bodleian's

323
00:39:22,180 --> 00:39:30,850
Rawlinson Collection.  This is a conventional 2D colour recording of a heavily corroded and oxidised mezzotint plate.

324
00:39:31,660 --> 00:39:38,680
The burrs, which make up the design, are incredibly short, less than 25 microns or a quarter of the width of a human hair.

325
00:39:44,170 --> 00:39:48,880
And here is the render of the three dimensional surface recorded with the Selene.

326
00:39:52,270 --> 00:39:58,629
By combining these two images, we can create a composite image which contains some of the original's colour and the

327
00:39:58,630 --> 00:40:02,740
design which has been revealed through creating a render the 3D surface.

328
00:40:06,010 --> 00:40:11,680
And this is the original 18th century print, mirrored and then perfectly aligned with the recordings of the printing plate.

329
00:40:13,460 --> 00:40:20,370
Here is the full print. The mezzotint plate recorded conventionally.

330
00:40:21,620 --> 00:40:23,870
And finally, the composite image makes for ARCHiOx.

331
00:40:27,710 --> 00:40:35,780
This engraved brass plate may simply have been made as a decorative object rather than a printing plate as the engraved lines are incredibly shallow and faint.

332
00:40:36,380 --> 00:40:39,470
Despite this, the plate has clearly been used to make prints,

333
00:40:39,830 --> 00:40:43,640
and it's perhaps because of this that the central design has become very difficult to see.

334
00:40:49,350 --> 00:40:54,420
This is the 3D recording showing the surface of the plate, which is revealed the design very effectively.

335
00:40:55,410 --> 00:40:59,800
And for small originals. These 3D recordings are relatively easy and quick to make.

336
00:41:00,310 --> 00:41:03,640
The image of the render in this example is not been enhanced in Photoshop.

337
00:41:03,970 --> 00:41:13,000
This is simply a recording of the original's surface. But it's not just printing plates that have been recorded for ARCHiOx.

338
00:41:14,070 --> 00:41:18,510
Recordings of the prints themselves also reveal details which are almost impossible to see.

339
00:41:19,170 --> 00:41:23,430
This Japanese ukiyo-e print was made using woodblocks to apply colour.

340
00:41:26,430 --> 00:41:31,560
Here you can see the impression as the paper is pushed down onto the woodblock from above.

341
00:41:38,590 --> 00:41:47,780
And if we zoom in. It's also possible to see these textural elements which are skilfully applied as part of the printing process.

342
00:41:48,380 --> 00:41:54,590
Perhaps these can only be fully appreciated through this kind of recording, as they're extremely difficult to see when studying the original.

343
00:41:54,980 --> 00:41:57,710
And would be impossible to record with conventional photography.

344
00:42:00,640 --> 00:42:06,340
This beautiful painted page belongs to a 17th century album made in the Mughal Empire.

345
00:42:13,740 --> 00:42:18,450
The recording made with the Selene not only shows the individual layers of paint which had been applied to the paper,

346
00:42:18,690 --> 00:42:22,200
but also hundreds of pricked holes which can't be seen from the colour image.

347
00:42:25,140 --> 00:42:33,840
The Sinhalese text from this palm leaf manuscript has been incised into the surface using an iron, stylus or reed pen before being inked.

348
00:42:34,500 --> 00:42:40,800
This is the colour recording. A render of the 3D recording.

349
00:42:42,610 --> 00:42:52,649
And the composite. And if we zoom in. We can see not only the incised text, but also the careful pricking around the edge of this deer,

350
00:42:52,650 --> 00:42:57,330
which suggests that this design may have been used and transferred to make copies.

351
00:43:00,180 --> 00:43:07,830
And this process of copying, using pinholes or pouncing marks can also be seen on the largest original, which has so far been captured for the project.

352
00:43:09,970 --> 00:43:16,630
Dating to the 14th century. The Gough map is one of the earliest maps to show Great Britain in a geographically recognisable form.

353
00:43:21,030 --> 00:43:26,370
This is the render of the 3D surface of the map captured at over 700,000 pixels per square inch.

354
00:43:27,880 --> 00:43:30,040
And in order to capture the map at that resolution,

355
00:43:30,040 --> 00:43:35,920
it was necessary to record the map as 85 image tiles, which were then stitched together in software.

356
00:43:38,420 --> 00:43:45,290
This is the highest resolution recording of the map to date, allowing researchers to zoom in to specific areas of interest.

357
00:43:47,710 --> 00:43:51,550
When we look at the render of the surface, we can see groupings of tiny pinholes.

358
00:43:51,880 --> 00:43:55,690
These holes are evidence that the position of geographical

359
00:43:55,690 --> 00:43:59,920
features were transferred to the Gough Map from an older map, or precursor map.

360
00:44:04,360 --> 00:44:08,740
Studying the holes allows researchers to understand what this precursor map may have looked like

361
00:44:09,100 --> 00:44:14,469
and may help to date the map more accurately. Without the recordings made firstly with the Lucida in

362
00:44:14,470 --> 00:44:17,560
2015 and this year with the higher resolution

363
00:44:17,560 --> 00:44:22,120
Selene. This research would be extremely difficult or even impossible.

364
00:44:25,190 --> 00:44:29,030
But the Bodleian is perhaps best known for its rich collection of medieval manuscripts.

365
00:44:29,390 --> 00:44:34,430
This manuscript, written in Old High German dates the first half of the 9th century.

366
00:44:35,240 --> 00:44:40,310
The catalogue record for this volume describes an annotation on the lower margin of this recto.

367
00:44:44,570 --> 00:44:49,730
A hunting scene barely visible, but clear enough to make a digital annotation.

368
00:44:56,560 --> 00:45:00,700
But the 3D recording made with the Selene reveals far more of the original annotation.

369
00:45:03,020 --> 00:45:06,610
Not just the hunting scene. But also for camouflaged

370
00:45:06,610 --> 00:45:11,350
letters R, O, D, A. A new discovery recorded for the first time.

371
00:45:13,870 --> 00:45:18,130
And if we put all of the annotations together from the colour image.

372
00:45:20,720 --> 00:45:27,500
The illustration from the 3D render. And the newly discovered text.

373
00:45:29,440 --> 00:45:34,390
We now have a complete recording of the annotation. Perhaps 1200 years after it was made.

374
00:45:37,180 --> 00:45:44,080
In this image the Bodleian's Librarian Richard Ovenden, is holding a small Rembrandt portrait, brought to the Library by Agnew's Gallery in London.

375
00:45:46,870 --> 00:45:49,570
This is the colour, or albedo recording made with the Selene.

376
00:45:52,390 --> 00:45:58,360
And this is the render of the 3D surface, which reveals the brushstrokes, the application of thick paint under the eyes.

377
00:45:58,750 --> 00:46:06,100
The substrate, in this case, a wooden panel. And also three brush hairs, two of which had never been noticed before.

378
00:46:07,410 --> 00:46:13,800
This demonstrates how 3D recordings have the potential to support research and perhaps provide provenance for an original.

379
00:46:17,240 --> 00:46:24,290
This is one of the Bodleian's treasures, a beautifully bound Bible owned by Elizabeth I, presented to her in 1584.

380
00:46:28,940 --> 00:46:31,399
Rather than making a render of the 3D surface.

381
00:46:31,400 --> 00:46:38,750
If we load the depth map and the colour images into software, it's possible to fly over the surface and relight the recording from any direction.

382
00:46:39,440 --> 00:46:43,670
The incredibly high resolution recording has captured every element from the embroidered design.

383
00:46:44,180 --> 00:46:47,930
Even the tiny seed pearls below the central Tudor rose.

384
00:46:51,720 --> 00:46:56,670
But the final sets of recordings I'd like to share are from a much older original.

385
00:47:05,370 --> 00:47:09,040
This is a two and a half thousand year old clay seal or letter-bulla.

386
00:47:09,720 --> 00:47:15,420
The impression was made with a cylindrical seal and it's tiny, just four centimetres wide.

387
00:47:20,100 --> 00:47:24,120
It's in fact, one of eight seals, which accompanied letters written by Aršāma,

388
00:47:24,270 --> 00:47:29,190
a Persian prince and regional governor to the steward of his estates in Egypt.

389
00:47:29,820 --> 00:47:37,410
Seven of the seals were made using the same cylindrical seal. These tiny unfired clay seals are incredibly fragile.

390
00:47:37,950 --> 00:47:42,810
Recording them as best this technology will allow is not only important to ensure their preservation,

391
00:47:43,560 --> 00:47:48,330
but the recordings also allow researchers to study them in a way that has never been possible before.

392
00:47:50,100 --> 00:47:57,570
Here the Selene has been used to make a render of the surface of the eighth seal, which can be virtually relit from any height or angle.

393
00:48:00,490 --> 00:48:03,190
Or in this case, using colour to represent height.

394
00:48:05,660 --> 00:48:13,580
The Lucida has been used to record the general shape of the seals and then the two data sets combined for the purposes of making 3D facsimiles.

395
00:48:15,440 --> 00:48:21,350
But even the high resolution recordings made with the Selene may not be sufficient to record such tiny originals effectively.

396
00:48:23,440 --> 00:48:29,390
By using an even higher resolution camera and following the photometric stereo workflow developed for ARCHiOx,

397
00:48:30,010 --> 00:48:34,540
it has been possible to record the seals at over six and a half million pixels per square inch.

398
00:48:36,840 --> 00:48:44,880
By using other principles such as focus stacking the depth of field has been extended, so the recording of the seal is pin sharp from top to bottom.

399
00:48:47,830 --> 00:48:53,290
And the result is this new 3D recording where you can clearly see every tiny detail, histroric

400
00:48:53,290 --> 00:48:56,290
repair, even the fingerprints of the maker.

401
00:49:05,780 --> 00:49:09,250
And this is one of the 14 letters sent from Prince Aršāma.

402
00:49:09,740 --> 00:49:17,720
The Aramaic text is remarkably well preserved. In this case, almost all of the text has been transcribed.

403
00:49:19,420 --> 00:49:24,190
The translation suggests that Aršāma valued not only horses, two of which featured on his seal,

404
00:49:26,030 --> 00:49:34,130
But also three dimensional artworks. In his letter, he commissions statues of horses and horsemen to be made by his favourite stonecutter.

405
00:49:34,310 --> 00:49:43,250
Thought to be Hinzani. So it seems fitting that we should carry out that request, albeit two and a half millennia later.

406
00:49:44,450 --> 00:49:47,720
Perhaps not by making statues of horses and horsemen.

407
00:49:49,090 --> 00:49:52,780
But by recreating his seal as a scaled up 3D facsimile.

408
00:49:53,990 --> 00:49:59,180
And even reimagining what the cylindrical seal, which has been lost to time, may have looked like.

409
00:50:04,860 --> 00:50:05,729
Through these examples,

410
00:50:05,730 --> 00:50:13,740
I hope that we have demonstrated the enormous potential which recording using ARCHiOx technology has. The ability to make discoveries to record,

411
00:50:13,740 --> 00:50:20,310
what has never been possible before, to make measurements and to recreate originals as 3D facsimiles.

412
00:50:25,630 --> 00:50:32,500
Shortly you'll hear from experts who will explain how ARCHiOx recordings have been used in their own research. With further development.

413
00:50:32,500 --> 00:50:39,160
perhaps the workflows used by ARCHiOx could be democratised, to the benefit of institutions everywhere.

