1
00:00:25.425 --> 00:00:28.880
Let's explore how to use AI for animation production

2
00:00:28.904 --> 00:00:30.730
Including sound, music and

3
00:00:30.754 --> 00:00:35.687
Idea generation

4
00:00:36.336 --> 00:00:40.071
AI Tools that Can be Used for Idea Generation

5
00:00:40.705 --> 00:00:43.782
Currently, using AI tools

6
00:00:43.806 --> 00:00:47.057
For idea generation

7
00:00:47.081 --> 00:00:49.304
Such as the widely known ChatGPT is

8
00:00:49.609 --> 00:00:51.675
A very popular method

9
00:00:51.700 --> 00:00:55.901
And you can achieve very accurate results with it

10
00:00:56.077 --> 00:01:01.186
When you launch ChatGPT, you'll see version 3.5

11
00:01:01.210 --> 00:01:02.743
If you subscribe

12
00:01:02.768 --> 00:01:06.171
You'll be able to access version 4.0

13
00:01:06.483 --> 00:01:10.251
While I may not know all the technical details

14
00:01:10.275 --> 00:01:14.203
It is stated that

15
00:01:14.227 --> 00:01:17.417
ChatGPT 4.0 has significantly greater computational abilities

16
00:01:17.441 --> 00:01:20.835
Capable of making many more judgments

17
00:01:21.657 --> 00:01:25.929
Compared to version 3.5

18
00:01:25.953 --> 00:01:29.581
Therefore

19
00:01:30.102 --> 00:01:33.366
I believe subscribing to get more accurate results is a good idea

20
00:01:34.126 --> 00:01:36.993
As you may have experienced

21
00:01:37.099 --> 00:01:41.939
Using version 3.5 to research various basic things

22
00:01:42.537 --> 00:01:47.357
Often yields inaccurate results

23
00:01:47.381 --> 00:01:52.275
So, since we're not searching for specific information

24
00:01:52.300 --> 00:01:53.831
Or anything like that right now

25
00:01:54.383 --> 00:01:56.361
Using version 3.5 is fine

26
00:01:56.753 --> 00:02:01.517
For example, to make sentences sound more natural

27
00:02:01.541 --> 00:02:04.749
I think using 4.0

28
00:02:04.773 --> 00:02:07.368
Would not be a bad idea for that purpose

29
00:02:07.392 --> 00:02:10.898
Typically, when you come up with ideas

30
00:02:11.403 --> 00:02:14.562
You write prompts in ChatGPT

31
00:02:14.980 --> 00:02:19.175
Due to ChatGPT being widely used in Korean as well

32
00:02:19.200 --> 00:02:23.666
You can obtain satisfactory results now that ChatGPT is being utilized in Korean

33
00:02:24.611 --> 00:02:28.005
But it's a essential to know how to write prompts effectively

34
00:02:28.029 --> 00:02:31.507
So, it would be beneficial

35
00:02:31.653 --> 00:02:35.809
To observe how other experts write prompts

36
00:02:36.654 --> 00:02:40.075
Also to see what I'm doing right now

37
00:02:40.100 --> 00:02:43.524
then mix them well to create effective prompts

38
00:02:43.984 --> 00:02:47.451
For instance, when I'm creating a story

39
00:02:47.916 --> 00:02:51.896
About a knight and a dragon

40
00:03:00.107 --> 00:03:02.800
And relevant information

41
00:03:06.964 --> 00:03:11.545
And I hit enter

42
00:03:11.819 --> 00:03:16.728
It will provide me brief plot summaries related to this

43
00:03:16.752 --> 00:03:20.888
But if you also want details

44
00:03:21.262 --> 00:03:30.228
Like shot breakdowns

45
00:03:30.252 --> 00:03:34.557
Character descriptions or camera movements included

46
00:03:34.581 --> 00:03:38.513
You can specify that

47
00:03:39.585 --> 00:03:41.881
When you write prompts

48
00:03:43.446 --> 00:03:45.134
In this manner

49
00:03:45.159 --> 00:03:48.999
You can obtain more accurate results

50
00:03:49.023 --> 00:03:49.803
For example

51
00:03:56.544 --> 00:04:02.042
If you're arranging it like a framed composition

52
00:04:07.599 --> 00:04:08.699
Or protagonist

53
00:04:16.392 --> 00:04:17.325
Background

54
00:04:25.387 --> 00:04:27.648
When you write prompts

55
00:04:27.672 --> 00:04:30.810
When you create an idea

56
00:04:31.379 --> 00:04:34.048
According to the 5W1H

57
00:04:35.053 --> 00:04:36.888
Listing the necessary things

58
00:04:37.363 --> 00:04:40.774
Will help create

59
00:04:41.136 --> 00:04:42.139
A more specific story

60
00:04:42.164 --> 00:04:44.632
For instance, if I were to go to this far

61
00:04:44.991 --> 00:04:48.209
ChatGPT now arranges scenes

62
00:04:48.233 --> 00:04:51.771
Into five segments as I desire

63
00:04:56.142 --> 00:04:58.770
Then arrange them into framed compositions

64
00:04:58.794 --> 00:05:02.550
And automatically generates descriptions of the characters

