All language subtitles for 5. ControlNet SD1.5 - Setup and Scribble, Lineart and Openpose

af Afrikaans
sq Albanian
am Amharic
ar Arabic
hy Armenian
az Azerbaijani
eu Basque
be Belarusian
bn Bengali
bs Bosnian
bg Bulgarian
ca Catalan
ceb Cebuano
ny Chichewa
zh-CN Chinese (Simplified)
zh-TW Chinese (Traditional)
co Corsican
hr Croatian
cs Czech
da Danish
nl Dutch
en English
eo Esperanto
et Estonian
tl Filipino
fi Finnish
fr French
fy Frisian
gl Galician
ka Georgian
de German
el Greek
gu Gujarati
ht Haitian Creole
ha Hausa
haw Hawaiian
iw Hebrew
hi Hindi
hmn Hmong
hu Hungarian
is Icelandic
ig Igbo
id Indonesian
ga Irish
it Italian
ja Japanese
jw Javanese
kn Kannada
kk Kazakh
km Khmer
ko Korean
ku Kurdish (Kurmanji)
ky Kyrgyz
lo Lao
la Latin
lv Latvian
lt Lithuanian
lb Luxembourgish
mk Macedonian
mg Malagasy
ms Malay
ml Malayalam
mt Maltese
mi Maori
mr Marathi
mn Mongolian
my Myanmar (Burmese)
ne Nepali
no Norwegian
ps Pashto
fa Persian
pl Polish
pt Portuguese
pa Punjabi
ro Romanian
ru Russian
sm Samoan
gd Scots Gaelic
sr Serbian
st Sesotho
sn Shona
sd Sindhi
si Sinhala
sk Slovak
sl Slovenian
so Somali
es Spanish
su Sundanese
sw Swahili
sv Swedish
tg Tajik
ta Tamil
te Telugu
th Thai
tr Turkish
uk Ukrainian
ur Urdu
uz Uzbek
vi Vietnamese Download
cy Welsh
xh Xhosa
yi Yiddish
yo Yoruba
zu Zulu
or Odia (Oriya)
rw Kinyarwanda
tk Turkmen
tt Tatar
ug Uyghur
Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated: 1 00:00:00,256 --> 00:00:01,280 So we'll take a look at the 2 00:00:01,536 --> 00:00:04,096 First and it's kind of a fun control net this is 3 00:00:04,352 --> 00:00:05,376 Scribble control 4 00:00:06,144 --> 00:00:09,216 Just see what's happening here we've got a number of outputs 5 00:00:09,728 --> 00:00:12,800 And it's taking a scribble that I put together on canva 6 00:00:13,312 --> 00:00:15,616 And it's turned it into these really interesting outcomes 7 00:00:17,152 --> 00:00:17,920 Let's take a look 8 00:00:18,176 --> 00:00:19,968 Look at how everything set up so 9 00:00:20,224 --> 00:00:20,992 We're dealing with 10 00:00:21,760 --> 00:00:23,296 SD 1.5 so 11 00:00:23,808 --> 00:00:25,856 We've got a 512 by 512 I'm 12 00:00:26,368 --> 00:00:29,696 And this is giving us this outcome are in the saved images 13 00:00:30,208 --> 00:00:33,280 The control the load control is loading up 14 00:00:33,536 --> 00:00:34,304 Control 15 00:00:35,328 --> 00:00:41,472 311 PSD 15 scribble very nice name and I might sometimes to keep the original names because 16 00:00:41,728 --> 00:00:43,520 Because it is much easier to 17 00:00:44,032 --> 00:00:46,848 Set of know exactly which file I downloaded 18 00:00:47,104 --> 00:00:51,200 Rather than changing them to a more friendly name with sometimes 19 00:00:51,712 --> 00:00:55,040 Sometimes makes sense sometimes it's not that helpful in the long run 20 00:00:57,088 --> 00:01:01,184 Checkpoint I'm loading is dream shaper aids for a change 21 00:01:01,952 --> 00:01:03,744 And we've got the standard 22 00:01:04,512 --> 00:01:05,536 Standard VA 23 00:01:07,072 --> 00:01:09,888 Light steel statues Starry Sky 24 00:01:10,144 --> 00:01:11,680 Beautiful desert landscape 25 00:01:12,192 --> 00:01:16,288 And we go back to the history we can see what the previous ones were 26 00:01:19,872 --> 00:01:22,688 2 Kings walking under a starry sigh 27 00:01:27,040 --> 00:01:28,576 So this is the one with 28 00:01:28,832 --> 00:01:29,856 Let's go back to 29 00:01:31,648 --> 00:01:34,464 I'll have to redo that 30 00:01:36,512 --> 00:01:38,048 With that bit of a reminder to 31 00:01:38,304 --> 00:01:39,328 Save the other 32 00:01:39,584 --> 00:01:40,608 What's the workflows before 33 00:01:40,864 --> 00:01:42,144 Playing around in the