414
00:50:43,190 --> 00:50:55,780
Thank you.

415
00:50:58,620 --> 00:51:03,780
So hello, everyone. I'm Andy Irving from Bodleian Digital Library Systems and Services.

416
00:51:04,530 --> 00:51:09,660
And we've been very lucky to work with John and Carlos and Jorge on this project.

417
00:51:11,010 --> 00:51:14,100


418
00:51:15,250 --> 00:51:23,559
So we run and maintain Digital Bodleian, which first launched in 2015 with version 2.0 being launched in late 2020.

419
00:51:23,560 --> 00:51:26,980
So actually almost two years ago to the day.

420
00:51:28,150 --> 00:51:34,360
And it's our primary showcase for our digitised collections here at the Bodleian Libraries as well as partner college libraries,

421
00:51:34,720 --> 00:51:39,310
the vast majority of which are captured by John and his colleagues in the Imaging Studio.

422
00:51:39,310 --> 00:51:42,940
And the majority of which are what you would consider conventional photography.

423
00:51:45,190 --> 00:51:51,760
We have a large range of objects present, digitised at the moment, and we strive to present them all in a reasonably similar way.

424
00:51:52,420 --> 00:51:56,680
Built on standards like IIIF, which we use for interoperability reasons.

425
00:51:57,670 --> 00:52:01,450
And we also have our kind of custom view built into the site,

426
00:52:01,450 --> 00:52:07,660
which is like a happy medium between a very complex way of looking at things and the more user friendly one.

427
00:52:09,610 --> 00:52:14,770
But we've always offered the option to let users open our content in other tools, which might be more appropriate.

428
00:52:15,130 --> 00:52:21,860
If you're interested in an advanced study of an object, then perhaps our viewer is not the most appropriate one. And because we use IIIF

429
00:52:21,920 --> 00:52:25,659
we're able to give people the opportunity to open in their own viewers.

430
00:52:25,660 --> 00:52:30,220
And so, you know, ones that we don't know about. But we've always offered at least two.

431
00:52:30,490 --> 00:52:34,330
And here's that same object open in the open source Mirador viewer,

432
00:52:34,750 --> 00:52:40,000
which has good support for things like annotations, where we've carefully described this object.

433
00:52:40,030 --> 00:52:46,420
Or perhaps if you want to view an object side by side with another version of it held up in a partner's library, that kind of thing.

434
00:52:47,410 --> 00:52:51,969
So this gives researchers the ability to use tools they're familiar with and use features that we wouldn't

435
00:52:51,970 --> 00:52:57,010
necessarily enable by default because they can be quite complex and offputting for casual users.

436
00:52:59,740 --> 00:53:04,930
So as we've shown, the kind of beauty of this approach is that we're just using images and it would have been

437
00:53:04,930 --> 00:53:10,150
very easy for us to kind of take the composite images that John's produced and call it a day,

438
00:53:10,660 --> 00:53:12,220
because they do look quite spectacular.

439
00:53:12,610 --> 00:53:19,270
But we felt that we needed to explore as part of this project how best we could make this data available.

440
00:53:20,440 --> 00:53:21,940
Who the audience for this might be.

441
00:53:21,940 --> 00:53:28,540
How can we convey this extra information and how can we do it in a way that's usable not just for us here at the Bodleian Libraries,

442
00:53:28,540 --> 00:53:30,400
but outside as well?

443
00:53:31,780 --> 00:53:40,590
So within the IIIF community, there is an established pattern for how to show kind of alternative views of an image or a canvas in IIIF speak.

444
00:53:40,600 --> 00:53:43,660
And it's usually illustrated with an example of multispectral imaging.

445
00:53:44,170 --> 00:53:51,100
And this technique is in some ways very similar to multispectral imaging in that we have different views of the same objects in different cases,

446
00:53:51,850 --> 00:53:56,260
and we thought this would be a good starting point for us as it's really the same class of problem.

447
00:53:57,670 --> 00:54:00,309
And we implemented this very quickly ourselves.

448
00:54:00,310 --> 00:54:07,880
It's supported in the Mirador viewer, which you've just seen, and we could very quickly now label our images and provide different views.

449
00:54:08,800 --> 00:54:14,530
But again, we felt that this might not be sufficient for fully exploring. Here

450
00:54:14,530 --> 00:54:19,870
For example, you know, the albedo image looks, we think, slightly better than the composite image as a kind of default view.

451
00:54:19,870 --> 00:54:23,130
So we can just kind of make a blanket kind of case.

452
00:54:25,030 --> 00:54:28,630
And as I hope this video demonstrates,

453
00:54:30,190 --> 00:54:36,399
the native Mirador three kind of layers view showing these different images as different choices lets

454
00:54:36,400 --> 00:54:41,470
you switch between them so you can still see the different views of the images that we've captured.

455
00:54:43,000 --> 00:54:49,060
You can blend them together in certain ways, which may and may not be captured very well in the in the video.

456
00:54:50,200 --> 00:54:59,290
And it looks okay. For kind of materials like this with a very shiny surface or where we've got relatively high relief

457
00:54:59,290 --> 00:55:05,740
The effect is not terrible, but it's also not as great as using the software that John has just demonstrated.

458
00:55:06,160 --> 00:55:11,860
And so while better than nothing, it's not necessarily the the end point of our of our journey.

459
00:55:12,900 --> 00:55:16,680
And of course you can use other plugins to adjust things like contrast and so on.

460
00:55:20,880 --> 00:55:27,960
And where we have like a shallower depth or relief or a different material type the kind of effect is much less impressive.

461
00:55:28,740 --> 00:55:34,500
So we thought about how we might extend IIIF in a very small way with an extension.

462
00:55:35,220 --> 00:55:42,000
And one of the benefits of using these kind of choice patterns is that most viewers

463
00:55:42,000 --> 00:55:45,030
will often just display the first image if they don't understand the details,

464
00:55:45,030 --> 00:55:48,330
which is great for us, kind of as a fallback.

465
00:55:49,290 --> 00:55:52,740
But obviously, these aren't just images, they contain data.

466
00:55:53,280 --> 00:55:59,540
And so we thought we could perhaps add some extra metadata to describe what type of lighting map we're actually using here,

467
00:55:59,550 --> 00:56:03,000
whether it's the albedo or depth map and so on and so forth.

468
00:56:03,840 --> 00:56:09,950
So we created a tiny extension, which we now include our own vocabulary in these contexts.

469
00:56:09,960 --> 00:56:14,640
And again, another important thing here is that if clients don't understand it, they simply ignore it.

470
00:56:14,820 --> 00:56:24,870
So by not breaking anything, but by enhancing, we're able to hopefully demonstrate that for an individual image section here,

471
00:56:25,230 --> 00:56:28,650
we can say that this particular one is an albedo and so on and so forth.

472
00:56:28,950 --> 00:56:34,530
Now I will hand over to Richard now to try to hopefully explain how we took this further.

473
00:56:34,730 --> 00:56:41,550
Morning, everyone.

474
00:56:42,120 --> 00:56:46,920
My name's Richard Allen, and I work with Andy BDLSS working on Digital Bodleian but

475
00:56:47,070 --> 00:56:50,910
I've been working on this for the last few months, this project.

476
00:56:51,180 --> 00:57:01,049
And I'm going to briefly go over the story of how we took some of the data that we produced from the Imaging

477
00:57:01,050 --> 00:57:15,150
Studio by John and how we think we can best visualise that and deliver it through our existing systems.

478
00:57:19,930 --> 00:57:28,270
So we've established earlier that ARCHiOx produces a number of derivatives from the Selene and Lucida scanners

479
00:57:28,270 --> 00:57:33,099
But what can we do with them in addition to servicing them as layers via the

480
00:57:33,100 --> 00:57:40,569
Mirador viewer, as Andy explained. One option would be to take the depth map which can

481
00:57:40,570 --> 00:57:45,309
be used in combination with a program called a shader to deform a mesh and produce

482
00:57:45,310 --> 00:57:50,050
a terrain model of the object to create a truly 3D render of the object.

483
00:57:50,440 --> 00:57:54,700
However, to get really good results

484
00:57:54,700 --> 00:58:00,489
We would want to have a vertex point for each pixel potentially,

485
00:58:00,490 --> 00:58:08,560
and because we have gigapixel images we'd soon run into the limits of what was capable of rendering in real time in a web browser.

486
00:58:09,070 --> 00:58:20,140
And so we also have to think about the way our viewers are used with our regular material, and it's usually a top down view.

487
00:58:20,770 --> 00:58:30,549
And so this would be a waste and we'd be investing a lot of computational power into rendering something in 3D which wouldn't look 3D

488
00:58:30,550 --> 00:58:32,950
if we were looking from a top down perspective,

489
00:58:33,790 --> 00:58:42,580
and so we decided not to go with the the depth map approach and instead we have some other derivatives that we can use.

490
00:58:42,940 --> 00:58:51,850
We have a normal map which, as we have learnt, is a way of encoding the way a 3D object surface is lit in two dimensions using RGB values.

491
00:58:52,720 --> 00:58:56,860
Here on the right here you can see an animation which uses a normal map.

492
00:58:56,860 --> 00:59:03,790
On the left of this image you'll see true 3D. On the right you'll see a flat surface with a normal map applied,

493
00:59:04,360 --> 00:59:09,520
and this is sometimes referred to as 2.5D, because when you look at the render from the top down view,

494
00:59:09,550 --> 00:59:12,790
the right hand side looks similar to the full 3D view on the left,

495
00:59:12,970 --> 00:59:18,370
and this technique is very cheap to compute in comparison to using full 3D.

496
00:59:18,760 --> 00:59:23,110
And so this suits our top down view a lot better.

497
00:59:25,940 --> 00:59:30,360
And so we decided to go with this technique and create a way for the user to interact with the normal map.

498
00:59:32,570 --> 00:59:36,710
Now to start with, we had a standalone example using

499
00:59:38,570 --> 00:59:42,110
A normal rendering tool, which is written by Jorge.

500
00:59:42,620 --> 00:59:46,570
And this took a normal map and allowed you to move a torch around relighting the object.

501
00:59:46,580 --> 00:59:52,100
But this was limited to running it through a standalone application called Processing, and you had to

502
00:59:52,400 --> 00:59:57,920
download and install this on a local desktop and we needed something that worked in a web browser.

503
00:59:58,070 --> 01:00:04,240
So. How can you render a normal map in a web browser?

504
01:00:04,270 --> 01:00:08,650
Well, we initially looked at what other institutions were using for their 3D renederers and viewers.

505
01:00:09,100 --> 01:00:16,030
And here you'll see the Smithsonian's Voyager 3D viewing application rendering Neil Armstrong's spacesuits in a web browser.

506
01:00:17,170 --> 01:00:24,190
The Voyager application itself makes use of a JavaScript library calls three.js. And what is three.js?

507
01:00:24,970 --> 01:00:31,120
three.js is a JavaScript library designed to make using WebGL easier,

508
01:00:31,270 --> 01:00:40,000
and WebGL is an API that allows people to render high performance 2D and 3D graphics in their web browsers. And in order to make good use of it

509
01:00:40,030 --> 01:00:45,070
Developers need to know how to write their own shaders, and this is a complete discipline within itself.

