1
00:00:01,251 --> 00:00:03,920
[Music] 
Jeremy: One of our favorite questions that we get,

2
00:00:03,920 --> 00:00:07,799
all the time, is what happens
when you fall into a black hole?

3
00:00:08,008 --> 00:00:11,636
It's a fascinating question
because it gets to all of the cool,

4
00:00:11,636 --> 00:00:14,472
weird, wild things about black holes.

5
00:00:14,472 --> 00:00:17,142
But it's technically a hard question to answer,

6
00:00:17,142 --> 00:00:18,893
especially with the visualizations.

7
00:00:19,019 --> 00:00:20,770
[Dramatic music]

8
00:00:24,899 --> 00:00:29,237
[Plunge: Behind the Scenes Creating a NASA Visualization]

9
00:00:29,362 --> 00:00:32,157
Narrator: The key to creating this visualization is scientist

10
00:00:32,157 --> 00:00:35,243
Jeremy Schnittman at NASA's Goddard Space Flight Center.

11
00:00:36,369 --> 00:00:41,708
Jeremy: All of the ways that we model black holes
with, with math and equations

12
00:00:41,708 --> 00:00:46,838
and computers, it all kind of falls apart
at the edge, at the event horizon,

13
00:00:46,838 --> 00:00:50,175
because everything either goes to zero or infinity,

14
00:00:50,467 --> 00:00:53,470
and it makes it really hard to solve the equations.

15
00:00:53,720 --> 00:00:56,639
So, this year we wanted to try to really

16
00:00:56,639 --> 00:01:01,644
to solve that problem, to get around
all of the mathematical challenges

17
00:01:01,644 --> 00:01:05,148
at the event horizon and take those cameras

18
00:01:05,148 --> 00:01:07,484
and take those visualizations inside.

19
00:01:09,069 --> 00:01:11,571
It was all this new physics that we were exploring,

20
00:01:11,571 --> 00:01:13,656
and that's what we do as scientists.

21
00:01:13,656 --> 00:01:17,410
We, we love to go beyond and learn new things.

22
00:01:18,244 --> 00:01:22,707
Narrator: Jeremy's computer code applies the 
laws of physics, including relativity,

23
00:01:22,957 --> 00:01:26,127
to individual photons, and tracks their paths.

24
00:01:26,920 --> 00:01:29,422
Jeremy: To get really just the photons that hit the camera,

25
00:01:29,422 --> 00:01:31,966
we have to go backwards in time.

26
00:01:31,966 --> 00:01:34,969
We start with the photons at the camera,

27
00:01:35,053 --> 00:01:38,807
and then we just follow them backwards
to see where they came from.

28
00:01:39,140 --> 00:01:42,185
Fortunately, all of the equations

29
00:01:42,644 --> 00:01:47,190
of relativity, and physics in general, are time reversible,

30
00:01:47,190 --> 00:01:50,318
so they work basically the same forwards and backwards.

31
00:01:51,653 --> 00:01:54,864
But if there's one place in the universe where they

32
00:01:54,864 --> 00:01:57,242
don't behave so well, it’s at the event horizon.

33
00:01:57,242 --> 00:02:00,036
Because things don't go forwards and backwards across

34
00:02:00,036 --> 00:02:01,246
the event horizon, 

35
00:02:01,246 --> 00:02:03,414
you know, so those kind of conceptual things,

36
00:02:03,957 --> 00:02:07,627
had to be really careful and think carefully about.

37
00:02:08,294 --> 00:02:13,424
All of this, of course, starts with just brainstorming
in a storyboard,

38
00:02:13,424 --> 00:02:15,093
kind of hashing out

39
00:02:15,552 --> 00:02:18,763
what this might look like
and what we hope it will look like.

40
00:02:19,013 --> 00:02:23,017
Then we have to try it again
in kind of low resolution to get

41
00:02:23,017 --> 00:02:28,273
a couple more sophisticated snapshots
to see how it will look.