history 34 00:01:42,400 --> 00:01:45,728 So the workflow here which I will share with you 35 00:01:45,984 --> 00:01:50,848 Basically runs from the control net which is Scribblenaut goes to dream shaper 36 00:01:51,360 --> 00:01:53,408 And then we're using the standard 37 00:01:54,688 --> 00:01:55,456 VA 38 00:01:55,712 --> 00:01:58,272 Two men likes steel statues Starry Sky 39 00:01:58,784 --> 00:02:00,576 Beautiful desert landscape 40 00:02:02,880 --> 00:02:05,952 What is what it comes up with using this particular scribble 41 00:02:06,976 --> 00:02:08,256 Did this particular 42 00:02:09,280 --> 00:02:11,072 Control net is actually 43 00:02:11,328 --> 00:02:12,608 Play pretty flexible 44 00:02:13,120 --> 00:02:14,656 And it does try to follow 45 00:02:15,680 --> 00:02:17,472 Play the shapes that it sees but 46 00:02:17,728 --> 00:02:19,008 Sometimes it interprets things 47 00:02:19,520 --> 00:02:25,152 Some mathematically so it is sort of put some Stars which didn't necessarily look like these giant Stars 48 00:02:25,664 --> 00:02:27,200 So it's a very flexible one 49 00:02:27,456 --> 00:02:33,600 Which I think pays to experiment with quite a bit to see how it works because it can produce some amazing results 50 00:02:34,112 --> 00:02:35,648 From very very simple diagram 51 00:02:36,160 --> 00:02:36,928 I control the 52 00:02:37,184 --> 00:02:39,232 Play Scrabble yourself you can get them online 53 00:02:39,744 --> 00:02:43,072 And working with a strength of 0.9 54 00:02:43,584 --> 00:02:48,192 That we've seen I think in previous projects 55 00:02:48,704 --> 00:02:51,520 What's a little bit different is that we are starting with 56 00:02:52,032 --> 00:02:55,104 The image that would normally be handled by the preprocessor 57 00:02:55,360 --> 00:02:59,968 So we just importing something that's already ready for the controller 58 00:03:02,272 --> 00:03:03,808 To the controller I think the 59 00:03:04,320 --> 00:03:06,368 The link you can see the link that 60 00:03:06,624 --> 00:03:10,208 So that's basically how it works I'll share this with you and you should be able to use this 61 00:03:10,720 --> 00:03:12,000 For your scribbles and 62 00:03:12,256 --> 00:03:13,792 It might work for quite a few other 63 00:03:14,048 --> 00:03:14,816 Control Nest as 64 00:03:17,376 --> 00:03:19,168 Now the line 65 00:03:19,680 --> 00:03:23,008 Control net this one is sd-15 line 66 00:03:23,520 --> 00:03:25,056 This one actually comes in 67 00:03:26,080 --> 00:03:27,616 With a similar kind of 68 00:03:29,408 --> 00:03:29,920 Play 69 00:03:30,176 --> 00:03:33,248 As the as the other one describe the one that we were dealing with 70 00:03:35,808 --> 00:03:38,112 I've taken a hand 71 00:03:38,368 --> 00:03:39,136 Lion 72 00:03:40,416 --> 00:03:41,440 Use it to create 73 00:03:41,696 --> 00:03:43,744 Delete a set of roads in the 74 00:03:44,000 --> 00:03:44,768 In the country 75 00:03:45,024 --> 00:03:46,816 Anderson for tropical area 76 00:03:51,936 --> 00:03:52,960 Road Network 77 00:03:53,216 --> 00:03:54,496 That seems to resemble 78 00:03:54,752 --> 00:03:56,032 What's happening within the hand 79 00:03:56,288 --> 00:04:00,128 It may be a higher number of steps 80 00:04:00,640 --> 00:04:02,944 And the control net here 81 00:04:03,200 --> 00:04:03,712 Is 82 00:04:05,248 --> 00:04:06,272 Working with 83 00:04:06,528 --> 00:04:07,040 Play 84 00:04:08,320 --> 00:04:09,856 A house in the first scene 85 00:04:10,112 --> 00:04:11,648 Tropical scene from above 86 00:04:11,904 --> 00:04:14,464 Trees Road hand like outline of the building 87 00:04:14,720 --> 00:04:17,791 So instead of taking the building and making it hand like 88 00:04:18,303 --> 00:04:19,839 It's instead made