Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:00,200 --> 00:00:00,900
in this lesson
2
00:00:00,900 --> 00:00:03,966
we're gonna learn about film grain and digital noise
3
00:00:04,600 --> 00:00:06,966
so as we've made our travels through the camera
4
00:00:06,966 --> 00:00:08,133
we've gotten through the lens
5
00:00:08,133 --> 00:00:09,500
we've gotten through the aperture
6
00:00:09,500 --> 00:00:11,700
and we are finally hitting the sensor
7
00:00:11,866 --> 00:00:16,133
now here you might expect that everything is good to go
8
00:00:16,133 --> 00:00:19,566
however there are some problems that arise
9
00:00:19,566 --> 00:00:21,833
in the sensor or on our film
10
00:00:21,933 --> 00:00:24,933
namely film grain and digital noise
11
00:00:24,933 --> 00:00:26,266
so let's take a look at a
12
00:00:26,266 --> 00:00:29,400
few examples of this before we add it into our comp
13
00:00:30,000 --> 00:00:32,633
so this is a very um
14
00:00:34,366 --> 00:00:36,200
extreme version of film grain
15
00:00:36,200 --> 00:00:39,466
and we can see here that are areas
16
00:00:39,466 --> 00:00:42,666
even areas that are in focus have this strange pattern
17
00:00:42,700 --> 00:00:45,833
and this pattern is in our different color channels
18
00:00:45,966 --> 00:00:48,666
and this if the picture was moving
19
00:00:48,666 --> 00:00:51,500
would actually be changing every single frame
20
00:00:52,266 --> 00:00:54,000
now on the other side of the world
21
00:00:54,000 --> 00:00:55,066
um excuse me
22
00:00:55,066 --> 00:00:58,200
in digital we have something called digital noise
23
00:00:58,200 --> 00:00:59,400
and this happens
24
00:00:59,400 --> 00:01:01,000
uh very similar to film grains
25
00:01:01,000 --> 00:01:03,866
so let's take a look at that oops
26
00:01:03,866 --> 00:01:06,300
so here is a picture without any noise
27
00:01:06,300 --> 00:01:09,400
and here is a picture shot at a much higher ISO
28
00:01:09,800 --> 00:01:11,500
if you recall our second lesson
29
00:01:11,500 --> 00:01:14,533
we talked about the ISO and the sensitivity
30
00:01:14,533 --> 00:01:18,533
as being aspects of film and digital sensors
31
00:01:18,533 --> 00:01:21,533
which controls how much light they pick up
32
00:01:21,533 --> 00:01:23,766
and how much light they amplify
33
00:01:24,200 --> 00:01:25,966
so let's take a look at why
34
00:01:26,133 --> 00:01:28,166
a film grain happens
35
00:01:28,400 --> 00:01:31,200
now this is actually a picture of film crystal
36
00:01:31,200 --> 00:01:34,200
silver highlight crystals on a piece of film
37
00:01:34,266 --> 00:01:36,366
and you'll notice that there are many different shapes
38
00:01:36,366 --> 00:01:38,100
and many different sizes
39
00:01:38,266 --> 00:01:41,200
now this is because we want our grains of film
40
00:01:41,200 --> 00:01:43,833
to be different and to be sort of random
41
00:01:44,000 --> 00:01:47,066
so what we get is we get this sort of random
42
00:01:47,066 --> 00:01:51,266
um collection of shapes and sizes
43
00:01:51,266 --> 00:01:54,066
and they combine to create film grain
44
00:01:54,600 --> 00:01:55,933
so we can see in this image
45
00:01:55,933 --> 00:01:58,900
we have random sizes of this film grain
46
00:01:58,900 --> 00:02:02,600
some of it is very small and some of it is very large
47
00:02:03,466 --> 00:02:04,333
now you might ask
48
00:02:04,333 --> 00:02:07,033
why don't they just have a single size
49
00:02:07,066 --> 00:02:08,566
and the reason for this is
50
00:02:08,566 --> 00:02:11,366