510
01:00:45,550 --> 01:00:52,330
So three.js abstracts the process of working with WebGL and it provides

511
01:00:52,900 --> 01:00:59,740
ready to use shaders and materials so that developers only need to know JavaScript in order to begin producing their own 3D scenes within minutes.

512
01:01:00,160 --> 01:01:04,989
So we decided to go with this. And using the power of three.js

513
01:01:04,990 --> 01:01:11,890
We had something written and working in a similar way to Jorge's application in a web browser within about an hour.

514
01:01:13,260 --> 01:01:19,270
However, the example I'm just shown you is a pure JavaScript application.

515
01:01:19,270 --> 01:01:25,150
And in order to get something working with our Mirador viewer we'd need to be able to use something called React.

516
01:01:25,330 --> 01:01:27,310
React is another JavaScript library.

517
01:01:28,270 --> 01:01:34,270
And I would argue it's also a particular method of writing applications that is quite different to regular JavaScript.

518
01:01:37,420 --> 01:01:41,680
But we managed to get something working. But it took longer than an hour this time.

519
01:01:43,150 --> 01:01:45,790
After a few afternoons, actually, of going through tutorials,

520
01:01:45,790 --> 01:01:54,879
we were able to recreate what we had created using three.js and using the relighting tool in an application that worked in a web browser again,

521
01:01:54,880 --> 01:01:58,090
but using the framework that Mirador is based upon.

522
01:02:00,010 --> 01:02:06,610
Now, the question is, how do we get this working in Mirador, which is the next logical step?

523
01:02:07,030 --> 01:02:12,610
And there were multiple challenges here, such as how do you get non-native code working in Mirador?

524
01:02:12,850 --> 01:02:20,380
How would we display the information in a way that doesn't break anything, and how would we relate to the other images we were surfacing?

525
01:02:22,230 --> 01:02:30,630
Well, helpfully the Mirador community already has a diversity of open source plugins available and each did little bits of what we wanted to do.

526
01:02:31,170 --> 01:02:37,370
And there is also plenty of documentation to get developers started with their own plugin projects.

527
01:02:37,380 --> 01:02:40,800
So we decided that we would take this route with the relighting example,

528
01:02:41,700 --> 01:02:47,430
and we spent a little while researching how Mirador worked and how to plugins interacted with the data that Mirador was handling.

529
01:02:49,600 --> 01:02:56,620
In order to get our plugin to work, we would need to take our normal map and the albedo map, past them into three.js, into a three.js

530
01:02:56,620 --> 01:03:04,990
material, and render our image in real time with a light like in the example we got working as a standalone application in React.

531
01:03:06,100 --> 01:03:07,719
And as Andy covered, later

532
01:03:07,720 --> 01:03:14,170
we needed to make some changes to our IIIF manifest to make these images and to get these images into Mirador in the first place.

533
01:03:14,530 --> 01:03:21,520
And so here is an early example of a static manifest we used for development and which we hosted via GitHub gists.

534
01:03:23,380 --> 01:03:26,590
But the question now, though, was how would we display our 3D render?

535
01:03:27,160 --> 01:03:34,390
We needed to make something and that followed the zooming and panning in Mirador whilst it rendered in order to create a seamless experience.

536
01:03:35,470 --> 01:03:40,870
But thankfully, we discovered through our research that a bulk of the work done by Mirador in terms of

537
01:03:40,870 --> 01:03:45,820
handling gigaapixel tiled images is actually done by a sub-application called OpenSeadragon,

538
01:03:46,360 --> 01:03:51,610
which is well documented and has an extremely supportive developer community for answering questions.

539
01:03:51,920 --> 01:03:56,290
And so we were able to learn how to take control of the images in our native plugin.

540
01:03:56,500 --> 01:04:04,270
And here is the documentation on using HTML overlays, which can allow custom overlays over the images in the viewer.

541
01:04:04,630 --> 01:04:10,470
And so we use this to position our three.js renderer directly over the images being presented in Mirador.

542
01:04:10,490 --> 01:04:14,500
And the result was this. Ta da!

543
01:04:16,210 --> 01:04:20,470
Here you'll see the application working as a plugin and Marinol making use of the OpenSeadragon

544
01:04:20,470 --> 01:04:26,380
API to recreate our relighting tool in a web browser using a three.js overlay.

545
01:04:26,860 --> 01:04:32,439
It looks good. However, as is par for the course with these things, there is always room for improvement.

546
01:04:32,440 --> 01:04:37,000
And one of the major limitations to our new technique was that in order to keep things simple,

547
01:04:37,480 --> 01:04:40,810
we were loading in the textures for the overlay as a single, large image.

548
01:04:40,960 --> 01:04:46,150
The problem here is that three.js will automatically downsized textures to a maximum

549
01:04:46,150 --> 01:04:50,110
of 4K resolution to help maintain performances on devices like mobile phones.

550
01:04:51,460 --> 01:04:57,760
What this meant was that our 2.5D renders would only ever be a maximum of 4K using this technique.

551
01:04:58,270 --> 01:05:02,470
Not a problem for this image being roughly that size. But what about larger images like the Gough Map?

552
01:05:04,420 --> 01:05:11,049
Well, thanks to the collaboration we've had with Factum, we were able to meet up in Madrid and we were

553
01:05:11,050 --> 01:05:17,290
able to tackle some of these challenges together by just having it in front of us and discussing it.

554
01:05:17,740 --> 01:05:30,100
And you can see here Jorge and myself, this is the moment when we managed to activate a debugging process in the Mirador viewer,

555
01:05:30,580 --> 01:05:36,040
which allowed us to visualise some of these complex issues because it's really difficult

556
01:05:36,040 --> 01:05:43,989
thinking about 3D in your head, alone! And within a couple of minutes of the pair-programming,

557
01:05:43,990 --> 01:05:47,320
we were able to get this working.

558
01:05:48,850 --> 01:05:52,750
We were able to display this information

559
01:05:55,330 --> 01:06:02,920
in front of us. And then a few weeks after we got back from the Madrid trip, we had solved the issue.

560
01:06:03,350 --> 01:06:15,370
And here is a view of the application with a tool menu added that enables you to change the direction of the light used in the overlayed scene.

561
01:06:15,370 --> 01:06:22,779
And you can see towards the end of the video that our plugin just happens to work with - on the right hand side

562
01:06:22,780 --> 01:06:28,149
You'll see a menu, there, from a application called Mirador Imaging Tools,

563
01:06:28,150 --> 01:06:34,360
and as Bob Ross would say, it was a happy little accident that it worked with it.

564
01:06:34,390 --> 01:06:48,500
You'll see in it in a few seconds. It changes the colours of our 3D render so we can use this other tool directly with our

565
01:06:48,770 --> 01:06:54,950
application which we've developed here and you can see the light is changing the direction of the,

566
01:06:56,620 --> 01:07:00,660
Well, the light is changing direction and you can see it interacting with the surface, using the normal map.

567
01:07:09,650 --> 01:07:16,710
But, where to now? The primary thing that we need to work on would be to get our images loading seamlessly as Mirador loads them.

568
01:07:17,160 --> 01:07:22,740
Currently, our tile servers do a lot of the heavy lifting to reload the images into a three.js rendering layer

569
01:07:22,920 --> 01:07:26,760
but this would not be ideal if we were to release it publicly.

570
01:07:27,210 --> 01:07:31,350
But the good news is we already have something close to solving this problem and again.

571
01:07:31,350 --> 01:07:35,820
It's thanks to the OpenSeadragon community who have been very helpful.

572
01:07:37,560 --> 01:07:47,580
What else? What else can we do? Well, the examples we have developed in our plugin currently use only one of the available materials,

573
01:07:47,910 --> 01:07:51,390
which is really good at rendering metal objects like the copper plates.

574
01:07:51,750 --> 01:07:58,110
But it's an inappropriate choice for other materials, like paper or vellum or wood and things which aren't shiny normally.

575
01:07:58,950 --> 01:08:02,270
And so we would like to investigate using the different materials available in three.js.

576
01:08:02,280 --> 01:08:08,730
But to do this, we'd also need to generate additional map types with the Selene.

577
01:08:11,060 --> 01:08:17,780
We'd also perhaps like to write their own shader code, and Jorge has kindly offered to to help there.

578
01:08:18,410 --> 01:08:26,180
And we'd like to try our own renderer, our own custom renderer and see what we can do and maybe use an alternative renderer to three.js,

579
01:08:26,180 --> 01:08:32,330
which is PixiJS, which might be a little bit more lightweight and be run things a lot smoother in the web browser.

580
01:08:32,810 --> 01:08:37,850
We'll see. And these are just some of the avenues for further research that we'd like to do.

581
01:08:38,540 --> 01:08:45,650
But I think it's important to say that after all, this one thing is clear we have a great foundation which we can build upon.

582
01:08:46,010 --> 01:09:00,350
Thank you.

583
01:09:00,360 --> 01:09:06,270
This is Damien Bove, who is a Gough Map expert and been working with me.

584
01:09:06,450 --> 01:09:09,930
I'm not an expert! Absolutely an expert! Working with me.

585
01:09:11,250 --> 01:09:15,239
I think you came during the actual recording of the Map. Yeah, that's right.

586
01:09:15,240 --> 01:09:18,690
And then lots of sort of follow up sessions examining pinholes and things like that.

587
01:09:19,740 --> 01:09:24,210
Yeah, Well, we've got 10 minutes, so no history. There won't be any history.

588
01:09:24,270 --> 01:09:27,970
This is the Gough Map and it's full of holes!

589
01:09:28,020 --> 01:09:39,630
That's right. That's the that's all you need to know. We didn't need a 3D scan to see the holes because you can see them with the naked eye.

590
01:09:39,660 --> 01:09:44,459
You can see them on conservators photographs or really high-res images.

591
01:09:44,460 --> 01:09:49,260
And even the hyperspectral images, you can see some of the holes.

592
01:09:49,830 --> 01:10:02,830
You can see some of them at the point. It's only recently that we've come to understand what the holes are because until, well,

593
01:10:02,890 --> 01:10:12,910
fairly recently they were thought to have been made by the engraver, James Basire, in the preparation of Richard Gough's British Topography.

594
01:10:16,930 --> 01:10:19,690
William Basevi Sanders was the first one and came up with this.

595
01:10:19,840 --> 01:10:26,710
He said, because Gough had made no allusion to it in his account, he thought that Basire must have done it, which is a bit unfair, I think.

596
01:10:27,730 --> 01:10:36,100
And Parsons, nearly a hundred years later, he was a librarian here, I think, he said the same thing.

597
01:10:37,300 --> 01:10:40,980
Well, they're entitled to their opinion, but we think Basire is innocent,

598
01:10:40,990 --> 01:10:43,590
and it's the other way round.

599
01:10:43,590 --> 01:10:50,290
That the holes were made, not while coppying from the map, but they were made to create the Gough Map.

600
01:10:53,480 --> 01:10:59,310
And coppied from a precursor that we call 'P'.

601
01:10:59,390 --> 01:11:11,490
So when I say 'P', that's the precursor to map. While some holes mark out the mountains and the river heads,

602
01:11:11,490 --> 01:11:19,350
most of them, the vast majority are marking out both the shape and the location of the signs.

603
01:11:19,890 --> 01:11:29,250
The place signs and these four here are just the standard common ones. There are various shapes with spires and towers and walls.

604
01:11:29,550 --> 01:11:37,490
This is the standard one. You see the little cottage?