42
00:02:28,857 --> 00:02:32,902
And it's an iterative process, that both involves
the technical problems

43
00:02:32,902 --> 00:02:34,654
that I'm trying to work through figuring

44
00:02:34,654 --> 00:02:38,074
“Oh, well, this this isn't working
because of a, of an equation.”

45
00:02:38,074 --> 00:02:41,077
And then feeding that back through the,

46
00:02:41,077 --> 00:02:44,539
the artists and the communicators that 

47
00:02:44,539 --> 00:02:48,126
will help say, “well, this isn't working
from a story point of view.” 

48
00:02:49,127 --> 00:02:52,422
Narrator: Jeremy worked with astrophysics science writer Frank Reddy

49
00:02:52,422 --> 00:02:54,424
and video producer Scott Wiessinger

50
00:02:54,424 --> 00:02:57,093
to refine and construct the final videos.

51
00:02:58,678 --> 00:03:01,514
Jeremy: That's, I really, actually one of the parts
of the problem I like the best:

52
00:03:01,514 --> 00:03:06,728
this, this teamwork between the technical
and the narrative part of it.

53
00:03:07,353 --> 00:03:11,441
I got to really invoke my Kip Thorne, “Interstellar,”

54
00:03:12,108 --> 00:03:15,486
physicist-artist combination, right?

55
00:03:15,737 --> 00:03:18,656
The strict physical reality
is that the camera's just going

56
00:03:18,656 --> 00:03:21,659
faster and faster as it gets closer to the black hole,

57
00:03:21,701 --> 00:03:25,663
which would make it realistic,
but very difficult to watch.

58
00:03:25,663 --> 00:03:28,666
It would just go “zooooooom” and you're done.

59
00:03:28,958 --> 00:03:32,837
And of course, all the interesting stuff
happens at the “zoooom” at the end.

60
00:03:33,171 --> 00:03:36,466
This is one of those places where you have to choose

61
00:03:36,758 --> 00:03:40,220
between the strict physical reality

62
00:03:40,553 --> 00:03:43,973
and the fact that you're really trying to tell a story.

63
00:03:44,307 --> 00:03:47,727
So, we had to take our artistic license and,

64
00:03:47,727 --> 00:03:51,981
and slow down the camera to help
share that exciting experience.

65
00:03:54,567 --> 00:03:57,195
To get a feel for what it's going to look like,

66
00:03:57,195 --> 00:04:01,449
you know, we start off with a low-resolution
image or maybe a few snapshots.

67
00:04:01,950 --> 00:04:04,702
And that sort of level of calculation

68
00:04:04,702 --> 00:04:08,456
is not too difficult to do on a laptop computer.

69
00:04:08,706 --> 00:04:09,332
Try it.

70
00:04:09,332 --> 00:04:10,041
It doesn't work.

71
00:04:10,041 --> 00:04:13,169
Try this. Doesn't work. Fix this.
Ah, it works.

72
00:04:13,419 --> 00:04:16,631
Okay, now that we've got something
that basically works,

73
00:04:17,298 --> 00:04:21,928
then we need to, to ramp up the resolution,
make a whole lot more frames.

74
00:04:21,928 --> 00:04:26,140
And at that point, the laptop computer
is really not powerful enough

75
00:04:26,140 --> 00:04:29,560
if we want to make this video in less than 100 years.

76
00:04:30,103 --> 00:04:33,648
So, at that point, we essentially take the same computer code

77
00:04:33,940 --> 00:04:37,151
and port it over to a supercomputer that can run

78
00:04:37,485 --> 00:04:41,155
many, many, many, many calculations all at the same time.

79
00:04:42,782 --> 00:04:44,826
Narrator: Jeremy loaded his code on

80
00:04:44,826 --> 00:04:47,203
the Discover supercomputer at Goddard.