larger crystals are more sensitive to light
51
00:02:11,400 --> 00:02:13,500
but you need smaller crystals so that you
52
00:02:13,500 --> 00:02:15,100
can have fine details
53
00:02:15,100 --> 00:02:17,900
so they have to make a very measured
54
00:02:17,900 --> 00:02:21,166
uh sort of mixture whenever you create film
55
00:02:21,466 --> 00:02:24,400
now this picture you see here is one layer
56
00:02:24,400 --> 00:02:28,566
uh you can have or you have a unique
57
00:02:28,566 --> 00:02:31,200
layer of crystals in the red
58
00:02:31,300 --> 00:02:33,700
green and blue channels
59
00:02:33,700 --> 00:02:36,700
so what that means is film green changes
60
00:02:36,700 --> 00:02:38,200
whether it's in the red
61
00:02:38,266 --> 00:02:40,266
green or blue channels
62
00:02:41,000 --> 00:02:43,400
so if you want more sensitive film
63
00:02:43,400 --> 00:02:46,266
or if you wanna shoot in a darker environment
64
00:02:46,266 --> 00:02:49,333
you need to have larger grain sizes or a larger
65
00:02:49,333 --> 00:02:51,266
ISO speed film
66
00:02:51,466 --> 00:02:55,166
however that leads to more grain and more
67
00:02:55,366 --> 00:02:57,466
visually noticeable grain
68
00:02:57,466 --> 00:02:59,233
so you always have to be careful
69
00:02:59,300 --> 00:03:01,466
uh to get the shot you want
70
00:03:01,466 --> 00:03:03,500
but with the least amount of grain
71
00:03:04,366 --> 00:03:05,500
now on the digital side
72
00:03:05,500 --> 00:03:06,366
you might be thinking well
73
00:03:06,366 --> 00:03:08,166
all the digital sensors are the same
74
00:03:08,166 --> 00:03:10,100
why would we have this problem here
75
00:03:10,100 --> 00:03:12,166
and the problem is not a a
76
00:03:12,200 --> 00:03:13,533
a problem of size
77
00:03:13,533 --> 00:03:16,100
it is a problem of intensity
78
00:03:16,133 --> 00:03:19,433
so what the sensor is doing whenever it is
79
00:03:19,566 --> 00:03:23,600
increasing its ISO or increasing its sensitivity
80
00:03:23,600 --> 00:03:26,066
is to increase the
81
00:03:26,266 --> 00:03:29,033
amount of energy it is pushing or the
82
00:03:29,100 --> 00:03:32,466
or decrease the amount of energy it needs to
83
00:03:32,466 --> 00:03:34,166
light up a cell
84
00:03:34,166 --> 00:03:36,000
so what happens is the noise
85
00:03:36,000 --> 00:03:37,666
just the general noise
86
00:03:37,700 --> 00:03:39,133
uh from different lights
87
00:03:39,133 --> 00:03:40,400
from electrical noise
88
00:03:40,400 --> 00:03:41,666
from heat uh
89
00:03:41,666 --> 00:03:44,733
begin to create these digital noise effects
90
00:03:44,733 --> 00:03:45,733
and as humans
91
00:03:45,733 --> 00:03:47,333
digital noise is
92
00:03:47,333 --> 00:03:49,366
not as nice as film grain
93
00:03:49,466 --> 00:03:51,233
film grain has a
94
00:03:51,333 --> 00:03:53,300
uh very positive quality
95
00:03:53,300 --> 00:03:54,200
it is random
96
00:03:54,200 --> 00:03:55,600
it is randomly distributed
97
00:03:55,600 --> 00:03:57,266
it has random shapes
98
00:03:57,266 --> 00:04:00,566
and those are all aspects of grain that we like however
99
00:04:00,566 --> 00:04:01,900
with digital noise
100
00:04:01,966 --> 00:04:04,366
our our sensors are based off of uh
101
00:04:04,366 --> 00:04:07,366
what's called a Bayer pattern or a Bayer filter
102
00:04:07,500 --> 00:04:09,200
which essentially looks like this
103
00:04:09,200 --> 00:04:10,900
so you can see here of course
104
00:04:10,900 --> 00:04:14,500
our pixels are arranged in a very orderly fashion
105