605
01:11:37,490 --> 01:11:44,320
They've got three generally. Actually, when they started off in the north of Scotland, they did four for the cottage, one in each corner.

606
01:11:44,560 --> 01:11:49,510
And then as they moved south, they became more economical with their designs.

607
01:11:50,140 --> 01:11:57,230
And moved on to three. You find some of them have got dozens of holes and some have only got a few.

608
01:11:57,230 --> 01:12:01,870
So. The signs are small.

609
01:12:02,140 --> 01:12:06,730
Some of them are less than centimetres, maybe seven or eight millimetres across.

610
01:12:07,120 --> 01:12:13,750
So you'd expect there to be some discrepancy between the inking and the pinholes.

611
01:12:14,620 --> 01:12:19,240
But in many cases, dozens of cases, the discrepancies are systematic.

612
01:12:20,370 --> 01:12:27,480
The top ones you can see are where the the river has been painted across the pinholes.

613
01:12:27,870 --> 01:12:32,490
So they've moved the sign away from the river. They've drawn the sign away from the river.

614
01:12:33,570 --> 01:12:42,210
The lower ones. That's where they seem to have picked it through and then decided that the pages have slipped.

615
01:12:42,360 --> 01:12:46,829
And so they've pricked through twice. That's why you've got. Two identical shapes,

616
01:12:46,830 --> 01:12:56,240
Forms, but slightly offset. Edinburgh.

617
01:12:59,740 --> 01:13:06,260
There are dozens of these kind of things, here the sign's been drawn quite a way south of the -

618
01:13:06,310 --> 01:13:09,970
that's the Firth of Forth on your left and it's been pricked

619
01:13:11,280 --> 01:13:14,100
There, and then somebody decided to draw it.

620
01:13:15,980 --> 01:13:22,670
there instead almost exactly in the same shape, so they must have had the other one to work from because it's so similar.

621
01:13:27,500 --> 01:13:39,520
Two or three of you have mentioned pouncing. I don't think this is pouncing in that pouncing is

622
01:13:40,650 --> 01:13:45,090
Where powders fall through the hole to to create a design.

623
01:13:45,330 --> 01:13:50,280
This is more direct. This is just poking a holes through, basically. Poking holes through the top layer.

624
01:13:50,940 --> 01:14:00,930
Through the precursor map into the new map. I don't know what else to call it.  Pricking, that's it.

625
01:14:03,990 --> 01:14:09,430
We've not seen it on any other maps. Some other maps may turn up, but we haven't seen any.

626
01:14:09,450 --> 01:14:12,509
We've been looking around. We've seen pouncing, but pricking patterns - 

627
01:14:12,510 --> 01:14:19,040
we haven't seen this. Well, we think they were improvising just a great extent.

628
01:14:19,050 --> 01:14:24,000
You can see where they've been changing their technique as they go, and I think they're making it up as they go along really.

629
01:14:24,050 --> 01:14:28,230
They weren't mapmakers. Probably. They were probably used to making books.

630
01:14:28,230 --> 01:14:31,880
And this may have been the first map they ever copied. I don't know.

631
01:14:35,670 --> 01:14:44,880
The first Factum scan - we're the three lucky recipients of two - revealed something we hadn't seen with the naked eye and that was the scoring.

632
01:14:45,690 --> 01:14:49,170
There are these scored marks around the cartouches, the regional cartouches.

633
01:14:49,710 --> 01:14:57,390
And what's interesting about scoring is it matches, almost exactly matches the pinholes rather than the inking.

634
01:14:57,780 --> 01:15:03,420
So we think it is another. 

635
01:15:03,580 --> 01:15:09,780
Well, it's another method of of copying that they they abandoned halfway I

636
01:15:09,780 --> 01:15:16,710
think because it's a few cartouches have it and the rest don't. It's another abandoned method.

637
01:15:18,270 --> 01:15:21,790
This is a map of all the pinholes I have found to date.

638
01:15:23,160 --> 01:15:29,360
And from this you can see. We have an idea of what the precursor looked like.

639
01:15:30,140 --> 01:15:35,600
You can see that vertical line down the middle there, the pricked wall, Hadrian's Wall.

640
01:15:38,090 --> 01:15:43,970
2123,

641
01:15:48,160 --> 01:15:58,120
Holes so far. And this is settlements by pricking, so the red ones are the pricked settlements and the blue ones are the unpricked ones.

642
01:16:04,000 --> 01:16:11,260
And then comes the second scan.

643
01:16:12,790 --> 01:16:14,680
Which means I'm going to go back and do these again,

644
01:16:15,010 --> 01:16:22,990
because we're finding all kinds of new things. You see where the blue is, the blue, unpricked ones.

645
01:16:23,030 --> 01:16:31,629
This is we presume, they were revising those areas, especially the coastline around East Anglia

646
01:16:31,630 --> 01:16:39,310
And Kent there, they were revising those areas so they left them unpricked.  Along the wall as well there's a lot, along Hadrian's Wall.

647
01:16:41,670 --> 01:16:47,340
So we come to the second scan and this is great because then we can draw lines across the

648
01:16:49,260 --> 01:16:52,970
Pinholes and take a profile.

649
01:16:56,370 --> 01:17:04,500
Which solves all our problems. But it doesn't really, I'm afraid to say. Now,

650
01:17:06,020 --> 01:17:11,420
I asked John to look at Bolton Castle because it's a very difficult one to look at.

651
01:17:12,720 --> 01:17:20,340
And his reply to me began, 'I can see now why research into this map is a) exciting,

652
01:17:20,730 --> 01:17:24,810
b) infuriating and c) probably a lifetime's work'

653
01:17:26,220 --> 01:17:32,910
John took vertical cross-sections and horizontal cross-sections and diagonal cross-sections and he measured the depth of each hole.

654
01:17:33,960 --> 01:17:37,050
And he looked at oblique images and moved the light sources around,

655
01:17:37,950 --> 01:17:46,170
which I quote 'hasn't led me to be any more or less confident about my conclusions'.

656
01:17:48,090 --> 01:17:52,860
He did find three triangular indentations roughly equal in size and orientation.

657
01:17:53,760 --> 01:17:56,820
We don't know what to make of them, but a different shape

658
01:17:58,150 --> 01:18:02,560
May represent a different tool, a different person or different episode.

659
01:18:03,830 --> 01:18:07,819
I suggested that rather than looking at a single hole, we should look at clusters.

660
01:18:07,820 --> 01:18:16,540
And if the clusters come in the recognisable form of of a sign, then we cold maybe lower the threshold for proof.

661
01:18:16,840 --> 01:18:21,879
I think that might be a bit dangerous because you can basically see anything. That's what you learn with the Gough Map

662
01:18:21,880 --> 01:18:26,920
If you look hard enough you see anything and once you see something in your mind, you can't unsee it then.

663
01:18:30,880 --> 01:18:37,960
John finished his letter, his email by saying, 'Please let me know your thoughts' before telling me I was going to France for a fortnight!

664
01:18:38,410 --> 01:18:42,250
So I think that maybe he's fed up after the few weeks of this!

665
01:18:42,980 --> 01:18:46,050
Can you explain what Bolton Castle is important?

666
01:18:46,060 --> 01:18:54,460
Bolton Castle, well, I don't quite know yet, but we think the dating of Bolton Castle - we have an idea of the dating of the precursor map.

667
01:18:55,690 --> 01:19:04,300
And if it was pricked through from the precursor map, it might make that later than the imagined date.

668
01:19:04,320 --> 01:19:05,830
So it's a bit complicated,

669
01:19:06,070 --> 01:19:15,880
but it seems we sorted it yet. It might be the machine learning solves this problem but we can't. Well maybe John can.

670
01:19:21,810 --> 01:19:24,860
Yes. We often talk about pinholes.

671
01:19:26,060 --> 01:19:29,240
But they are more 'impressions'

672
01:19:29,720 --> 01:19:35,660
Some of them are poked through, but some of them are just impressions. The tool is being pressed into the parchment rather than poked through.

673
01:19:37,050 --> 01:19:40,860
And this is what the parchment looks like close up.

674
01:19:40,860 --> 01:19:45,810
So if you imagine with dirt and ink and damp and 600 years of abuse,

675
01:19:47,220 --> 01:19:53,520
it makes a bit of a challenge to identify what's what's an intentional hole and what's just a mark.

676
01:19:57,450 --> 01:20:06,850
We often call these pinholes, by the way. But the tool was almost certainly not a pin was more likely an awl or a knife point or something.

677
01:20:08,140 --> 01:20:11,740
We call it a pinhole because it's more euphonious than awl impression.

678
01:20:12,800 --> 01:20:17,540
And also my wife calls me 'Professor Pinhole', and it's an honorary title.

679
01:20:17,540 --> 01:20:23,020
I don't want to lose it! This is Winchester.

680
01:20:23,030 --> 01:20:29,809
This is a problem because these look like holes, but they don't show up very well on the 3D.

681
01:20:29,810 --> 01:20:39,200
I should give them to you, John, to look at. There's an inked cross below, but there's an uninked cross - 

682
01:20:39,860 --> 01:20:47,260
a pricked cross above. And two smaller crosses.

683
01:20:47,500 --> 01:20:52,980
To the left and right. Which are just small indentations.

684
01:20:52,990 --> 01:20:56,080
They show up in photographs, but the profiles are inconclusive.

685
01:20:59,300 --> 01:21:04,750
Same at Chichester. There there's a large pricked cross.

686
01:21:06,150 --> 01:21:10,890
Which hasn't been inked-in. And interestingly, there seems to be a kind of fleur-de-lis

687
01:21:11,940 --> 01:21:15,809
On the hyperspectral, the Multi-spectral image, which we hadn't seen before.

688
01:21:15,810 --> 01:21:21,960
So maybe that's what was being poked through or pressed through from the precursor map.

689
01:21:24,450 --> 01:21:31,260
So next thing is trying to recreate a Gough Map hole with some old parchment and various tools.

690
01:21:34,810 --> 01:21:42,640
And I've been doing this. I've got an envelope full of holes which I'm giving to John so you can test them.

691
01:21:42,880 --> 01:21:51,490
I've got them under microscope. I'm going to try and get them under 3D scanner for comparison and see if we can recreate them.

692
01:21:52,810 --> 01:22:06,840
And that's it.

693
01:22:06,840 --> 01:22:11,640
Good morning, everybody. Just while we're getting the slides sorted out, I can get started.

694
01:22:11,850 --> 01:22:19,470
My name is Jo Story and this is Jess Hodgkinson, who is a PhD student at the University of Leicester working with me.

695
01:22:19,860 --> 01:22:26,459
It's a very great pleasure to be here this morning, not least because coming to the Bodleian Library is always a thrill,

696
01:22:26,460 --> 01:22:31,980
because it contains some of the most important manuscripts that I am interested in at the moment.

697
01:22:32,970 --> 01:22:40,890


698
01:22:40,890 --> 01:22:46,740
My particular interest at the moment are on what we call insular

699
01:22:46,740 --> 01:22:52,170
manuscripts, so manuscripts that were written on the islands of Britain and Ireland in the period

700
01:22:52,470 --> 01:23:00,900
between round about 600 and round about 900 A.D. and on manuscripts that were written using

701
01:23:00,900 --> 01:23:07,590
the similar types of scripts so, insular scripts on the continents in Carolingian and Francia

702
01:23:08,280 --> 01:23:15,390
in the same kind of period, English and Irish missionaries left the islands and founded monasteries on the continent,

703
01:23:15,690 --> 01:23:20,070
both in Central Fancia and in Eastern Francia.