81
00:04:47,495 --> 00:04:50,373
With 213,000 cores,

82
00:04:50,373 --> 00:04:54,085
Discover can process code at 8,280 trillion

83
00:04:54,085 --> 00:04:56,087
floating point operations every second.

84
00:04:57,171 --> 00:05:01,092
Data scientist Brian Powell helped
load and run Jeremy's code on Discover.

85
00:05:01,634 --> 00:05:05,138
The initial run required about 10,000 CPU hours

86
00:05:05,513 --> 00:05:08,516
and generated 10 terabytes of raw data.

87
00:05:10,727 --> 00:05:14,063
Jeremy: And it's a different job for each frame, right?

88
00:05:14,063 --> 00:05:17,984
We're making one image, and that takes a few hours.

89
00:05:17,984 --> 00:05:21,321
But then we have tens and tens of thousands of these frames

90
00:05:21,321 --> 00:05:22,488
to make the movie.

91
00:05:22,488 --> 00:05:26,326
And then the last part is the Discover generates

92
00:05:26,326 --> 00:05:30,788
all of these numerical files, just lists of numbers,

93
00:05:31,289 --> 00:05:35,335
and I have to pull those back and
convert those then into images.

94
00:05:35,335 --> 00:05:39,422
So, I have a couple of the last steps of converting

95
00:05:39,422 --> 00:05:43,009
just raw numbers into Jpeg files of images,

96
00:05:43,426 --> 00:05:46,846
also just running on my personal computer, 
because that's where

97
00:05:47,013 --> 00:05:50,808
you can kind of play around with them
and preview them in an easier way.

98
00:05:51,642 --> 00:05:55,938
Putting all that together,
we would find some weird properties,

99
00:05:55,980 --> 00:05:59,776
and you have to go back and figure out
what the problem was, work it out

100
00:05:59,776 --> 00:06:04,113
again on the local computer
and think you've got it fixed

101
00:06:04,113 --> 00:06:07,075
and put it all back through the Discover cluster again.

102
00:06:07,992 --> 00:06:12,330
Narrator: To show everything an observer might see
on the trips into or around a black hole,

103
00:06:13,289 --> 00:06:16,709
Jeremy generated two map projections of the entire sky.

104
00:06:18,044 --> 00:06:21,047
One worked within standard 360 options,

105
00:06:21,172 --> 00:06:24,592
and the other, called a Mollweide projection, is often

106
00:06:24,592 --> 00:06:28,137
used to show the sphere of sky from astronomical observations.

107
00:06:28,721 --> 00:06:31,974
Jeremy: And then there was really a third one
that that we worked

108
00:06:31,974 --> 00:06:33,976
on, which was the zoom in, right?

109
00:06:33,976 --> 00:06:36,604
So, we also have these zoom-in frames that are,

110
00:06:36,604 --> 00:06:38,773
I would say, more like a conventional camera,

111
00:06:38,773 --> 00:06:41,734
you know, a small field of view, right, right in front of you.

112
00:06:41,734 --> 00:06:45,029
And that was great for picking out
some of the really cool

113
00:06:45,363 --> 00:06:49,033
relativity features, like the photon rings. 

114
00:06:49,033 --> 00:06:54,372
As we both zoom in with the zoom-in frames
and physically zoom in, bringing

115
00:06:54,956 --> 00:06:58,835
the observer closer and closer
to the black hole, you get closer

116
00:06:58,835 --> 00:07:03,047
to this really cool feature
of the black hole called the photon ring.

117
00:07:03,423 --> 00:07:08,469
And this is where light gets bent
by the black hole 360 degrees around.

118
00:07:08,803 --> 00:07:11,597
And then beneath that, there's another set of photons

119
00:07:11,597 --> 00:07:15,726
that got wrapped around twice
by the black hole. Three times. 

120
00:07:15,768 --> 00:07:18,146
Like, it just goes forever and ever.