65
00:05:02.983 --> 00:05:05.528
Sets the background

66
00:05:05.987 --> 00:05:08.684
In Joseon Dynasty era

67
00:05:08.708 --> 00:05:12.324
Here, when you use the expression

68
00:05:17.040 --> 00:05:23.277
"in more detail"

69
00:05:23.302 --> 00:05:25.531
ChatGPT provides a more detailed plot for the story

70
00:05:27.784 --> 00:05:30.322
But if you read the story

71
00:05:30.346 --> 00:05:33.104
generated by ChatGPT

72
00:05:33.289 --> 00:05:39.026
You'll probably realize that

73
00:05:39.050 --> 00:05:42.336
Using it as a script would be quite challenging

74
00:05:43.029 --> 00:05:48.528
I also once conducted a class

75
00:05:48.815 --> 00:05:52.082
Where we used AI to create animation videos

76
00:05:53.434 --> 00:05:59.040
Most of the time, the resulting videos were quite typical

77
00:05:59.719 --> 00:06:02.839
For certain shots

78
00:06:03.071 --> 00:06:05.731
Students had to resort to drawing since there was no other option

79
00:06:05.756 --> 00:06:07.426
I also tried using it once

80
00:06:07.853 --> 00:06:10.367
It was quite challenging to get exactly what you need

81
00:06:10.391 --> 00:06:14.712
So I think it's best to use AI

82
00:06:14.736 --> 00:06:18.979
With the mindset of getting some ideas

83
00:06:19.743 --> 00:06:21.481
Rather than relying on it completely

84
00:06:22.361 --> 00:06:26.073
In a way, these AI tools can be seen as assistants to writers

85
00:06:26.097 --> 00:06:30.625
Helping them with certain aspects of their work

86
00:06:31.278 --> 00:06:33.712
And there's something called Bard

87
00:06:34.204 --> 00:06:39.280
Which is an AI tool for generating ideas

88
00:06:40.436 --> 00:06:44.606
Bard has now been replaced by Gemini

89
00:06:44.925 --> 00:06:47.843
Which assets with interview preparation, household routines

90
00:06:48.362 --> 00:06:51.259
Language learning plans, and place recommendations

91
00:06:51.283 --> 00:06:58.227
So, typically

92
00:06:58.251 --> 00:07:00.610
You might not use it for everyday idea generation

93
00:07:01.116 --> 00:07:05.457
The basic presets are indeed for interview preparation and household routines

94
00:07:05.942 --> 00:07:10.346
If you write prompts in a similar way here

95
00:07:13.870 --> 00:07:18.505
You can expect to get similar results

96
00:07:19.813 --> 00:07:25.107
This one might provide more details about each scene

97
00:07:25.486 --> 00:07:27.720
And the characters involves, right?

98
00:07:27.901 --> 00:07:31.838
It might give a more detailed description of the events

99
00:07:33.101 --> 00:07:35.147
It might come out in a more concise, summarized format

100
00:07:36.153 --> 00:07:39.141
Rather than in full sentences

101
00:07:39.726 --> 00:07:42.287
It seems like a good tool

102
00:07:42.703 --> 00:07:44.549
For generating ideas

103
00:07:45.234 --> 00:07:47.859
Once you've generate the ideas

104
00:07:48.517 --> 00:07:53.929
You can use AI tools like Midjourney or Runway

105
00:07:53.953 --> 00:07:57.307
To visualize them into images

106
00:07:57.433 --> 00:08:00.220
After visualizing them

107
00:08:01.813 --> 00:08:05.663
You can edit the images and add music on top

108
00:08:07.001 --> 00:08:10.310
If needed, you can also add voiceovers to the characters

109
00:08:11.079 --> 00:08:12.852
Which can be quite essential

110
00:08:13.256 --> 00:08:16.504
So, you can use AI for all these tasks

111
00:08:17.115 --> 00:08:21.772
Of course, to create higher-quality videos

112
00:08:22.338 --> 00:08:24.761
AI cannot accurately provide

113
00:08:25.367 --> 00:08:28.509
The exact sound effects needed for a scene

114
00:08:28.533 --> 00:08:31.275
Or the specific voice you envision for a character

115
00:08:31.299 --> 00:08:35.377
While it can't provide exact details

116
00:08:35.401 --> 00:08:38.750
It's still a useful tool for generating guidelines

117
00:08:38.774 --> 00:08:41.241
Like "roughly this kind of feel"

118
00:08:41.733 --> 00:08:44.145
Or if you consider this approach sufficient

119
00:08:44.169 --> 00:08:49.554
For creating high-quality videos

120
00:08:49.578 --> 00:08:53.076
Then that's okay too

121
00:08:53.100 --> 00:08:58.300
So, for example, with the Voice Over Generative Voice AI

122
00:08:58.905 --> 00:09:06.100
You choose your desired language

123
00:09:06.365 --> 00:09:09.558
Then write a script

124
00:09:09.583 --> 00:09:10.718
It's important to accept

125
00:09:11.125 --> 00:09:13.340
That it may not be perfect

126
00:09:13.937 --> 00:09:17.468
Then, you write a scrip like this and

127
00:09:17.492 --> 00:09:25.942
You select the voices for the characters that match the script