00:04:14,500 --> 00:04:18,166
what this creates is very orderly digital noise
106
00:04:18,166 --> 00:04:18,966
and our eyes
107
00:04:18,966 --> 00:04:21,900
and our brains can pick up on that repeating pattern
108
00:04:21,900 --> 00:04:23,500
and we really don't like it
109
00:04:23,500 --> 00:04:26,166
especially if it's moving around in our frame
110
00:04:26,166 --> 00:04:27,900
a digital noise is
111
00:04:28,066 --> 00:04:30,166
uh almost always to be avoided
112
00:04:30,166 --> 00:04:30,900
unless of course
113
00:04:30,900 --> 00:04:32,300
you're going for a cheap
114
00:04:32,400 --> 00:04:33,500
camera shot
115
00:04:33,500 --> 00:04:35,433
and then you can add in digital noise
116
00:04:35,533 --> 00:04:36,600
however in our case
117
00:04:36,600 --> 00:04:40,000
we're gonna want to remove as much noise as possible
118
00:04:40,000 --> 00:04:42,400
and then add in grain
119
00:04:43,133 --> 00:04:44,400
so let's jump over to nuke
120
00:04:44,400 --> 00:04:47,133
this is gonna be a relatively simple operation after
121
00:04:47,133 --> 00:04:48,500
all the work we've done
122
00:04:48,533 --> 00:04:51,600
and let's just drop in a grain node
123
00:04:51,600 --> 00:04:53,000
and this comes with many
124
00:04:53,000 --> 00:04:57,033
useful presets for creating grain inside of our image
125
00:04:57,366 --> 00:04:58,966
now if we take a look at this node
126
00:04:58,966 --> 00:05:00,900
uh separate from everything else
127
00:05:00,900 --> 00:05:02,233
we can see that
128
00:05:02,466 --> 00:05:03,533
if we turn off
129
00:05:03,533 --> 00:05:04,900
apply only through alpha
130
00:05:05,466 --> 00:05:07,266
well I guess we don't have any black grain
131
00:05:07,266 --> 00:05:09,100
and I don't wanna add any just yet
132
00:05:09,100 --> 00:05:10,566
so let's plug this back in
133
00:05:10,566 --> 00:05:11,966
we'll notice that the grain
134
00:05:11,966 --> 00:05:13,766
changes every single frame
135
00:05:13,766 --> 00:05:15,000
if we jump forward
136
00:05:15,000 --> 00:05:15,733
and jump back
137
00:05:15,733 --> 00:05:16,766
one or two frames
138
00:05:16,766 --> 00:05:18,400
give us a second to load up
139
00:05:19,300 --> 00:05:21,800
so we can see the grain is changing size
140
00:05:21,800 --> 00:05:23,300
shape and position
141
00:05:23,300 --> 00:05:24,600
every single frame
142
00:05:24,600 --> 00:05:25,500
and that is of course
143
00:05:25,500 --> 00:05:27,000
because a piece of film
144
00:05:27,000 --> 00:05:28,300
or a roll of film
145
00:05:28,366 --> 00:05:29,866
every single grain is
146
00:05:29,866 --> 00:05:30,866
unique to that
147
00:05:30,866 --> 00:05:32,166
explicit frame
148
00:05:32,166 --> 00:05:34,566
so it's gonna change every single frame
149
00:05:35,000 --> 00:05:36,500
so one thing we need to do
150
00:05:36,500 --> 00:05:37,933
before calling this good to go
151
00:05:37,933 --> 00:05:38,966
is we need to match
152
00:05:38,966 --> 00:05:40,500
the size of our grain
153
00:05:40,500 --> 00:05:41,533
because currently
154
00:05:41,533 --> 00:05:44,500
the grain note is set to a 2K image
155
00:05:44,700 --> 00:05:45,866
if you have a weird question Mark
156
00:05:45,866 --> 00:05:46,866
you can see here
157
00:05:46,866 --> 00:05:47,900
these are the correct
158
00:05:47,900 --> 00:05:50,600
sizes for 2K scans
159
00:05:50,666 --> 00:05:53,100
now we are not using a 2K image
160
00:05:53,100 --> 00:05:54,733
so we're gonna have to do a little math
161
00:05:54,733 --> 00:05:56,100
I'm sorry to
162