121
00:07:18,146 --> 00:07:21,274
Ring after ring after ring as we skirt

122
00:07:21,649 --> 00:07:26,028
that critical surface, getting closer
and closer to the event horizon

123
00:07:26,028 --> 00:07:29,991
and the, and the singularity. We're seeing more
and more of these light rings

124
00:07:30,199 --> 00:07:33,953
and also copies of the accretion disk and the galaxy

125
00:07:33,953 --> 00:07:38,166
get bent around and around the black hole, 

126
00:07:38,166 --> 00:07:42,170
and we're going to see more and more copies
of basically the entire sky,

127
00:07:42,503 --> 00:07:47,884
every, every loop around is another copy of the galaxies.

128
00:07:48,259 --> 00:07:49,886
And now we're almost hitting

129
00:07:49,886 --> 00:07:53,723
the singularity in the front and everything,
just goes [crunch sound] right at the end.

130
00:07:55,141 --> 00:07:58,478
Narrator: Even for someone as familiar with
black hole physics as Jeremy,

131
00:07:58,644 --> 00:08:02,565
watching the new visualization
led to some unexpected discoveries.

132
00:08:03,191 --> 00:08:04,901
Jeremy: A completely genuine surprise is

133
00:08:04,901 --> 00:08:07,945
we see at some point, very close to the end, we see,

134
00:08:08,321 --> 00:08:10,615
like the sky almost shrinks.

135
00:08:10,615 --> 00:08:12,825
And we figured out that was because of some of these

136
00:08:12,825 --> 00:08:16,370
relativistic effects of the whole, sky getting pinched

137
00:08:16,370 --> 00:08:19,999
towards you as all of those photons
hit you in the face.

138
00:08:20,917 --> 00:08:23,920
Narrator: Not all the surprises were welcome.

139
00:08:23,961 --> 00:08:27,340
After Jeremy produced the first full batch of zoom frames,

140
00:08:27,340 --> 00:08:29,717
a strange jitter was apparent.

141
00:08:30,176 --> 00:08:33,137
Jeremy: These jitter problems were the kind of things that,

142
00:08:33,137 --> 00:08:36,265
by definition, happen at the end
because we only saw them

143
00:08:36,265 --> 00:08:40,978
in the really high resolution, smooth, videos

144
00:08:40,978 --> 00:08:44,357
where you're getting a lot of frames right
next to each other.

145
00:08:45,149 --> 00:08:47,568
And what we think was happening

146
00:08:47,568 --> 00:08:53,533
was that you're basically stepping the camera
forward in time by, you know,

147
00:08:53,533 --> 00:09:00,206
big steps and this part in between
those steps wasn't perfectly smooth.

148
00:09:00,414 --> 00:09:04,293
So, it was like literally the difference
between going up and down stairs

149
00:09:04,544 --> 00:09:06,087
and going on a slide.

150
00:09:06,087 --> 00:09:09,048
And we were taking these steps,
which were kind of

151
00:09:09,048 --> 00:09:11,634
almost like jerking
the camera back and forth.

152
00:09:11,634 --> 00:09:17,014
So, we had to come up with a way of making
those steps into a nice, smooth ride.

153
00:09:17,682 --> 00:09:21,936
And then at the very end, where it's
almost like the inescapable jitters

154
00:09:21,936 --> 00:09:25,731
at the singularity, at the center of the black hole.

155
00:09:26,065 --> 00:09:29,485
This is where even our best equations have trouble.

156
00:09:29,485 --> 00:09:33,614
There are matrices and inversions and integrals

157
00:09:33,614 --> 00:09:37,785
that just struggle with the infinities

158
00:09:37,785 --> 00:09:41,581
that happen at the center of a black hole,
and there's almost no way around it.

159
00:09:46,002 --> 00:09:48,170
Narrator: With Black Hole Week 2024 looming,

160
00:09:48,170 --> 00:09:51,716
Jeremy and Brian Powell
rushed to solve the issue of the jitter

161
00:09:51,966 --> 00:09:55,052
and generate new, smoother frames for the final videos.