704
01:23:20,640 --> 01:23:28,650
And those monasteries continued to produce books in insular style for generations after the original missionaries had died.

705
01:23:29,040 --> 01:23:39,930
So my interests in particular are in that phenomenon of the existence and continuation of insular book production,

706
01:23:40,200 --> 01:23:44,790
both within Anglo-Saxon England and on the continent in this period.

707
01:23:45,120 --> 01:23:50,940
And indeed, Jess when she was doing her Masters degree was one of my sort of assistants on a on a sort of proof of concept of this project.

708
01:23:51,270 --> 01:24:00,299
She's since moved on to a to do a Ph.D. and is including insular manuscripts in her research.

709
01:24:00,300 --> 01:24:06,690
And she's going to tell us about the amazing things that we found with John's help in one of them.

710
01:24:07,080 --> 01:24:10,700
But I just want to say just a couple of things about the importance of this new technique.

711
01:24:10,950 --> 01:24:19,649
for my research and for research in medieval manuscripts in particular, this type of research looking at this micro topography,

712
01:24:19,650 --> 01:24:30,870
these micro landscapes, is as Adam said at the beginning, going to really transform the way in which we understand two aspects of these books.

713
01:24:31,170 --> 01:24:36,030
One is the readership and engagement with these books after they were created.

714
01:24:36,210 --> 01:24:41,040
As Jess is going to to show you we can see things that we have never seen before through this technique.

715
01:24:41,460 --> 01:24:48,810
The other is in understanding in much greater detail and with completely different techniques and in a totally different way.

716
01:24:49,110 --> 01:24:56,580
Some of the methods of the making of these books, and I'm thinking here in particular about the way the vellum was prepared,

717
01:24:57,210 --> 01:25:08,520
both from the point at which it came off the carcase of the animal to the production of the of the prepared vellum that we can see in front of us.

718
01:25:08,850 --> 01:25:16,320
I'm working with another team at the moment on insular vellum, and we're very interested there in things like hair follicles,

719
01:25:16,560 --> 01:25:21,299
things like the marks that were made with the lunellum when

720
01:25:21,300 --> 01:25:25,230
the parchment maker was scraping the hair and the fat off of the skin.

721
01:25:25,500 --> 01:25:29,969
Those are the types of things that we can start to pick up with this method that

722
01:25:29,970 --> 01:25:34,140
we have never really been able to see in any kind of systematic way before.

723
01:25:34,770 --> 01:25:40,140
Those lunellum marks could be kind of like a fingerprint of a manuscript maker, of a parchment maker.

724
01:25:40,380 --> 01:25:46,140
And this type of technology is going to give us a tool to see that in a way that we hadn't before.

725
01:25:46,440 --> 01:25:52,259
I also really, really want one of those Mirador viewer things for ourselves.

726
01:25:52,260 --> 01:25:55,620
Can we have one? Can that be your next trial please? Because this is great,

727
01:25:56,370 --> 01:26:02,580
because being able to take these images but then open them up to that wider community so that we can surface them on the web

728
01:26:02,760 --> 01:26:09,690
and then to start to play them with the community is just going to be transformational with what we can do with this material.

729
01:26:10,260 --> 01:26:13,800

730
01:26:14,310 --> 01:26:17,510


731
01:26:18,660 --> 01:26:24,390


732
01:26:26,940 --> 01:26:32,009
Okay. So I'm going to begin by providing a quick introduction to this manuscript,

733
01:26:32,010 --> 01:26:39,270
which is Bodleian Library Manuscript Selden Supra 30. I'll then discuss some of the newly discovered inscriptions,

734
01:26:39,570 --> 01:26:43,020
how they were found and revealed, and what the next steps might be.

735
01:26:43,860 --> 01:26:47,250
Manuscript Selden Supra 30 is a biblical manuscript.

736
01:26:47,310 --> 01:26:52,350
It's a copy of the New Testament book, The Acts of the Apostles, and it's written in Latin.

737
01:26:52,830 --> 01:26:56,700
It's quite a small volume, only slightly bigger than an A5 piece of paper.

738
01:26:57,720 --> 01:27:01,130
It was copied by two scribes in a type of scrap called uncial

739
01:27:01,520 --> 01:27:08,910
And the opening displayed here shows most of page 18 and a little bit of page 19, and that's within the first scribe's stint.

740
01:27:09,530 --> 01:27:13,640
Certain features of the manuscript, including the style of uncial script used

741
01:27:14,090 --> 01:27:18,080
indicate that it was written in the geographical area we now know as England,

742
01:27:18,560 --> 01:27:22,639
most likely somewhere in the Kingdom of Kent in the first half of the eighth century.

743
01:27:22,640 --> 01:27:25,650
So that's sometime between around 700 and 750AD.

744
01:27:27,320 --> 01:27:35,000
Very early in its history manuscript Selden Supra 30 appears to have been used by a woman or a group of women. Two Latin prayers were 

745
01:27:35,030 --> 01:27:40,790
added to page 70, which was originally left blank, by different hand in a style of uncial script,

746
01:27:40,790 --> 01:27:43,430
which appears to be roughly contemporary with the main text.

747
01:27:44,090 --> 01:27:52,370
The first prayer is a petition to God made by an anonymous woman described as God's unworthy servant, which in Latin is 'indignam famulam'.

748
01:27:53,390 --> 01:27:56,450
This strongly suggests that at the time the prayer was added,

749
01:27:56,720 --> 01:28:03,620
the manuscript was being used by at least one woman and the prayers may have even been copied into the manuscript by a female scribe.

750
01:28:04,650 --> 01:28:07,230
Prior to the new discoveries, which I'm going to talk about.

751
01:28:07,440 --> 01:28:18,120
One marginal inscription was known about and this was the letters E, A, D, B and E, surrounded by two crosses incised into the lower margin of page 47.

752
01:28:18,840 --> 01:28:26,250
This was first reported by Elias Avery Lowe in the second volume of  Codices Latini Antiquiores, in 1935.

753
01:28:27,120 --> 01:28:30,360
Given the manuscript's apparent links to women established by the prayer,

754
01:28:30,870 --> 01:28:37,650
Lowe suggested that these letters were abbreviations of the Old English female name Eadburg.

755
01:28:39,440 --> 01:28:47,240
Earlier this year, I discovered another inscription etched into the margins of manuscript in Selden Supra 30, which had never been noticed before.

756
01:28:48,230 --> 01:28:54,050
At the bottom of page 18 I found Eadburg's name, written in full, preceded by a cross.

757
01:28:54,620 --> 01:29:01,730
The inscription is very faint and almost invisible to the naked eye. In an image under normal light,

758
01:29:03,210 --> 01:29:06,200
you can't really see the inscription at all on it, as you can see here.

759
01:29:06,830 --> 01:29:12,830
So ARCHiOx took a 3D photometric stereo recording of the inscription using the Selene scanner.

760
01:29:15,020 --> 01:29:16,400
And this is what it revealed.

761
01:29:17,670 --> 01:29:25,249
And this was then subsequently enhanced by Jorge Cano using virtual relighting, image stacking and principal components analysis.

762
01:29:25,250 --> 01:29:34,910
And that allowed us to see it a bit more clearly. And so now we've digitally annotated in blue the letters we think we can currently see.

763
01:29:35,420 --> 01:29:40,100
However, deciphering the inscription is still very much a work in progress.

764
01:29:40,280 --> 01:29:47,060
It's increasingly difficult to read towards the end, but you can very clearly see the name Eadburg, preceded by a cross at the start.

765
01:29:48,470 --> 01:29:54,230
Certain letters within the inscription do appear to suggest that it was written in the old English vernacular language,

766
01:29:54,440 --> 01:30:02,150
not in Latin, like the main text, which is really exciting. In addition to taking recordings of the inscription on page 18,

767
01:30:02,600 --> 01:30:09,260
John took recordings of other pages of manuscript Selden Super 30, which appeared to have something etched into them.

768
01:30:09,830 --> 01:30:18,370
This has revealed an astonishing number of marginal additions which have never been noticed before. Unlike the letters spotted by Lowe on page 47,

769
01:30:18,560 --> 01:30:22,160
which were etched into the parchment with a sharp implement, probably a knife,

770
01:30:22,520 --> 01:30:26,960
and are easy to see, these new inscriptions are almost invisible to the naked eye.

771
01:30:27,590 --> 01:30:31,490
The lines are far shallower, some less than a fifth of the width of a human hair.

772
01:30:31,910 --> 01:30:38,090
This indicates that they were made using a different type of implement, perhaps a stylus, using a technique known as drypoint.

773
01:30:39,290 --> 01:30:43,880
The newly discovered inscriptions reveal that Eadburg's name, written in full,

774
01:30:44,300 --> 01:30:49,760
was incised into the parchment of manuscript Selden Supra five times on five different pages.

775
01:30:52,140 --> 01:30:56,160
The instances of Eadburg's name which have been discovered so far are shown here.

776
01:30:56,230 --> 01:31:03,130
And for each you've got the recording taken by ARCHiOx and then a digitally annotated image just immediately below that.

777
01:31:04,200 --> 01:31:11,280
Certain letter forms, particularly the distinctive A, U and G, are common to all the newly identified inscriptions.

778
01:31:12,120 --> 01:31:16,170
This could suggest that the same scribe made all of these editions.

779
01:31:16,770 --> 01:31:20,910
Though I can't currently say definitively who added Eadburg's name to this book,

780
01:31:21,300 --> 01:31:23,670
It is possible that it was the Eadburg herself.

781
01:31:24,990 --> 01:31:33,210
Although small in scale and feint, these inscriptions showed that someone was evidently keen to preserve Eadburg's name in the pages of this book.

782
01:31:33,870 --> 01:31:36,990
Whoever was responsible for these editions,

783
01:31:37,110 --> 01:31:44,310
they provide clear evidence of a reader's interaction with this manuscript and of that active engagement with the text it contains.

784
01:31:45,180 --> 01:31:53,910
I am hopeful that as I study the inscriptions further, I may uncover more clues about who this reader was and why they added Eadburg's name to this book.

785
01:31:54,690 --> 01:31:57,480
Eadburg's identity also remains a mystery for now.

786
01:31:57,930 --> 01:32:04,530
There are at least nine known English women who have that name who lived sometime between the mid-sixth and mid-tenth centuries.

787
01:32:05,070 --> 01:32:11,130
It's also possible that the Eadburg of manuscript Selden Supra 30 left no other trace on the historical record.

788
01:32:12,210 --> 01:32:18,570
Further analysis of these editions may provide more clues about who Eadburg was and how she's linked to this book.

789
01:32:21,010 --> 01:32:28,000
So alongside Eadburg's name - everyone loves these, they're so cool! - alongside Eadburg's name,

790
01:32:28,270 --> 01:32:35,050
Several small, intriguing drypoint drawings have also been discovered, and just four examples are shown here.

791
01:32:35,920 --> 01:32:42,940
Some are clearly human figures, though further investigation is needed to establish exactly what or who they depict.

792
01:32:43,930 --> 01:32:51,340
The inscription added to the low margin of page 13 which is the top right image has been mostly lost when the lower margin was cut away.