00:05:56,100 --> 00:05:57,766
fix the grain node
163
00:05:58,966 --> 00:06:00,466
so let's go to our size here
164
00:06:00,466 --> 00:06:02,233
and we have our full size format
165
00:06:02,300 --> 00:06:04,266
I'm just gonna create this
166
00:06:04,466 --> 00:06:06,533
and fix the pixel aspect ratio
167
00:06:06,533 --> 00:06:09,300
so this is gonna be our 7 20 p size
168
00:06:09,300 --> 00:06:11,600
so we have 1280 by 7 20
169
00:06:11,900 --> 00:06:13,900
and so now let's go into our
170
00:06:13,900 --> 00:06:16,200
let's first write down what the values are
171
00:06:16,200 --> 00:06:17,700
for a 2 k plate
172
00:06:17,800 --> 00:06:20,200
we can find this in our project settings
173
00:06:20,200 --> 00:06:21,633
full size format
174
00:06:21,766 --> 00:06:23,233
and it's the 2 k
175
00:06:23,366 --> 00:06:27,300
uh Super Cinema or excuse me super 35 millimeter
176
00:06:27,400 --> 00:06:33,300
so 2,048 by 1556 okay
177
00:06:33,800 --> 00:06:36,833
so let me just bring up a quick notepad here
178
00:06:37,700 --> 00:06:42,100
so our full size are 2K format is a
179
00:06:42,400 --> 00:06:44,400
2,048 by 1 5
180
00:06:44,400 --> 00:06:46,700
5 6 and our current format
181
00:06:46,866 --> 00:06:49,866
is 1280 by 7 20
182
00:06:49,900 --> 00:06:52,033
so what we need to do is we need to calculate
183
00:06:52,066 --> 00:06:54,333
how much smaller the grain should be
184
00:06:54,333 --> 00:06:56,300
so it is the same size
185
00:06:56,333 --> 00:06:58,233
in our smaller frame
186
00:06:58,300 --> 00:06:59,000
and of course
187
00:06:59,000 --> 00:07:00,800
depending on what size frame you're using
188
00:07:00,800 --> 00:07:02,766
what size footage you're working with
189
00:07:02,766 --> 00:07:05,333
uh you'll need to put your own numbers here
190
00:07:05,333 --> 00:07:07,166
and then do the calculations
191
00:07:07,666 --> 00:07:10,266
so if I bring up a handy dandy calculator
192
00:07:10,266 --> 00:07:11,833
and I plug the numbers in
193
00:07:12,400 --> 00:07:15,266
let's plug in our current frame size 1280
194
00:07:15,300 --> 00:07:17,000
and then let's divide this by
195
00:07:17,000 --> 00:07:19,400
the target frame size of a 2K
196
00:07:19,666 --> 00:07:20,933
so for the x axis
197
00:07:20,933 --> 00:07:22,966
we get point six two five
198
00:07:23,066 --> 00:07:25,266
and for the y axis
199
00:07:25,266 --> 00:07:27,166
we get 7 20 / 5
200
00:07:27,166 --> 00:07:31,666
5 6.462 okay
201
00:07:31,700 --> 00:07:34,133
now I'm just gonna use point six two five
202
00:07:34,133 --> 00:07:37,366
I'm not going to change the aspect ratio of the grain
203
00:07:37,366 --> 00:07:38,366
now that's gonna
204
00:07:38,366 --> 00:07:40,566
make things look very strange
205
00:07:40,800 --> 00:07:41,600
so to do this
206
00:07:41,600 --> 00:07:43,000
let's go into our grain
207
00:07:43,000 --> 00:07:45,633
and this is controlled in the size
208
00:07:45,766 --> 00:07:48,366
so let's do 3.3 times
209
00:07:48,533 --> 00:07:50,566
our point six two five
210
00:07:50,666 --> 00:07:52,000
hit enter and nuke
211
00:07:52,000 --> 00:07:52,766
nicely enough
212
00:07:52,766 --> 00:07:55,066
does the math for us so again
213
00:07:55,066 --> 00:07:56,400
multiply by point two
214
00:07:56,400 --> 00:07:59,233
six five multiply by 2.265
215
00:07:59,400 --> 00:08:00,466
excuse me point six
216
00:08:00,666 --> 00:08:01,766
six two five
217
00:08:01,766 --> 00:08:03,000
uh good old dyslexia
218
00:08:03,533 --> 00:08:06,166
and we have the correct grain size