the 89 00:04:20,351 --> 00:04:24,959 Rodeway the main theme of the of the design so we've got a roadway 90 00:04:25,471 --> 00:04:29,567 Because this image is 91 00:04:30,591 --> 00:04:31,615 Where is the 92 00:04:33,151 --> 00:04:37,759 The Mt Layton damage is 512 so we could increase this to 768 93 00:04:40,319 --> 00:04:42,367 Think it allows us to work at that level 94 00:04:42,623 --> 00:04:44,159 Without too much distortion 95 00:04:46,207 --> 00:04:48,767 What happens and you can see that it is extremely quick 96 00:04:50,303 --> 00:04:51,839 When we're working at 97 00:04:52,095 --> 00:04:53,631 What's SD 1.5 98 00:04:53,887 --> 00:04:54,911 And here we have 99 00:04:55,423 --> 00:05:00,543 Something which may be resembles the prompt a bit more you've got a really unusual building 100 00:05:00,799 --> 00:05:02,847 Which takes the outline of the hand 101 00:05:03,103 --> 00:05:06,431 So this is a really powerful one and it's one that I think 102 00:05:06,943 --> 00:05:08,479 Can be quite inventive 103 00:05:10,271 --> 00:05:13,087 So we can take a look at the ones that I've created so far this is the 104 00:05:17,951 --> 00:05:19,743 Shape of that of that hand 105 00:05:20,255 --> 00:05:22,559 It's really kind of impressive how it works 106 00:05:23,071 --> 00:05:27,167 And you can see the previous one where is combined house and 107 00:05:28,447 --> 00:05:28,959 And then 108 00:05:29,215 --> 00:05:34,335 With the men with the two men two Stickman you can see it's a little bit more 109 00:05:35,103 --> 00:05:36,383 It's a little bit more 110 00:05:37,151 --> 00:05:38,175 I would say 111 00:05:39,455 --> 00:05:41,503 It's a little bit less creative in some ways 112 00:05:41,759 --> 00:05:43,039 Then these crippled one 113 00:05:44,319 --> 00:05:46,879 Follows the outlines a little bit more 114 00:05:47,135 --> 00:05:47,903 Authentically 115 00:05:48,159 --> 00:05:49,183 Play bendy the scrib 116 00:05:51,231 --> 00:05:53,535 So there's that minor difference but otherwise 117 00:05:53,791 --> 00:05:56,095 They do seem to carry out a similar 118 00:05:56,351 --> 00:05:57,119 Kind of function 119 00:05:57,631 --> 00:05:58,655 This one here 120 00:05:58,911 --> 00:06:01,215 Literally the outline that we were looking at 121 00:06:01,727 --> 00:06:03,519 With the two gentlemen 122 00:06:04,543 --> 00:06:09,151 And you can see it's a very different idea to the to the previous one the scribble one 123 00:06:09,663 --> 00:06:12,991 So use those two in similar but different situations 124 00:06:15,807 --> 00:06:17,087 So this one 125 00:06:17,343 --> 00:06:18,623 Which comes from comfy himself 126 00:06:19,903 --> 00:06:21,183 Little bit more complicated 127 00:06:21,951 --> 00:06:28,095 And I will give it to you as is it's a complicated one that produces two outcomes which you can easily do 128 00:06:28,351 --> 00:06:29,631 Do with the company UI 129 00:06:29,887 --> 00:06:31,679 And it starts off with the pose 130 00:06:31,935 --> 00:06:36,031 Estimation is that uses a technique known as pose estimation 131 00:06:38,591 --> 00:06:44,735 Unpause destination is a neural network technique that allows us to take a simple pose 132 00:06:46,271 --> 00:06:49,599 Well it can take an actual subject and then interpret what 133 00:06:49,855 --> 00:06:51,135 Posse must have been 134 00:06:51,391 --> 00:06:52,927 Turning it into lines like this 135 00:06:55,743 --> 00:06:56,511 Make a 136 00:06:56,767 --> 00:06:58,815 And then turn it into an image so in this case 137 00:06:59,071 --> 00:07:00,095 This particular image 138 00:07:00,607 --> 00:07:02,911 It's turned into this worshipful 139 00:07:03,167 --> 00:07:06,751 Image it's not 100% successful I have to say