793
01:32:51,610 --> 01:32:54,190
And we don't currently know when or why that happened.

794
01:32:54,940 --> 01:33:01,570
All the figures, like the written inscriptions, are very small and some like those on page 11, which is the bottom left image,

795
01:33:01,840 --> 01:33:07,330
might have been made by incising a line around thumb or finger to form the outline of the figure.

796
01:33:08,710 --> 01:33:15,430
Analysis of these drawings and of the other inscriptions is still in its early stages, but I'm hopeful that, as I study them further,

797
01:33:15,910 --> 01:33:24,280
I will be able to better understand their meanings and significance. Without the technology used by ARCHiOx to record these inscriptions.

798
01:33:24,580 --> 01:33:30,250
These drawings and Eadburg's name would have remained hidden in the margins of manuscripts Selden Supra 30.

799
01:33:30,880 --> 01:33:36,460
The discovery of Eadburg's name provides tantalising new evidence of apparent links between women,

800
01:33:36,700 --> 01:33:39,700
books and literate culture in the early medieval period.

801
01:33:40,570 --> 01:33:46,540
It also shows the potential of this technology to help us learn more about surviving early medieval manuscripts

802
01:33:46,720 --> 01:33:52,450
and to bring to light the readers and users of these books in a whole new way through the marks they left behind.

803
01:33:52,960 --> 01:34:05,980
Thank you. And just to follow up, in case you're wondering, the one on the top right was SpongeBob!

804
01:34:07,810 --> 01:34:17,110
And just to finish off with really, this I think is a a perfect example of serendipity in academic research,

805
01:34:17,350 --> 01:34:22,900
but that serendipity is not just Jess's sharp eyes spotting something nobody had seen before,

806
01:34:23,200 --> 01:34:32,920
but the coming together of that observation with this new initiative in the Bodleian and and the Bodleian's willingness to try it out,

807
01:34:32,950 --> 01:34:41,589
to see what would happen, to see what would come next. And, and it's a real moment to thank the Bodleian for their openness,

808
01:34:41,590 --> 01:34:45,430
not just allowing PhD students in to look at some of their most precious books,

809
01:34:46,030 --> 01:34:52,780
but also for that open mindedness to try something new with these extraordinary resources.

810
01:34:52,930 --> 01:35:05,489
So thank you. Perhaps we should mention as well, Matthew Holford, who's one of our fantastic curators

811
01:35:05,490 --> 01:35:13,500
and Martin Kauffmann for making the connection between the project and Jess's research, which has made this happen.

812
01:35:14,340 --> 01:35:23,940
So I'm going to introduce Alessandro Bianchi, who is our Manager of the Japanese Library and Curator of Japanese Rare Books and Manuscripts.

813
01:35:23,940 --> 01:35:29,220
And he's going to talk about Recreating Materiality: 3D Recordings from Japanese Originals.

814
01:35:29,970 --> 01:35:34,850
Thank you very much, John.

815
01:35:36,230 --> 01:35:39,950
So looking at selected case studies from our Japanese collections,

816
01:35:39,950 --> 01:35:44,840
I would like to show how we can use 3D recordings to study the materiality of an object,

817
01:35:45,170 --> 01:35:51,740
investigate its manufacturing process, and even examine and describe its condition and damage.

818
01:35:52,370 --> 01:35:57,710
So my first example is a painted album

819
01:36:00,190 --> 01:36:08,139
that comprise of a set of decorated portrait cards which I will use to illustrate how

820
01:36:08,140 --> 01:36:12,550
this technology can improve remote consultation and examination of primary sources.

821
01:36:12,850 --> 01:36:16,480
So what happens if there is an image online and you can't see the original, right?

822
01:36:16,900 --> 01:36:17,320


823
01:36:18,280 --> 01:36:26,590
If you look at traditional photography, we see that the painting layer is almost as if this portrait card was painted on top of a white page.

824
01:36:27,370 --> 01:36:35,320
But if we compare the 3D recordings, we'll see that the portrait card actually sits on the paper and it's pasted on top of it.

825
01:36:36,460 --> 01:36:45,120
This can be further seen by looking at this slanted 3D rendering and the high resolution surface data

826
01:36:45,140 --> 01:36:50,150
extracted using photometric technology reveals some of the material features on the manuscript

827
01:36:50,150 --> 01:36:55,280
which again are not seen it with standard photography and we can use this to study the

828
01:36:55,280 --> 01:37:02,240
complexity of the painting layers as well as the different texture of the paper.

829
01:37:02,450 --> 01:37:13,790
How thick the card is and so forth. So this technology brings the remote user and reader much closer due to regional.

830
01:37:13,970 --> 01:37:17,060
Despite looking at an object through a screen.

831
01:37:17,930 --> 01:37:27,240
Now, the same technology can be used also to identify different printing techniques and gain insights about the manufacturing process of various works.

832
01:37:27,830 --> 01:37:37,820
And for example, if we take a look at this, this is a page from an artist book, and we compare traditional photography with the 3D rendering.

833
01:37:38,210 --> 01:37:46,430
We see that there is a lot going on, in the centre, a rectangular area, rather flat and slightly sunken.

834
01:37:46,670 --> 01:37:51,380
And then there are a lot of creases and all around the printed text.

835
01:37:52,310 --> 01:37:58,550
Now these are traces left during the production of this and of this object,

836
01:37:58,850 --> 01:38:04,610
and they suggest that a special technique of stencil printing, called katazome was used to produce this book.

837
01:38:05,480 --> 01:38:12,020
So the artist first applied the thick coating of a special resist dye paste using a mask.

838
01:38:12,140 --> 01:38:16,250
And this is the flat, rectangular area that you see around the printed text.

839
01:38:17,030 --> 01:38:21,200
Then the text was dyed. It was printed in yellow.

840
01:38:21,230 --> 01:38:31,070
Using a yellow dye. And finally, the excess paste was removed, washing the page in abundant water.

841
01:38:31,340 --> 01:38:38,990
And so the creases are the results of this process of the page and the paper drying.

842
01:38:39,500 --> 01:38:43,309
So none of this is possible to see from normal photography.

843
01:38:43,310 --> 01:38:49,220
But again, these 3D renderings allow us to look at the objects as if we had an original in front of us.

844
01:38:51,780 --> 01:38:59,969
Conversely, if we look at an example of a 3D rendering, a 3D recording of a woodblock print, so relief printing

845
01:38:59,970 --> 01:39:05,850
in this case, we'll see that there is a highly textured surface.

846
01:39:06,000 --> 01:39:15,660
It recreates the outline of the colourful image. Now, these raised and sunken areas, and I would like to show you here and a blown up detail.

847
01:39:16,560 --> 01:39:23,520
These areas are the traces left on the surface of the paper by the wood blocks that were used during the manufacturing process.

848
01:39:24,510 --> 01:39:32,760
Now, woodcuts like this one were created by hand pressing a sheet of paper against the printing block using a flat bat,

849
01:39:33,210 --> 01:39:36,870
and in the case of polychrome prints, multiple blocks were used.

850
01:39:37,800 --> 01:39:40,650
And this process was repeated over and over again for each colour block.

851
01:39:42,030 --> 01:39:48,300
For example, if we look at the kimono, the flower pattern on this kimono here,

852
01:39:48,570 --> 01:39:54,670
since the original blocks of this print were preserved, we know that it was created in two steps.

853
01:39:54,690 --> 01:40:02,489
First, the silhouettes, the pink silhouette of the flower was printed, and then using a second block, the red accents were added.

854
01:40:02,490 --> 01:40:11,400
And so all the details of the petals. This means that one single sheet of paper was pressed multiple times against various blocks.

855
01:40:11,820 --> 01:40:19,460
And understandably, during this process, each block left a microscopic dent on the paper's surface and these dents, which are

856
01:40:19,620 --> 01:40:24,060
not visible with an observation of the original

857
01:40:24,870 --> 01:40:30,270
just with our naked eyes, are revealed by photometric technology,

858
01:40:30,630 --> 01:40:34,950
which allow us to see how the paper's surface was altered during the printing process.

859
01:40:35,790 --> 01:40:39,360
So if we take a look at the detail from this Kabuki actor print,

860
01:40:40,170 --> 01:40:45,630
we can understand how 3D recordings reveal this alteration on the surface.

861
01:40:46,590 --> 01:40:51,810
Now you'll notice that there are sunken areas corresponding to the gold accents on the kimono.

862
01:40:52,620 --> 01:40:56,850


863
01:40:58,110 --> 01:41:05,820
You can see them here, here, and on the hilt of the sword.

864
01:41:06,120 --> 01:41:13,110
This is because the metallic pigment was overlaid on top of the blue and in the red areas using the second block.

865
01:41:13,440 --> 01:41:18,330
Therefore,

866
01:41:18,430 --> 01:41:27,150
Firstly the blue and red layers were printed, and then using a second block, the gold was was applied, thus creating a deeper sunken area.

867
01:41:29,310 --> 01:41:35,580
This is, it's very important for us to look at these new material evidence that we managed

868
01:41:35,580 --> 01:41:41,030
to record digitally because they can provide information about how a print was made,

869
01:41:41,040 --> 01:41:44,760
even though the original printing blocks are not surviving.

870
01:41:46,980 --> 01:41:54,690
If we look at the detail of this kimono sleeve, for example, and we look at the 3D rendering, we see that there is a lot going on here.

871
01:41:55,950 --> 01:42:06,000
We can distinguish three levels, a raised area in white, and medium layer in pink and this sunken area in red.

872
01:42:07,650 --> 01:42:11,430
So what does this tell us? What does this rendering tell us about the production of this print?

873
01:42:12,120 --> 01:42:16,710
If we look at this small portion and we try to isolate each colour,

874
01:42:17,610 --> 01:42:23,219
we understand that the pink pigment was applied first and this little misregistration in the print

875
01:42:23,220 --> 01:42:27,570
can give it away that they didn't create a complex pattern like the one that we see here,

876
01:42:27,900 --> 01:42:33,870
but it was most likely one single block with a circular shape cut into it, in the surface.

877
01:42:34,950 --> 01:42:41,160
Then when the pink layer was printed, the woodblock

878
01:42:41,170 --> 01:42:48,569
compressed the paper's surface and the white circles which correspond to the area left unprinted look raised because again,

879
01:42:48,570 --> 01:42:53,160
no block was pressed against them.  The red pattern,

880
01:42:53,160 --> 01:43:01,920
was then printed on top of the pink layer and thus this is why it appears sunken here in the 3D renderings.

881
01:43:01,920 --> 01:43:10,350
And we have these three layers, a printed layer on top with the white circles, first pressing with the pink layer, and then the red area which is sunken.

882
01:43:11,920 --> 01:43:17,550
So even if we no longer possess the original blocks, by examining sunken areas and raised areas,

883
01:43:17,670 --> 01:43:21,600
in these 3D renderings, we can, so to speak, reverse engineer the printing process.

884
01:43:21,990 --> 01:43:27,900
And we can understand how this kimono and certain effects created with relief printing were made.

885
01:43:29,930 --> 01:43:35,480
Now, while studying Japanese woodcuts, we stumbled across some unexpected results.

886
01:43:36,140 --> 01:43:44,360
For example, in this print, which looks in perfect condition from normal photography and even from observation with the naked eye,

887
01:43:45,200 --> 01:43:53,180
reveals a very clear crease here in the middle and multiple creases horizontal on top.