128
00:09:29.011 --> 00:09:30.970
Saying, "It's important to accept"

129
00:09:30.995 --> 00:09:32.955
"That failure is inevitable in life"

130
00:09:33.566 --> 00:09:36.897
Where you don't have to rely on voice actors for this

131
00:09:37.389 --> 00:09:40.757
It's something commonly used

132
00:09:40.887 --> 00:09:44.723
That many YouTubers are using these tools these days

133
00:09:44.747 --> 00:09:47.545
They often use AI tools

134
00:09:47.569 --> 00:09:51.803
To create virtual influencers

135
00:09:52.221 --> 00:09:54.947
Who can then host live broadcasts

136
00:09:55.160 --> 00:09:57.959
Thus, you can download these

137
00:09:57.983 --> 00:10:00.866
And use them later for video editing

138
00:10:01.544 --> 00:10:05.519
Of course, you can use them in other languages as well

139
00:10:05.544 --> 00:10:10.653
If you want your content to be broadcasted in Russia

140
00:10:10.677 --> 00:10:13.899
Or if you want Russians to understand it

141
00:10:13.923 --> 00:10:16.352
You can choose Russian

142
00:10:16.376 --> 00:10:17.906
Or you can choose German

143
00:10:22.823 --> 00:10:26.190
Then you can use the same script in German

144
00:10:26.908 --> 00:10:30.476
For the same content

145
00:10:31.261 --> 00:10:34.911
There are tools like that

146
00:10:36.240 --> 00:10:38.078
There's also a tool called Lovo

147
00:10:39.005 --> 00:10:41.934
You can think of this as

148
00:10:41.958 --> 00:10:43.660
A similar type of AI tool

149
00:10:43.684 --> 00:10:49.627
So, if you're creating a test

150
00:10:58.548 --> 00:11:02.412
You would start by creating a project

151
00:11:05.135 --> 00:11:06.582
Then selecting "AI Voice and Video Start Project"

152
00:11:11.960 --> 00:11:15.246
And then perhaps

153
00:11:15.270 --> 00:11:16.948
"Short Voiceover" to begin

154
00:11:20.100 --> 00:11:21.534
Then, the script would be generated

155
00:11:22.339 --> 00:11:25.264
By default

156
00:11:25.289 --> 00:11:28.781
You can choose the voice you want

157
00:11:28.805 --> 00:11:33.367
It can also be accompanied by a user-friendly face

158
00:11:34.712 --> 00:11:38.221
You can select the age

159
00:11:38.245 --> 00:11:41.302
Whether it's a child's voice or a more mature one

160
00:11:41.326 --> 00:11:43.842
As well as choose between female and male voices

161
00:11:44.362 --> 00:11:46.196
The purpose for which it's used

162
00:11:47.161 --> 00:11:48.650
Whether it's for audiobooks

163
00:11:48.674 --> 00:11:50.314
Education

164
00:11:50.338 --> 00:11:54.216
Marketing, etc., the tone will vary accordingly

165
00:11:54.240 --> 00:11:55.508
So, you can choose the tone of voice

166
00:11:56.104 --> 00:12:01.766
That's appropriate for your purpose

167
00:12:03.225 --> 00:12:06.695
If you want to do it in Korean

168
00:12:06.719 --> 00:12:10.302
You can change the language here to Korean

169
00:12:17.545 --> 00:12:19.076
After changing it to Korean

170
00:12:19.595 --> 00:12:22.430
You can then choose characters that match here

171
00:12:23.175 --> 00:12:25.145
Then when you press "Generate"

172
00:12:25.847 --> 00:12:27.147
Press "Generate" button

173
00:12:30.902 --> 00:12:33.517
It will speak with a slightly awkward

174
00:12:33.541 --> 00:12:37.754
Korean/English pronunciation

175
00:12:37.778 --> 00:12:40.463
If I write it in Korean

176
00:12:41.129 --> 00:12:42.817
"Hello, I'm a student"

177
00:12:45.896 --> 00:12:49.438
"Nice to meet you," and if I generate it

178
00:12:50.286 --> 00:12:53.281
Then I get the generated tones saying

179
00:12:54.241 --> 00:12:57.298
"Hello, I'm a student. Nice to meet you"

180
00:12:57.322 --> 00:12:59.108
When you listen to it

181
00:12:59.279 --> 00:13:03.569
You'll understand the tone of this character and cans use it later for

182
00:13:04.206 --> 00:13:07.236
When you need a character with this kind of tone

183
00:13:08.154 --> 00:13:12.948
Write the script in that person's tone, and use it for voice-over

184
00:13:12.972 --> 00:13:15.472
Where you can then incorporate it into your work

185
00:13:15.891 --> 00:13:19.100
Working like this should suit your needs

186
00:13:19.540 --> 00:13:24.092
If you want to translate

187
00:13:24.116 --> 00:13:25.835
You can use the Generative Voice AI tool

188
00:13:26.014 --> 00:13:29.030
From ElevenLabs, like we tried earlier

189
00:13:29.202 --> 00:13:32.107
It just automatically translates here

190
00:13:32.266 --> 00:13:36.722
If I write the script in Korean

191
00:13:36.973 --> 00:13:38.046
Shall I give it a try first?

