1
00:00:00,000 --> 00:00:11,437
[MUSIC]

2
00:00:11,437 --> 00:00:13,881
So, I am Sergio Nascimento.

3
00:00:13,881 --> 00:00:21,005
I work on color vision for many years at Minho University in Portugal.

4
00:00:21,005 --> 00:00:25,908
And I've tried to understand how we perceive colors and specifically and

5
00:00:25,908 --> 00:00:30,901
been interested in the last few years how we perceive colors in paintings.

6
00:00:30,901 --> 00:00:36,535
So one of the things I've been interested in is trying to quantify as I said,

7
00:00:36,535 --> 00:00:39,881
the colors of paintings and try to perceive or

8
00:00:39,881 --> 00:00:44,743
measure information that is invisible light that we cannot see.

9
00:00:44,743 --> 00:00:50,319
So actually, our color vision is limited because we sample the visible spectrum

10
00:00:50,319 --> 00:00:55,499
only in three regions, in the red, green and blue region of the spectrum.

11
00:00:55,499 --> 00:00:58,530
And this means that the result of information

12
00:00:58,530 --> 00:01:01,411
in the visual spectrum that we cannot see.

13
00:01:01,411 --> 00:01:05,518
So the techniques that we use that are called Spectral Imaging,

14
00:01:05,518 --> 00:01:08,188
they allows us to see much more than that.

15
00:01:08,188 --> 00:01:15,492
So because they sample the visible spectrum in many more spectral bands.

16
00:01:15,492 --> 00:01:21,515
So it's like if we have a system that has ten times more resolution than our eye for

17
00:01:21,515 --> 00:01:22,232
colors.

18
00:01:22,232 --> 00:01:26,526
So and we use that to digitalize paintings and

19
00:01:26,526 --> 00:01:31,046
then to get more information from paintings and

20
00:01:31,046 --> 00:01:34,325
we actually can see with our eyes.

21
00:01:34,325 --> 00:01:38,011
This technique allows us to have a very detailed information from the painting,

22
00:01:38,011 --> 00:01:39,156
spectral information.

23
00:01:39,156 --> 00:01:43,380
So from each pixel of the painting, so from each small area of the painting,

24
00:01:43,380 --> 00:01:45,954
we have a spectrum, a full visible spectrum.

25
00:01:45,954 --> 00:01:48,618
And that allows us to do a lot of things, for

26
00:01:48,618 --> 00:01:53,221
example allows us to compute very precisely the colors of the paintings.

27
00:01:53,221 --> 00:01:58,710
And these computations can give

28
00:01:58,710 --> 00:02:03,589
estimates of how many colors

29
00:02:03,589 --> 00:02:07,861
we see in the paintings.

30
00:02:07,861 --> 00:02:10,658
And then we can compare it for example for

31
00:02:10,658 --> 00:02:14,939
the same painter,different paintings and how they evolve or

32
00:02:14,939 --> 00:02:19,483
they change the technique to produce a certain number of colors.

33
00:02:19,483 --> 00:02:26,122
So for example, there are painters that statically using colors but

34
00:02:26,122 --> 00:02:31,515
in very limited ways, just some colors, a few colors.

35
00:02:31,515 --> 00:02:34,935
And then they evolve to use many more colors and

36
00:02:34,935 --> 00:02:39,764
to create the perception of many more colors in their paintings.

37
00:02:39,764 --> 00:02:44,138
So another thing that we can do with this kind of data is to simulate

38
00:02:44,138 --> 00:02:48,604
the painting is going to be perceived in different light sources.

39
00:02:48,604 --> 00:02:54,078
For example, how it;s going to be perceived with LED or with a specific LED.

40
00:02:54,078 --> 00:02:59,442
And that is very important because in the museums when they have to select a light

41
00:02:59,442 --> 00:03:04,725
source for their paintings, they need to know what is the effect that the light

42
00:03:04,725 --> 00:03:09,876
source is going to have on the painting, on the appearance of the painting.

43
00:03:09,876 --> 00:03:12,629
How does spectral vision work in practice?

44
00:03:12,629 --> 00:03:16,181
So we have one painting here that is illuminated so

45
00:03:16,181 --> 00:03:18,389
we can see all sorts of colors.

46
00:03:18,389 --> 00:03:20,846
And we have the spectral imaging system here.

47
00:03:20,846 --> 00:03:26,271
What it does is to take multiple pictures at different wavelength bands,

48
00:03:26,271 --> 00:03:28,466
different chromatic bands.

49
00:03:28,466 --> 00:03:32,383
And with that information it builds spectrum for

50
00:03:32,383 --> 00:03:36,022
each big cell for a small bit of the debating.

51
00:03:36,022 --> 00:03:40,595
For example, this part here that is that you can see that is red,

52
00:03:40,595 --> 00:03:43,489
we have the spectral information here.

53
00:03:43,489 --> 00:03:48,312
So you can see the spectrum here that there's more intensity in

54
00:03:48,312 --> 00:03:50,594
the red part of the spectrum.

55
00:03:50,594 --> 00:03:54,568
So we have this information for all pixels of the painting.

56
00:03:54,568 --> 00:03:58,970
So in summary, we can say that we can build the spectral fingerprint of

57
00:03:58,970 --> 00:04:09,520
the painting with this data.