888
01:43:55,280 --> 01:44:05,150
This gave us the idea to use this technology to study damage and improve our assessment of the condition of an object.

889
01:44:05,450 --> 01:44:08,959
And we started experimenting with other works

890
01:44:08,960 --> 01:44:17,600
On paper in particular, we tried out mapping the damage, the extent and type of damage in this 14th century, printed hand scroll.

891
01:44:18,140 --> 01:44:28,070
And we obtained very good results using a depth map or a heat map filter to identify the extent of the damage, but also the nature of the damage.

892
01:44:28,220 --> 01:44:34,860
Whether the paper was and raised and

893
01:44:35,120 --> 01:44:44,120
whether it was detached from the mounting or whether it was a sunken area,

894
01:44:44,120 --> 01:44:48,140
for example, worm-hole or other kinds of damage such as wear and tear.

895
01:44:48,530 --> 01:44:56,810
And you can see the orange part, the raised one and the lime-green ones are the worm-holes, for example, or parts where the paper is missing.

896
01:44:57,730 --> 01:45:04,580
But this technology has far greater potential and can be used to analyse, describe and study damage in a completely new way.

897
01:45:05,270 --> 01:45:10,520
And so we joined hands with colleagues in our Conservation Department, Marinita Stiglitz and Robert Minte,

898
01:45:11,390 --> 01:45:18,740
and we began experimenting with more complex scenarios, such as analysing the damage in the painted layer of the 17th century manuscript.

899
01:45:19,880 --> 01:45:26,720
And you can see here from the initial recordings already that the damaged areas are very clear.

900
01:45:26,720 --> 01:45:38,900
And if we apply the depth map, the crackling of the gold foil here in the clouds on top, flaking pigments, other abrasions,

901
01:45:39,020 --> 01:45:48,830
And even you can see here that there are fibres in the paper substrate that are creating damage to the painting layer.

902
01:45:48,980 --> 01:45:52,090
Become very clear. But there is more.

903
01:45:52,100 --> 01:46:01,590
And using these 3D renderings we could, so to speak, we can create a digital environment

904
01:46:01,770 --> 01:46:12,740
that allows you to perform incredibly complex observations as well as very detailed and precise, accurate measurements.

905
01:46:13,280 --> 01:46:23,389
And we can recreate therefore a non-invasive and contactless digital environment to study and look at damages, and look at objects in a very

906
01:46:23,390 --> 01:46:36,890
new and effective way that is incredibly respectful of of the original objects and allow us to look at things under a different light,

907
01:46:37,700 --> 01:46:41,260
but also collaborate remotely with colleagues across the globe.

908
01:46:41,270 --> 01:46:49,070
So, for example, if you were to share this data with colleagues in Japan, we would be allowed to do that without having to send the objects there.

909
01:46:49,730 --> 01:46:54,170
And it's also possible to, as has been shown before,

910
01:46:54,350 --> 01:47:04,970
recreate illuminations and create a raking light effect without having to shine lights directly onto objects,

911
01:47:05,240 --> 01:47:09,710
which is a great thing in the case of Japanese paintings, because the pigments are very volatile.

912
01:47:10,040 --> 01:47:12,589
So there is a lot of potential here.

913
01:47:12,590 --> 01:47:22,880
And I think that if we continue working on this technology, implementing AI and other machine learning technology,

914
01:47:23,000 --> 01:47:28,010
it would be really a game-changer and it would allow us to change the way we look at

915
01:47:28,010 --> 01:47:33,290
things and the way we work and research and approach matteriality in a digital way.

916
01:47:33,620 --> 01:47:49,190
Thank you. Our last speaker is Chiara Betti. The starting point for ARCHiOx was to record,

917
01:47:49,220 --> 01:47:53,030
lots of the copper plates from the Bodleian's collections, the Rawlinson copper plates.

918
01:47:53,540 --> 01:48:00,920
And she's going to talk about the collection, which is being researched by her.

919
01:48:05,430 --> 01:48:11,380
Thank you, John. As John said, I'm Chiara Betti and I'm a third year Collaberative Doctoral Partnership PhD

920
01:48:11,490 --> 01:48:21,510
student, and I'm studying the Rawlinson copper plates, 755 of them. And they are kept at the Bodleian Library.

921
01:48:22,770 --> 01:48:30,630
So in the next few slides, I will demonstrate how the ARCHiOx renderings have really transformed my research.

922
01:48:31,060 --> 01:48:38,940
I will start with a very brief introduction on Richard Rawlinson and his collection, and again, an introduction to his printing plates.

923
01:48:38,940 --> 01:48:47,970
And then I will delve into two case studies the first, a mezzotint plate, the portrait of Lady Brandon, and then the so called Invidia plates.

924
01:48:49,860 --> 01:48:59,810
So Richard Rawlinson was an 18th century antiquarian collector and clergyman, and under many aspects he embodied the typical early modern collector.

925
01:48:59,820 --> 01:49:04,500
He came from an upper class family. He had a very good education.

926
01:49:05,070 --> 01:49:12,930
He was a fellow of the Royal Society in the Society of Antiquities of London, and therefore he was very well-connected with other peers.

927
01:49:13,770 --> 01:49:21,089
However, he was one of the very few gentlemen who was lucky enough to go on a Grand Tour before it became popular

928
01:49:21,090 --> 01:49:30,720
Late in the 18th century. He spent six years between France and Italy, and there he had the chance to see many art collections, many private libraries.

929
01:49:31,350 --> 01:49:41,280
And he also purchased many books and objects in general, which are scattered between the Bodleian, the Ashmolean and the various GLAM museums at Oxford.

930
01:49:42,720 --> 01:49:49,380
So Rawlinson amassed a very impressive collection. He's very often described as a universal collector.

931
01:49:49,770 --> 01:49:57,809
I also call him a hoarder because he amassed over 5000 rare books and manuscripts, prints,

932
01:49:57,810 --> 01:50:05,340
drawings, coins, hundreds of seals and seal matrices and various curiosities,

933
01:50:05,340 --> 01:50:14,880
including skeletons, herbaries, chalices, most of which today lost and of course, 755 printing plates.

934
01:50:16,710 --> 01:50:21,210
So the Rawlinson plates are a unique collection

935
01:50:21,690 --> 01:50:30,090
Because of its scale, no other private collector has amassed such a big collection of printing plates.

936
01:50:30,600 --> 01:50:38,220
And the production and acquisition of this collection is exceptionally well documented by manuscript material, which is mostly at the Bodleian

937
01:50:38,310 --> 01:50:46,800
And at St John's College The collection of plates consists of two groups, the commissioned copper plates,

938
01:50:47,160 --> 01:50:57,000
which Rawlinson commissioned to illustrate his other collections and the second-hand ones that he purchased from auctions or directly from engravers.

939
01:50:57,810 --> 01:51:07,650
The subjects range from topographical views, seals, portraits, medals, many, many subjects.

940
01:51:08,220 --> 01:51:14,610
And we find some of the most prominent engravers in the Rawlinson collection of plates like George Vertue,

941
01:51:14,880 --> 01:51:22,410
George White, Michael Burghers, Wenceslaus Hollar, John Smith and hundreds of other famous British engravers.

942
01:51:23,490 --> 01:51:34,770
So let's delve into the first case study, the mezzotint plate of Lady Brandon. So you can see an impression and a traditional colour recording.

943
01:51:35,940 --> 01:51:38,630
This is a mezzotint by John Smith,

944
01:51:39,120 --> 01:51:47,850
also called John Smith, the mezzotinter after William Wissing. And this is one of three of Smith's copperplate in the Rawlinson collection,

945
01:51:49,080 --> 01:51:54,690
so mezzotint plates are the worst preserved and that is due to their intrinsic fragile nature.

946
01:51:55,380 --> 01:52:02,910
A mezzotint plate can only pull a handful of very good impressions before having to be reworked or re engraved.

947
01:52:06,060 --> 01:52:15,150
But ARCHiOx renderings have really transformed  my research into mezzotint plates and allowed me to see details I could not really see with the naked eye.

948
01:52:15,150 --> 01:52:24,330
And features that do not show in the impressions. So Smith's plates were defaced, so cancelled after he his death.

949
01:52:24,660 --> 01:52:31,080
As per his will, but the printer, publisher decided to cancel the cancelling signs,

950
01:52:31,080 --> 01:52:36,180
the scratches, to continue printing the plates and of course, the impressions.

951
01:52:36,930 --> 01:52:41,729
And in one of the plates you can see the scratch with the naked eye.

952
01:52:41,730 --> 01:52:47,970
But on this one I could only see it by zooming in thanks to the ARCHiOx rendering.

953
01:52:48,360 --> 01:52:56,219
And there, there is a scratch which is really barely visible on the plates.

954
01:52:56,220 --> 01:53:00,480
But this demonstrates that it's an original John Smith plate.

955
01:53:00,480 --> 01:53:06,959
So this scratch was deleted, so polished away so that the plate could be reprinted again.

956
01:53:06,960 --> 01:53:16,530
But this is the definitive proof that it's an original Smith's plate, that it's not a copy which often happened in the 18th century.

957
01:53:16,980 --> 01:53:21,510
So this is the scratch. And again, it's not visible in the impression.

958
01:53:24,450 --> 01:53:38,129
The second case study, are the so-called Invidia plates. It's a group it's a group of three very small plates, which are part of a series of 12 small plates.

959
01:53:38,130 --> 01:53:47,040
And they're after much larger engravings of an Italian series, the Speculum Romanae Magnificentiae.

960
01:53:47,040 --> 01:53:50,640
And they all represent ruins from Rome.

961
01:53:52,140 --> 01:54:01,830
These 12 plates are likely to have been purchased by Rawlinson when he was in Italy or commissioned soon after his trip to Italy.

962
01:54:02,560 --> 01:54:07,139
And we have the original wrapper in the Rawlinson collection.

963
01:54:07,140 --> 01:54:17,580
So in the Rawl Copperplates a.0  there is the wrapper and you can probably read 'Views of different Palaces in Rome'.

964
01:54:18,090 --> 01:54:25,680
And we have a whole volume of original wrappers with annotations with their former content on the wrapper.

965
01:54:27,600 --> 01:54:37,259
So this is the first of the three plates. All of the 12 plates have holes at the top, nail holes, and it's not clear why they have these holes,

966
01:54:37,260 --> 01:54:46,390
but probably the plates were used at some point as decorative elements for some furniture or some other purposes,

967
01:54:46,410 --> 01:54:52,889
I don't know, to be honest. So the first plate represents the Cerchio di Antonino Callo,

968
01:54:52,890 --> 01:54:58,320
and it's after Giovanni Battista de' Cavalieri, the same subject, of a much, much larger engraving.

969
01:55:00,250 --> 01:55:05,380
The second one is the Tempio di Fortuna Virile after again, de' Cavalieri.

970
01:55:06,310 --> 01:55:11,400
And the third one is Li Trofei di Mario after Etienne Dupérac.

971
01:55:12,160 --> 01:55:15,940
So these three plates are etched on the reverse.

972
01:55:15,940 --> 01:55:21,849
So out of the 12th plates, only these three have another design on the reverse.

973
01:55:21,850 --> 01:55:32,379
And the first one shows a man in Elizabethan fashion holding a scroll that reads 'To Invidia' and it's engraved in the right direction.

974
01:55:32,380 --> 01:55:35,050
And this is important. I will get there in a moment.