192
00:13:48.699 --> 00:13:50.598
If I do it like this

193
00:13:53.300 --> 00:13:56.436
"Hello, I'm a student. Nice to meet you"

194
00:13:56.461 --> 00:13:57.429
Yes, like this

195
00:13:57.662 --> 00:14:02.375
You need to sign up here

196
00:14:02.399 --> 00:14:04.054
To enter

197
00:14:04.078 --> 00:14:06.884
The AI tool automatically translates

198
00:14:09.243 --> 00:14:14.388
The script you've created

199
00:14:14.574 --> 00:14:18.422
So, if I go into the tool

200
00:14:21.761 --> 00:14:26.426
I can skip and go to 'Voice Setting'

201
00:14:26.451 --> 00:14:29.585
Where I can choose the language, right?

202
00:14:30.325 --> 00:14:31.673
And then the person

203
00:14:31.893 --> 00:14:35.673
But here, you can see the default tone

204
00:14:35.890 --> 00:14:38.399
And other characteristics of this character

205
00:14:38.750 --> 00:14:41.872
like hashtags, right?

206
00:14:42.559 --> 00:14:47.039
Then you can also add your voice here

207
00:14:48.455 --> 00:14:54.035
And automatically

208
00:14:54.187 --> 00:14:56.511
if you've recorded your voice

209
00:14:56.535 --> 00:14:59.448
It takes your voice from here

210
00:14:59.472 --> 00:15:02.352
And automatically translates into other languages

211
00:15:02.986 --> 00:15:07.986
Thus, if you try using this tool at home

212
00:15:08.698 --> 00:15:10.631
You can get

213
00:15:12.151 --> 00:15:14.115
Some really interesting results

214
00:15:14.495 --> 00:15:17.257
Since I don't have the voiced file

215
00:15:17.282 --> 00:15:20.263
I hope you all give it a try yourselves

216
00:15:20.886 --> 00:15:22.888
Next is Heygen

217
00:15:22.912 --> 00:15:28.526
Heygen is a tool that's been used lately

218
00:15:28.551 --> 00:15:32.690
For things like news or for announcers

219
00:15:32.714 --> 00:15:36.938
I've heard stories about

220
00:15:37.524 --> 00:15:41.379
How there's a shortage of announcers in Jeju Island

221
00:15:41.403 --> 00:15:42.816
So they're using AI to broadcast the news

222
00:15:43.037 --> 00:15:47.532
Like such situation

223
00:15:47.776 --> 00:15:52.518
Heygen is an AI tool that uses AI to act as announcers

224
00:15:53.229 --> 00:15:57.943
Or to conduct question-and-answer sessions

225
00:15:58.644 --> 00:16:00.832
Which is becoming increasingly popular for these purposes

226
00:16:00.857 --> 00:16:03.271
Therefore, if you take a look at the demo here

227
00:16:08.931 --> 00:16:11.088
You'll notice how natural it sounds

228
00:16:11.273 --> 00:16:15.465
You can think of this as all being AI-generated content

229
00:16:15.490 --> 00:16:17.534
Yet it sounds quite natural

230
00:16:17.558 --> 00:16:21.147
Of course, the fundamentally AI-driven characters within it

231
00:16:21.459 --> 00:16:26.221
or this man, based on real people, likely had photos taken

232
00:16:26.245 --> 00:16:29.419
Collected data on things like closing eyes

233
00:16:29.443 --> 00:16:31.829
or opening his mouth

234
00:16:31.853 --> 00:16:34.996
So, this character was likely created using data collected from real people

235
00:16:35.289 --> 00:16:37.788
When you see the generated video

236
00:16:38.130 --> 00:16:40.103
It looks like it was shot in real-time, doesn't it?

237
00:16:40.657 --> 00:16:43.492
So, once you get in

238
00:16:43.516 --> 00:16:45.964
You can create an instant avatar

239
00:16:45.988 --> 00:16:48.864
A photo avatar or a studio avatar

240
00:16:49.070 --> 00:16:52.502
But with photo avatars, we can create them ourselves

241
00:16:52.656 --> 00:16:55.138
To create instant avatars like the ones we saw in the demo

242
00:16:55.162 --> 00:16:57.868
You typically need to subscribe

243
00:16:57.988 --> 00:17:02.067
Which can be a drawback

244
00:17:02.091 --> 00:17:04.720
So, for scenarios like

245
00:17:05.698 --> 00:17:10.491
Creating realistic avatars for YouTube

246
00:17:10.515 --> 00:17:12.398
or studio shoots, it could be quite useful and worth considering

247
00:17:12.422 --> 00:17:15.623
Especially if you plan to use it frequently

248
00:17:16.338 --> 00:17:19.744
When it comes to animation production

249
00:17:19.769 --> 00:17:22.758
It's questionable whether such tools would be as necessary