162
00:09:55,886 --> 00:09:57,805
Jeremy: And we, we've done this before.

163
00:09:57,805 --> 00:10:00,725
We've done this three or four times
before for Black Hole Week.

164
00:10:00,725 --> 00:10:01,851
We know the deadline,

165
00:10:01,851 --> 00:10:03,811
we know when we have to get everything done,

166
00:10:03,811 --> 00:10:07,356
we started over six months in advance because we knew

167
00:10:07,356 --> 00:10:10,359
we had to get all of this stuff done in time.

168
00:10:10,568 --> 00:10:13,571
And yet there's always that rush at the end.

169
00:10:13,738 --> 00:10:16,866
And it's stressful, but it's also fun, right?

170
00:10:17,199 --> 00:10:20,119
Because it's, you know, it's like,

171
00:10:20,119 --> 00:10:23,122
it's like real life! Physics!

172
00:10:23,456 --> 00:10:27,918
In my particular case, it was even more, ah, “exciting”

173
00:10:28,169 --> 00:10:32,757
because I was traveling out of town
for a family wedding that weekend.

174
00:10:33,090 --> 00:10:38,137
So, I was like, in this cabin in a location
wedding in the middle of nowhere,

175
00:10:38,137 --> 00:10:42,767
trying to get internet, working with Brian, on the Discover

176
00:10:42,767 --> 00:10:47,563
cluster, working with Frank and Scott,
trying to get the frames all set.

177
00:10:49,023 --> 00:10:50,441
So, it was a lot,

178
00:10:50,441 --> 00:10:54,945
but it was it was so worth it because the final
product was just tremendous, I think.

179
00:10:57,198 --> 00:10:59,241
Narrator: Despite all the math, science,

180
00:10:59,241 --> 00:11:02,244
and computer work underpinning this project,

181
00:11:02,286 --> 00:11:05,414
the end result has a strong emotional resonance.

182
00:11:05,831 --> 00:11:08,542
Jeremy: I have to admit I do get a little bit emotional

183
00:11:08,542 --> 00:11:13,047
when we watch the final project,
especially once we add the soundtrack in.

184
00:11:14,465 --> 00:11:17,551
The, the sound it just, you know,
there is a scene

185
00:11:17,551 --> 00:11:20,554
where we're going into the black hole
and we’re zooming in,

186
00:11:20,721 --> 00:11:25,226
and the music is getting like, [beat sound]
and it's just, it's it's really exciting.

187
00:11:25,226 --> 00:11:27,144
Even though I know it's going to happen.

188
00:11:27,603 --> 00:11:30,606
[Dramatic music]

189
00:11:43,244 --> 00:11:46,247
[Music fades]

190
00:11:46,247 --> 00:11:47,790
Jeremy: We the scientists, right?

191
00:11:47,790 --> 00:11:49,125
We're people, too.

192
00:11:49,125 --> 00:11:50,710
We are our audience.

193
00:11:51,252 --> 00:11:55,297
We love to have our imagination engaged.

194
00:11:55,297 --> 00:11:57,341
We love to see pretty pictures.

195
00:11:57,341 --> 00:12:00,010
We love to see the beauty of nature.

196
00:12:00,010 --> 00:12:03,097
I mean, that's why most of us do what we do.

197
00:12:03,097 --> 00:12:05,266
Because of the excitement.

198
00:12:05,266 --> 00:12:07,560
Because of the wonder.

199
00:12:07,560 --> 00:12:11,313
I think a lot of times
when I make the visualizations

200
00:12:11,313 --> 00:12:16,527
and we have the dramatic music
and the zoom ins and the zoom outs,

201
00:12:16,527 --> 00:12:21,157
like that's where I get to really express
how I feel about black holes.

202
00:12:23,284 --> 00:12:26,370
[Music fades out]

203
00:12:26,829 --> 00:12:29,915
[NASA]