975
01:55:36,520 --> 01:55:47,740
The second one shows some legs and the third, a woman screaming and probably running away and a man with a hat.

976
01:55:47,890 --> 01:55:56,110
So if we place these three plates next to another, like a jigsaw puzzle, we get a new image.

977
01:55:58,180 --> 01:56:08,500
So this is as I went with my own devices, with my camera and with my phone, although the images were much, much worse.

978
01:56:08,890 --> 01:56:13,060
But the ARCHiOx rendering allowed me to take a further step forward.

979
01:56:13,090 --> 01:56:23,040
So John kindly stitched the three images together and these allowed me to try and search for impressions of the original plate.

980
01:56:23,050 --> 01:56:28,900
So the three plates come from a larger plate that was etched with this image

981
01:56:28,900 --> 01:56:35,470
and then was cut up to be reused and make the the series of Roman engravings.

982
01:56:36,460 --> 01:56:43,060
But the problem is that I couldn't find any impressions, and I think the answer might be in the in the scroll.

983
01:56:43,510 --> 01:56:46,570
So as I said, it's engraved in the right direction.

984
01:56:47,020 --> 01:56:56,860
And printing plates contain text. So the text must be engraved in the reverse so that when you print, it will show in the correct direction.

985
01:56:57,340 --> 01:57:02,650
So I've been wondering if this again was a decorative plate or it was a mistake.

986
01:57:03,430 --> 01:57:15,460
I don't know. But I initially thought that this was an allegory of envy, Invidia, because 'To Invidia' is the Latin or Italian word for envy.

987
01:57:16,390 --> 01:57:20,200
And the iconography is consistent with the depiction of Invidia

988
01:57:20,920 --> 01:57:24,310
So this is how it is traditionally represented.

989
01:57:24,320 --> 01:57:30,850
But a few days ago when I was preparing for this conference and I took another look at the ARCHiOx

990
01:57:31,270 --> 01:57:35,640
stitched image, I saw a couple of details that I hadn't noticed before.

991
01:57:35,650 --> 01:57:41,440
So the man at the back has crutches and there is a table with a tablecloth.

992
01:57:43,970 --> 01:57:51,630
And that made me think because this is not the allegory of envy.

993
01:57:51,650 --> 01:57:58,660
This is a representation from the parable of the rich man and Lazarus, because this is how it's traditionally represented.

994
01:57:58,740 --> 01:58:04,980
So I have an image. So this is a German print from the Monogrammist DS.

995
01:58:05,420 --> 01:58:10,340
And that is Lazarus with crutches. And this is the table with a rich man.

996
01:58:10,400 --> 01:58:20,180
And the representation of Invidia in the Rawlinson plate is unusual but not unlikely so Invidia

997
01:58:20,180 --> 01:58:28,190
is usually represented as a cat in this kind of parable or representation of the parable and not as a woman.

998
01:58:28,190 --> 01:58:32,900
But it's consistent with the with the story, with the parable.

999
01:58:33,860 --> 01:58:40,339
So I hope I've demonstrated how the ARCHiOx rendering has really transformed my research.

1000
01:58:40,340 --> 01:58:44,780
Without those images, I wouldn't have been able to see those details.

1001
01:58:44,790 --> 01:58:51,790
And they also improved the fruition of the set of objects, printing plates are extremely difficult to photograph.

1002
01:58:51,800 --> 01:58:57,170
And every time I had to speak in public about my research and I have to use my own images,

1003
01:58:57,530 --> 01:59:05,330
it was very difficult and frustrating because you cannot convey the same excitement about your discoveries without good images.

1004
01:59:06,050 --> 01:59:10,040
It also helps with information for the general public because with good images,

1005
01:59:10,280 --> 01:59:16,399
the public, so non-experts can understand what we are talking about when we talk about printing plates,

1006
01:59:16,400 --> 01:59:21,200
especially the mezzotints you have seen, how difficult it is to photograph metal things,

1007
01:59:21,200 --> 01:59:29,109
but the 3D renderings have really, really transformed the experience for me, and for everybody else.

1008
01:59:29,110 --> 01:59:41,130
So thank you.

1009
01:59:42,190 --> 01:59:46,750
So, I've just been told that maybe isn't time for questions because we're running late.

1010
01:59:48,040 --> 01:59:57,230
But I would like to conclude by really saying a few words on behalf of Lady Hamlyn, who's arrived and has asked me to do it on her behalf.

1011
01:59:57,310 --> 02:00:04,660
So, first of all, I'd like to say a very, very big thank you to the Helen Hamlyn Trust, because without her support,

1012
02:00:05,260 --> 02:00:13,360
both what we're doing in Venice and what we're doing here with John and the team in the Imaging Services.

1013
02:00:13,360 --> 02:00:21,700
So, Lady Hamlyn, thank you very, very much. And I think there's a little present coming down for you to say thank you.

1014
02:00:23,500 --> 02:00:28,860


1015
02:00:41,070 --> 02:00:51,800
So I hope everyone here can feel the kind of excitement that comes out of the materiality of the things that are here in the Bodleian.

1016
02:00:52,440 --> 02:01:00,450
And, you know, for me, looking at all the different examples and thinking there are thousands and thousands

1017
02:01:00,450 --> 02:01:06,599
and thousands of objects in this collection that could benefit by the work.

1018
02:01:06,600 --> 02:01:10,470
That John's doing. So, of course, photography is critical.

1019
02:01:10,620 --> 02:01:15,780
Traditional photography and the great work that's gone on here for years and years is vital.

1020
02:01:16,410 --> 02:01:23,069
But this is another tool. It's another level that gives access to it.

1021
02:01:23,070 --> 02:01:26,520
And I kept thinking when I was listening, many years ago

1022
02:01:26,550 --> 02:01:38,700
Simon Schaffer and I curated an exhibition called Noise that was was mainly at Cambridge I'm afraid, but you know, we can't be perfect!

1023
02:01:39,420 --> 02:01:51,210
And in it we were looking, it was basically something looking at pattern recognition and looking at universal language and looking at data synthetics.

1024
02:01:51,780 --> 02:02:02,730
And there was one quote that really stuck in my mind when Patrick Blackett from Imperial College was asked what experimental physicists do.

1025
02:02:03,750 --> 02:02:07,020
He answered this question - and it's quite a long quote - but more or less

1026
02:02:07,020 --> 02:02:11,429
He says, You know, he has to blow glass, but he could never earn his money as a glassblower.

1027
02:02:11,430 --> 02:02:16,260
He has to make carpentry things, but he never earned his money as a carpenter.

1028
02:02:17,160 --> 02:02:19,230
He has to wire electrical circuits,

1029
02:02:19,530 --> 02:02:28,350
but he never, and I'm looking at Jorge, because that picture of the wooden frame with the lights and the improvisation that goes on.

1030
02:02:28,560 --> 02:02:35,490
And then he concludes and he said, You know, I do that for 95% of my time, but for the rest of my time,

1031
02:02:36,330 --> 02:02:44,940
I do what a physicist has to do, which is cultivate an intimacy with the behaviour of the physical world.

1032
02:02:45,690 --> 02:02:54,630
And I think what's going on in ARCHiOx is cultivating an intimacy with the surface of the objects that are here.

1033
02:02:54,960 --> 02:03:04,830
And you know, Jessica's research, I sit there, I feel in that sense that this is being brought forward because of the work,

1034
02:03:04,830 --> 02:03:10,410
John, and the team doing here and because of the work that goes on, in Factum, in Madrid.

1035
02:03:10,860 --> 02:03:18,480
And I think it's this sense of sharing, sharing insights, thoughts, ideas,

1036
02:03:18,690 --> 02:03:32,290
access and different kinds of visualisation that allows a real intimacy with knowledge to grow and expand and, you know, without Lady Hamlyn's support.

1037
02:03:33,300 --> 02:03:41,640
And it should be said, I mean, the Bodleian has been amazing. So they opened up access to what we're doing in Madrid.

1038
02:03:41,820 --> 02:03:45,459
It took quite a long time. Being amazing often

1039
02:03:45,460 --> 02:03:47,970
is something that happens over a period of time!

1040
02:03:48,510 --> 02:03:57,450
But, you know, now, once we started, I remember even, John, it's true, the first week when Jorge and Carlos and others and Celeste,

1041
02:03:57,450 --> 02:04:06,570
who was doing a lot of the Lucida recording here came, it was very obvious, very fast that something was happening.

1042
02:04:06,870 --> 02:04:09,870
And, you know, I feel we've just scratched the surface.

1043
02:04:10,170 --> 02:04:15,440
I feel there's four or five things that have come up that can really be followed.

1044
02:04:15,450 --> 02:04:23,550
I love the Japanese prints. I mean, for me, the sophistication of what was going on in those printmaking studios,

1045
02:04:23,910 --> 02:04:28,440
which is nonverbal, I mean, the printing of textures into the white.

1046
02:04:28,710 --> 02:04:33,720
So the white is reflecting differently in different parts of the print is an

1047
02:04:33,720 --> 02:04:39,240
awareness of the visual languages that artists and printmakers work with.

1048
02:04:39,630 --> 02:04:48,570
And I think how information is mediated and transformed from one state to another is exactly what ARCHiOx is focussed on.

1049
02:04:48,960 --> 02:04:57,060
And I'm incredibly proud and happy with the collaboration that's grown up and I hope that this is the beginning of something very significant.

1050
02:04:57,480 --> 02:05:02,280
So thank you all for being here. And there's an exhibition upstairs.

1051
02:05:02,280 --> 02:05:05,109
I don't quite know the protocol because I know there's certain

1052
02:05:05,110 --> 02:05:13,659
numbers here there and everywhere, but there's an exhibition upstairs and I hope many exhibitions will follow because actually, you know what

1053
02:05:13,660 --> 02:05:25,649
this technology does make possible is to show things in context where they've never been seen before and I think watching Nick Millea this morning,

1054
02:05:25,650 --> 02:05:29,460
and I'm very Nick's not here. But he had to go to London,

1055
02:05:30,030 --> 02:05:38,160
but looking at the facsimile of the Gough Map and he'd come running up because he'd been told the Gough Map was on display and

1056
02:05:39,910 --> 02:05:49,660
Of course it's a recreation of it, but it's a three dimensional print in colour with front and back that in exhibition is identical.

1057
02:05:50,080 --> 02:05:54,670
I mean, of course it doesn't contain exactly the same information because as you can see,

1058
02:05:54,940 --> 02:05:58,720
there's still a separation between the real world of what we're able to record,

1059
02:05:59,050 --> 02:06:05,800
and there are things that we're not recording and there's a lot of innovations that we still need to perfect.

1060
02:06:05,850 --> 02:06:15,760
So this is very much a work in progress, but the results are pretty amazing and I hope as many of you as possible can see them.

1061
02:06:16,210 --> 02:06:20,440
And I hope it'll be the beginning of many, many more exciting revelations.

1062
02:06:20,920 --> 02:06:24,890
So thank you all very much. And if you have questions, I'll be here.

1063
02:06:24,990 --> 02:06:30,100
I'm sure John will. And so off the record, please come and ask anything you like.

1064
02:06:30,140 --> 02:06:34,750
And Richard and Andy, and Jorge and Carlos, so, please.

1065
02:06:35,140 --> 02:06:38,740
And thank you all for coming. And thank you for sharing this great day.