250
00:17:23.911 --> 00:17:28.403
There are various tools available

251
00:17:28.681 --> 00:17:31.371
For generating music as well

252
00:17:31.638 --> 00:17:34.363
When it comes to tools for generating music

253
00:17:35.348 --> 00:17:37.282
There are tools for generating music

254
00:17:37.306 --> 00:17:40.498
As well as tools where you can download

255
00:17:40.522 --> 00:17:43.035
royalty-free music

256
00:17:43.320 --> 00:17:46.343
But now there are many subscription-based services

257
00:17:46.367 --> 00:17:49.813
That offer higher-quality music generation

258
00:17:50.305 --> 00:17:54.090
When I enter the website

259
00:17:54.669 --> 00:17:56.628
Called "Soundful"

260
00:17:57.401 --> 00:18:02.475
You can choose various genres

261
00:18:02.913 --> 00:18:06.912
and click on 'Create,' you'll need to mix here

262
00:18:06.936 --> 00:18:09.632
Once you select another option

263
00:18:09.656 --> 00:18:12.362
and create

264
00:18:15.634 --> 00:18:17.757
You can also determine

265
00:18:18.402 --> 00:18:21.963
The speed of the sound here

266
00:18:22.197 --> 00:18:26.816
You can also choose whether the key is in C or D code

267
00:18:27.570 --> 00:18:32.812
In this way, you can generate ambient music

268
00:18:32.836 --> 00:18:35.135
or background music needed for videos

269
00:18:35.273 --> 00:18:40.263
In SoundDraw, for example

270
00:18:41.345 --> 00:18:43.429
You can choose genres like hip-hop

271
00:18:44.793 --> 00:18:46.609
and create music

272
00:18:47.054 --> 00:18:51.540
Generally, the prompts are predefined in this way

273
00:18:51.564 --> 00:18:54.905
You can change the prompts here

274
00:18:54.929 --> 00:18:59.490
For instance, you can change the mood to something happy

275
00:19:00.265 --> 00:19:03.867
or set the theme to cinematic and the length to 10 seconds

276
00:19:04.832 --> 00:19:16.071
or set the tempo to normal and choose to use all instruments

277
00:19:23.862 --> 00:19:28.392
In this way, you can create music that you have composed

278
00:19:28.938 --> 00:19:31.461
Of course, as I mentioned earlier

279
00:19:31.485 --> 00:19:34.358
It might not be of the highest quality

280
00:19:34.625 --> 00:19:38.857
But you will be able to use it well for instant purposes

281
00:19:39.545 --> 00:19:44.559
Additionally, you can use royalty-free music, which means music that is free from copyright restrictions

282
00:19:44.793 --> 00:19:50.757
There are numerous websites where you can download royalty-free music

283
00:19:51.791 --> 00:19:54.715
Later, when you work on post-production after creating animation

284
00:19:55.310 --> 00:19:58.882
Using these resources can be incredibly helpful

285
00:19:59.362 --> 00:20:03.515
I've also directed about

286
00:20:03.539 --> 00:20:07.654
Five short animations

287
00:20:08.557 --> 00:20:11.740
I've found that after creating five short animations

288
00:20:12.234 --> 00:20:17.971
No matter how well the video is made, if the music is subpar

289
00:20:18.379 --> 00:20:20.235
The quality of the animation drop significantly

290
00:20:20.260 --> 00:20:23.100
Therefore, I always tell my students that

291
00:20:24.192 --> 00:20:26.407
Music is half of the work

292
00:20:26.432 --> 00:20:29.135
So, music is just as important

293
00:20:29.888 --> 00:20:32.623
Choosing the right music to complement your project is crucial

294
00:20:32.647 --> 00:20:36.292
Thus

295
00:20:37.252 --> 00:20:39.593
Even if you don't find the exact music you want here

296
00:20:40.112 --> 00:20:42.700
You can also use these guidelines

297
00:20:43.053 --> 00:20:45.728
To work with an actual music composer

298
00:20:46.155 --> 00:20:49.704
To create the perfect track for your project

299
00:20:51.396 --> 00:20:54.484
My student is also doing that right now

300
00:20:54.509 --> 00:20:57.323
That approach has been yielding excellent results

301
00:20:58.249 --> 00:21:00.583
When I produced animations

302
00:21:01.335 --> 00:21:03.623
I usually recorded my vocalizations as guidance for the music composer

303
00:21:04.076 --> 00:21:08.331
Providing them with an idea of what I'm aiming for

304
00:21:08.355 --> 00:21:11.153
Whereas, with the advancement of AI nowadays

305
00:21:11.478 --> 00:21:13.772
It seems that these processes have become much more convenient

306
00:21:14.447 --> 00:21:18.194
AI Tools for Animation Production Stages

307
00:21:18.260 --> 00:21:22.501
Next up is

308
00:21:22.525 --> 00:21:25.746
When making a movie, for example

309
00:21:25.770 --> 00:21:30.008
or when I want to overlay a 3D character on live-action background