219
00:08:06,166 --> 00:08:08,466
now if we compare this with the original size
220
00:08:08,466 --> 00:08:09,266
it's gonna be
221
00:08:09,266 --> 00:08:10,266
very different
222
00:08:10,266 --> 00:08:11,600
so let's take a look at here
223
00:08:11,600 --> 00:08:14,333
so the original size had very large grain
224
00:08:14,333 --> 00:08:15,866
while with our change
225
00:08:15,866 --> 00:08:17,366
we have much nicer
226
00:08:17,366 --> 00:08:19,200
much smaller grain
227
00:08:19,600 --> 00:08:20,666
now we could of course
228
00:08:20,666 --> 00:08:23,800
use the even smaller value of point four
229
00:08:23,800 --> 00:08:25,566
I believe 6 2 it was
230
00:08:25,600 --> 00:08:26,900
but again uh
231
00:08:26,900 --> 00:08:28,900
I believe this is close enough
232
00:08:30,000 --> 00:08:33,233
so now we have green on our frame
233
00:08:33,500 --> 00:08:34,100
and of course
234
00:08:34,100 --> 00:08:36,433
you can come in play with a different presets
235
00:08:36,566 --> 00:08:39,300
uh going and change your intensity values
236
00:08:39,300 --> 00:08:41,500
maybe this grain is a little bit too strong
237
00:08:41,566 --> 00:08:42,300
uh remember
238
00:08:42,300 --> 00:08:44,000
grain increases
239
00:08:44,000 --> 00:08:44,966
the darker it is
240
00:08:44,966 --> 00:08:45,800
you're trying to shoot
241
00:08:45,800 --> 00:08:48,433
so grain increases in
242
00:08:48,666 --> 00:08:50,033
uh darker shots
243
00:08:50,166 --> 00:08:52,400
now if you're outside with a lot of sun
244
00:08:52,400 --> 00:08:54,700
you're probably not gonna need to use ISO
245
00:08:54,900 --> 00:08:57,766
16 film unless you
246
00:08:57,766 --> 00:09:00,300
have 10 neutral density filters on your camera
247
00:09:00,300 --> 00:09:02,566
which it would be rather strange indeed
248
00:09:02,566 --> 00:09:03,566
so in this case
249
00:09:03,566 --> 00:09:04,933
when we have a very bright scene
250
00:09:04,933 --> 00:09:05,700
we can pull
251
00:09:05,700 --> 00:09:08,300
the grain back a little bit in its intensity
252
00:09:08,800 --> 00:09:09,533
because again
253
00:09:09,533 --> 00:09:13,900
we would be using a faster film stock to shoot outside
254
00:09:13,900 --> 00:09:14,733
and a lower
255
00:09:14,733 --> 00:09:16,800
ISO if we were using the digital
256
00:09:16,800 --> 00:09:17,833
counterpart
257
00:09:18,200 --> 00:09:20,100
now you notice the intensity values are actually
258
00:09:20,100 --> 00:09:21,100
rather interesting
259
00:09:21,100 --> 00:09:22,933
the red is the lowest
260
00:09:22,933 --> 00:09:24,266
next comes the green
261
00:09:24,266 --> 00:09:25,266
and then the blue
262
00:09:25,266 --> 00:09:26,900
is very large
263
00:09:27,766 --> 00:09:28,900
or very intense
264
00:09:28,900 --> 00:09:31,400
and this is a very common thing with film stocks
265
00:09:31,400 --> 00:09:32,666
the blue uh
266
00:09:32,666 --> 00:09:34,300
grain is much
267
00:09:34,300 --> 00:09:35,900
much more intense than the
268
00:09:35,900 --> 00:09:39,433
red or green uh grain
269
00:09:40,133 --> 00:09:40,966
so here we go
270
00:09:40,966 --> 00:09:42,466
we can see this nice
271
00:09:42,666 --> 00:09:43,400
subtle grain
272
00:09:43,400 --> 00:09:45,300
it's gonna add just a little bit of movement
273
00:09:45,300 --> 00:09:46,566
a little bit of
274
00:09:46,566 --> 00:09:48,500
uh what looks like sharpness
275
00:09:48,500 --> 00:09:50,400
even though it's just random noise
276
00:09:50,400 --> 00:09:51,333
it actually will
277
00:09:51,333 --> 00:09:53,700