310
00:21:30.179 --> 00:21:34.356
or when I've shot live-action footage, but there are unwanted elements like

311
00:21:34.746 --> 00:21:38.077
People walking in the background or unnecessary objects

312
00:21:38.556 --> 00:21:41.587
Tools that remove such elements are also available

313
00:21:41.611 --> 00:21:44.925
In the past, for instance

314
00:21:44.949 --> 00:21:48.665
In VFX movies where everything had to be removed frame by frame

315
00:21:48.836 --> 00:21:51.771
They would first connect the green screen

316
00:21:51.795 --> 00:21:54.250
and then they would remove the shots with people in them

317
00:21:54.522 --> 00:21:57.821
frame by frame

318
00:21:58.152 --> 00:22:01.150
Nowadays, AI has advanced to the point where

319
00:22:01.440 --> 00:22:03.114
You can simply erase these things with just one click of the mouse

320
00:22:03.138 --> 00:22:05.606
Even in videos

321
00:22:06.060 --> 00:22:09.130
Erasing things in photos isn't so difficult

322
00:22:09.442 --> 00:22:11.408
But it's quite challenging when it comes to videos

323
00:22:12.081 --> 00:22:16.312
But now, there are many tools available for such tasks, so

324
00:22:17.051 --> 00:22:19.849
If you get a chance to see a demo

325
00:22:20.701 --> 00:22:23.876
These days, even Photoshop has tools

326
00:22:23.900 --> 00:22:25.703
That allow you to erase backgrounds

327
00:22:25.727 --> 00:22:28.825
So you can remove unwanted elements from your images

328
00:22:28.849 --> 00:22:32.390
For example, in Photoshop

329
00:22:32.414 --> 00:22:34.728
You can right-click on the subject

330
00:22:35.165 --> 00:22:40.629
and use the prompt like 'remove'

331
00:22:40.655 --> 00:22:43.797
Then the subjects over here will be removed

332
00:22:43.822 --> 00:22:48.086
Right now, because I've selected a blurry background

333
00:22:48.452 --> 00:22:51.518
The selected objects are a bit unclear

334
00:22:51.824 --> 00:22:53.940
If the object is clear, then

335
00:22:54.045 --> 00:22:55.632
Should we try removing the face?

336
00:23:00.641 --> 00:23:02.607
Using 'remove face' resulted in

337
00:23:15.790 --> 00:23:18.496
A somewhat strange outcome

338
00:23:19.656 --> 00:23:23.427
Anyway, if that's what came up

339
00:23:23.883 --> 00:23:27.821
You can choose various options

340
00:23:27.845 --> 00:23:30.344
From the Properties panel next to it

341
00:23:33.667 --> 00:23:36.892
You can choose the options

342
00:23:37.304 --> 00:23:41.070
Then edit as you want from here

343
00:23:41.403 --> 00:23:45.987
So, by evaluating these results as

344
00:23:46.011 --> 00:23:49.181
Good or Poor

345
00:23:49.600 --> 00:23:53.432
After the AI learns from these evaluations

346
00:23:53.881 --> 00:23:57.229
When you generate something else

347
00:23:57.641 --> 00:24:00.773
It can produce results of higher quality

348
00:24:01.057 --> 00:24:03.542
And when you've made the video

349
00:24:03.566 --> 00:24:06.822
The most important thing in the end was

350
00:24:06.940 --> 00:24:09.531
If I said I made a video

351
00:24:09.555 --> 00:24:13.383
in HD resolution

352
00:24:13.407 --> 00:24:19.497
If I later said that I'm going to screen this video in a movie theater

353
00:24:19.521 --> 00:24:22.141
Then I would need to scale up the video's resolution

354
00:24:22.721 --> 00:24:26.957
When scaling up for the theater platform

355
00:24:27.202 --> 00:24:31.276
The resolution itself may degrade

356
00:24:31.695 --> 00:24:36.903
Many AI tools have emerged

357
00:24:36.927 --> 00:24:42.125
To upscale resolution without degrading it

358
00:24:42.645 --> 00:24:48.115
Tools like Topaz Labs and BigJPG are good examples of such AI-powered upscaling solutions

359
00:24:49.590 --> 00:24:52.909
and by in putting an image into these tools

360
00:24:53.627 --> 00:24:56.494
They perform upscaling while preserving the quality

361
00:24:56.518 --> 00:24:58.185
Making the final result look more vibrant and live

362
00:24:58.975 --> 00:25:00.825
There's also Runway

363
00:25:01.313 --> 00:25:04.099
Additionally, LeiaPix is a tool

364
00:25:04.205 --> 00:25:07.366
That specializes in

365
00:25:08.339 --> 00:25:12.011
Creating quick animations with efficiency

366
00:25:12.698 --> 00:25:15.811
Well, the tools I've been introducing are

367
00:25:15.835 --> 00:25:20.676
Primarily focused on

368
00:25:20.700 --> 00:25:24.705
Those commonly used AI animation production

369
00:25:24.729 --> 00:25:28.628
and there are many other tools available as well

370
00:25:29.155 --> 00:25:33.560
There are many tools avilable

371
00:25:33.585 --> 00:25:35.218
But I'm focusing on showing you the one

372
00:25:35.397 --> 00:25:38.610
That are still widely used

373
00:25:38.736 --> 00:25:41.054
As many have been phased out

374
00:25:41.079 --> 00:25:46.040
Like this, when you run LeiaPix

375
00:25:46.064 --> 00:25:49.282
After automatically calculating the depth values

376
00:25:49.800 --> 00:25:54.438
It separates the background and me

377
00:25:54.462 --> 00:25:55.748
Allowing for a tilt effect

378
00:25:55.772 --> 00:26:00.025
As if the camera were moving like this

379
00:26:00.544 --> 00:26:01.593
and it pans like this

380
00:26:02.925 --> 00:26:06.156
As you can see

381
00:26:06.180 --> 00:26:08.960
Hair is quite delicate

382
00:26:09.259 --> 00:26:14.171
So it's challenging for computers to accurately capture the details of hair

383
00:26:14.195 --> 00:26:18.218
Therefore as the hair stretches, it pans from side to side

384
00:26:19.257 --> 00:26:21.760
There are some quality issues

385
00:26:22.716 --> 00:26:25.046
But the zoom-in function works really well, doesn't it?

386
00:26:25.139 --> 00:26:26.092
Zoom in, zoom out

387
00:26:26.572 --> 00:26:30.121
You can also increase the amount of motion for better interaction

388
00:26:30.251 --> 00:26:33.622
or you can use it in various ways

389
00:26:35.034 --> 00:26:37.957
The AI tools I've introduced so far can be divided into

390
00:26:37.981 --> 00:26:40.859
Cloud-based tools and local tools

391
00:26:41.666 --> 00:26:46.742
In simple terms, cloud-based tools

392
00:26:47.377 --> 00:26:52.576
Use the cloud network of the company that created the AI tools

393
00:26:52.600 --> 00:26:55.612
Local tools, on the other hand, use own computer for their operations

394
00:26:55.906 --> 00:26:58.289
So, both have their pros and cons

395
00:26:58.313 --> 00:27:01.256
If the computer has high specifications

396
00:27:01.448 --> 00:27:04.742
Using local tools is feasible without any issues

397
00:27:04.778 --> 00:27:06.914
However, using local tools

398
00:27:06.938 --> 00:27:10.314
Consumes the computer's graphic card memory

399
00:27:10.646 --> 00:27:15.144
Which can slow down other tasks

400
00:27:15.168 --> 00:27:20.267
Therefore, the best option is often

401
00:27:20.291 --> 00:27:22.251
To use their cloud services

402
00:27:22.629 --> 00:27:27.154
If the number of users utilizing the cloud service increases significantly

403
00:27:27.559 --> 00:27:29.960
Without a properly built pipeline

404
00:27:30.099 --> 00:27:32.447
Even that could become slow

405
00:27:32.471 --> 00:27:35.571
Although there are pros and cons

406
00:27:36.658 --> 00:27:41.946
It's becoming increasingly common for cloud-based tools to dominate as time goes on

407
00:27:42.950 --> 00:27:45.943
If I were to create an animation

408
00:27:45.968 --> 00:27:50.657
Using the tools I've shown you so far

409
00:27:50.681 --> 00:27:55.471
I've briefly summarized some tools

410
00:27:55.643 --> 00:27:57.683
That I recommended for each process

411
00:27:57.982 --> 00:28:00.989
For ideas and scripts, of course

412
00:28:01.139 --> 00:28:06.000
Your own thoughts are the foundation

413
00:28:06.025 --> 00:28:09.166
Next, the most basic tool that would probably be used is

414
00:28:09.190 --> 00:28:13.098
ChatGPT

415
00:28:13.318 --> 00:28:17.760
And if you were to create an audio version, as I mentioned earlier

416
00:28:18.442 --> 00:28:23.500
You would obviously use Premiere

417
00:28:23.524 --> 00:28:25.749
You would use Premiere for editing

418
00:28:26.236 --> 00:28:32.756
You would use ElevenLabs and Stock Library

419
00:28:32.780 --> 00:28:37.192
For voiceovers and other audio elements

420
00:28:37.216 --> 00:28:40.959
For music, you'd use Artlist and

421
00:28:40.983 --> 00:28:44.103
Perhaps also Pixabay for additional resources

422
00:28:44.276 --> 00:28:46.663
And when it comes to creating shots

423
00:28:46.687 --> 00:28:48.943
or shot lists

424
00:28:49.275 --> 00:28:54.190
Similarly, as you've seen in the examples

425
00:28:54.214 --> 00:28:57.266
You would write prompts

426
00:28:57.459 --> 00:28:59.394
Requesting specific scenes or shot compositions