make our picture look sharper
278
00:09:53,700 --> 00:09:56,333
and be much more pleasing to look at
279
00:09:56,333 --> 00:09:57,100
because again
280
00:09:57,100 --> 00:09:59,200
it makes us feel like we're looking
281
00:09:59,200 --> 00:10:02,800
at a piece of film okay great
282
00:10:04,066 --> 00:10:06,733
so let's just talk briefly about a grain workflow
283
00:10:06,733 --> 00:10:07,666
again in CG
284
00:10:07,666 --> 00:10:09,533
you can just add grain here at the end
285
00:10:09,533 --> 00:10:12,400
uh in the real world or in real footage
286
00:10:12,400 --> 00:10:14,700
you wanna take grain out
287
00:10:14,700 --> 00:10:18,033
you usually wanna take grain out of your backgrounds
288
00:10:18,133 --> 00:10:21,300
if you have multiple real world elements to composite
289
00:10:21,300 --> 00:10:24,366
and then regrain them at the end however
290
00:10:24,366 --> 00:10:25,566
if you are shooting or if
291
00:10:25,566 --> 00:10:28,200
you're working off of a fixed back plate
292
00:10:28,200 --> 00:10:28,766
for example
293
00:10:28,766 --> 00:10:31,166
if this was real world camera footage
294
00:10:31,166 --> 00:10:32,600
we could bring uh
295
00:10:32,600 --> 00:10:33,566
this in here
296
00:10:33,566 --> 00:10:35,766
composite whatever we need on top
297
00:10:35,766 --> 00:10:39,066
and then add grain to these new elements
298
00:10:39,066 --> 00:10:41,366
that match our background grain
299
00:10:41,366 --> 00:10:42,266
and that would
300
00:10:42,266 --> 00:10:45,600
stay truest to the background and not modify it
301
00:10:45,600 --> 00:10:49,266
in case you're any more stricter pipeline okay great
302
00:10:49,266 --> 00:10:51,966
so that brings us to the end of this course on creating
303
00:10:51,966 --> 00:10:53,966
photo realistic camera lens effects
304
00:10:53,966 --> 00:10:55,366
here inside of nuke
305
00:10:55,366 --> 00:10:59,000
uh we started with a rather simple uh image
306
00:10:59,000 --> 00:11:02,600
and by using the power of nuke and no base compositing
307
00:11:02,600 --> 00:11:04,666
as well as learning about a few
308
00:11:04,766 --> 00:11:06,866
uh pieces of light physics
309
00:11:06,866 --> 00:11:08,466
we were able to take again
310
00:11:08,466 --> 00:11:11,666
what was a relatively boring 3D render
311
00:11:11,666 --> 00:11:14,033
and turn it into something nice
312
00:11:14,566 --> 00:11:16,466
or at least something nicer
313
00:11:17,066 --> 00:11:19,266
so the image quality uh
314
00:11:19,666 --> 00:11:21,266
has degraded a little bit
315
00:11:21,266 --> 00:11:23,533
but I think you'll agree that it looks much
316
00:11:23,533 --> 00:11:26,066
much nicer with all of our changes
317
00:11:26,066 --> 00:11:29,466
because we tried to mimic reality wherever we could
318
00:11:29,466 --> 00:11:31,500
and use the rules of nature
319
00:11:31,500 --> 00:11:32,666
the rules of
320
00:11:32,666 --> 00:11:35,200
that control our glass lenses
321
00:11:35,200 --> 00:11:38,200
and the rules of light to add in some of those
322
00:11:38,200 --> 00:11:40,466
subtle irregularities and problems
323
00:11:40,466 --> 00:11:43,800
that we will have when shooting in natural conditions
324
00:11:44,300 --> 00:11:44,966
so as always
325
00:11:44,966 --> 00:11:46,466
I hope you've enjoyed this course
326
00:11:46,466 --> 00:11:49,333
and I can't wait to see how you apply these techniques
327
00:11:49,333 --> 00:11:50,533
to your own project
22602
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.