427
00:28:59.578 --> 00:29:03.384
That's why tools like ChatGPT would be

428
00:29:03.408 --> 00:29:04.850
Heavily utilized for this purpose as well

429
00:29:05.119 --> 00:29:09.127
And even in platforms like Notion

430
00:29:09.151 --> 00:29:13.652
Which are widely used in companies these days

431
00:29:13.677 --> 00:29:18.590
For explaining internal policies or pipelines

432
00:29:19.452 --> 00:29:23.452
ChatGPT is also integrated

433
00:29:23.476 --> 00:29:27.038
Therefore, using ChatGPT within Notion

434
00:29:27.062 --> 00:29:30.115
Would likely yield similar results

435
00:29:30.888 --> 00:29:33.241
And for image generation, Midjourney would be used

436
00:29:33.441 --> 00:29:37.876
Image generation would be best done using Midjourney

437
00:29:38.290 --> 00:29:42.461
For editing images, you could use Midjoruney

438
00:29:42.485 --> 00:29:44.563
or Generative Fill

439
00:29:44.742 --> 00:29:47.246
For image editing, Midjourney would likely

440
00:29:48.095 --> 00:29:49.315
Be the most commonly used tool

441
00:29:49.340 --> 00:29:53.242
And for combining individual slideshows

442
00:29:53.266 --> 00:29:57.444
For assembling slideshows for your film

443
00:29:58.332 --> 00:30:01.128
You could use Midjourney

444
00:30:01.152 --> 00:30:04.502
For extracting and editing images

445
00:30:05.669 --> 00:30:09.282
Similarly, you could use a video editing tool

446
00:30:09.988 --> 00:30:12.310
To combine those images

447
00:30:13.047 --> 00:30:16.525
Processed with Midjourney

448
00:30:17.621 --> 00:30:21.203
And for animating images, you could use Runway

449
00:30:21.227 --> 00:30:25.539
And you could continue to mix and use Pika Labs

450
00:30:26.407 --> 00:30:27.693
 and Stable Diffusion as needed

451
00:30:27.819 --> 00:30:31.036
And for upscaling, as mentioned earlier

452
00:30:31.061 --> 00:30:35.020
You could use tools like Topaz

453
00:30:35.044 --> 00:30:38.261
To upscale while maintaining resolution

454
00:30:38.852 --> 00:30:42.459
For effects and compositioning

455
00:30:42.850 --> 00:30:44.761
Popular choices include After Effects

456
00:30:44.786 --> 00:30:48.074
Runway, and Pika Labs

457
00:30:48.444 --> 00:30:52.854
For editing, you could continue using the software you're already familiar with

458
00:30:52.878 --> 00:30:56.691
Focusing on Premiere, or for those using Mac

459
00:30:57.853 --> 00:31:01.707
They could use other editing tools specific to Mac

460
00:31:02.265 --> 00:31:05.979
Later, for releasing your work

461
00:31:06.237 --> 00:31:08.796
You could render

462
00:31:09.328 --> 00:31:11.274
and publish it

463
00:31:11.781 --> 00:31:14.263
On commonly used platforms

464
00:31:15.653 --> 00:31:17.426
Like YouTube or Vimeo

465
00:31:17.809 --> 00:31:21.886
With this, we've covered ideas and sound

466
00:31:22.040 --> 00:31:25.717
For animation production using AI

467
00:31:25.741 --> 00:31:26.392
Thank you

468
00:31:27.733 --> 00:31:29.124
1. AI tools for Animation Ideas, Sound, and Music Utilization
AI tools generally produce results at a standard level, so they should be used as guidelines to find direction

469
00:31:29.124 --> 00:31:30.473
Since music plays a significant role in animation, it's worth exploring various uses of AI tools in this aspect
Types of AI Tools 
Ideas and Scriptwriting: ChatGPT, Bard (Gemini)

470
00:31:30.473 --> 00:31:31.450
Voice Over / Dubbing: Generative Voice AI, Lovo
Translation: Heygen

471
00:31:31.450 --> 00:31:32.118
Music Creation: Soundful, SoundDraw
Video Background Removal: Photoshop
Upscaling Images: Topaz Labs, BigJPG

472
00:31:32.118 --> 00:31:32.654
Quick Animation: Leiapix

473
00:31:32.654 --> 00:31:33.369
AI Tools for Different Stages of Animation Production
1. Idea: ChatGPT
2. Script: ChatGPT
3. Create the Audio Version: ElevenLabs for Voice Overs,
Stock Library for Sound Effects

474
00:31:33.369 --> 00:31:34.134
4. Music from Stock Library: Artlist Pixabay
5. Create a Shot List: ChatGPT, Notion Table

475
00:31:34.134 --> 00:31:34.919
6. Pull Images: Midjourney

476
00:31:34.919 --> 00:31:35.687
7. Edit the Images: Midjourney, Generative Fill
8. Put Together a 'Slide Show' Version of Your Film: Midjourney

477
00:31:35.687 --> 00:31:36.437
9. Animate the Images: Runway, Pika Labs, Stable Video Diffusion

478
00:31:36.437 --> 00:31:37.037
10. Upscaling & Polish Your Footage: Topaz 
11. VFX and Composition: After Effects, Runway, Pika Labs

479
00:31:37.037 --> 00:31:37.624
12. Find Tune the Edit: Using Familiar Tools, Premiere, Mac, etc. 
13. Distribute: YouTube, Vimeo