so it's a little bit 140 00:07:07,263 --> 00:07:09,823 I'm more of a demonstration of what you can do 141 00:07:10,079 --> 00:07:11,103 With come for you 142 00:07:11,359 --> 00:07:14,687 But it is showing a gradient from 143 00:07:14,943 --> 00:07:15,967 The small 144 00:07:16,479 --> 00:07:21,087 Input through to the largest output so there's an enlargement going on as well 145 00:07:21,343 --> 00:07:24,415 And that is a little bit of a breaking down of the the image 146 00:07:24,927 --> 00:07:30,047 As as you can see where you get will you begin to see parts of the image 147 00:07:30,559 --> 00:07:31,583 Duplicating 148 00:07:31,839 --> 00:07:36,703 That kind of collapse does occur as you work with larger size images but it's more of 149 00:07:37,727 --> 00:07:40,031 Representation of how you can do 150 00:07:40,287 --> 00:07:42,335 Something that starts with a fairly small image 151 00:07:42,847 --> 00:07:46,943 Basic neural net and neural network input 152 00:07:47,455 --> 00:07:48,735 And this one 153 00:07:49,503 --> 00:07:51,551 Goes into the open pose 154 00:07:51,807 --> 00:07:55,135 What's fp16 we're using an fp16 control 155 00:07:55,647 --> 00:07:56,415 Dream shaper 156 00:07:56,671 --> 00:08:00,511 And then we're using the standard 84 157 00:08:01,023 --> 00:08:02,559 840000 158 00:08:03,583 --> 00:08:04,095 Stop PR 159 00:08:04,351 --> 00:08:05,119 Vae 160 00:08:05,375 --> 00:08:08,191 And then some non-descript problems 161 00:08:08,703 --> 00:08:13,567 In this particular case we've got young adults expensive religious robes 162 00:08:15,871 --> 00:08:18,175 For whatever reason cute natural look sanding 163 00:08:18,431 --> 00:08:20,479 Sunrise beautiful landscape Bo 164 00:08:20,735 --> 00:08:24,575 Boca is a photography term that means that sort of background blur 165 00:08:24,831 --> 00:08:30,975 I was getting a very sharp background and when you when you in input Boca it just blows out the background 166 00:08:33,535 --> 00:08:35,327 Play the photographic term which 167 00:08:36,095 --> 00:08:39,679 The neural networks here in Statesville diffusion they understand 168 00:08:39,935 --> 00:08:41,983 That particular word to indicate 169 00:08:42,495 --> 00:08:43,775 Blood part of the image 170 00:08:44,287 --> 00:08:46,335 Play pretty well trained in in that 171 00:08:47,615 --> 00:08:52,991 And we've got City Lights I'm going to change that to Boca City Lights so we might get a different type of 172 00:08:53,247 --> 00:08:55,807 City Lights if we put in Boca 173 00:08:59,647 --> 00:09:03,231 Say the complex one beginning at 12:08 / 7:04 174 00:09:03,487 --> 00:09:05,023 Which is kind of like the point of 175 00:09:05,535 --> 00:09:06,815 Things Begin to Fall Apart 176 00:09:07,071 --> 00:09:10,143 With dream shaper and SD 1.5 177 00:09:10,911 --> 00:09:15,007 So this is maybe something that might work a little bit better inside of sdx 178 00:09:16,287 --> 00:09:17,055 We've got to stop 179 00:09:17,311 --> 00:09:18,079 Add CL 180 00:09:18,335 --> 00:09:19,871 Lay it too so we're going to 181 00:09:22,687 --> 00:09:27,551 And then we've got a couple of Samplers one for the initial run and the second one 182 00:09:28,063 --> 00:09:29,855 Dot actually run 183 00:09:30,367 --> 00:09:32,415 For the second enlargement run 184 00:09:36,511 --> 00:09:38,303 And this is the latest UPS 185 00:09:38,559 --> 00:09:39,839 So with the latest upscale 186 00:09:40,095 --> 00:09:42,911 You quite often have to run a fairly 187 00:09:44,447 --> 00:09:45,727 Have a family 188 00:09:46,751 --> 00:09:48,287 Simple upgrade 189 00:09:48,543 --> 00:09:50,591 To to the image you have to sort of run 190 00:09:50,847 --> 00:09:52,383 You can't just do 191 00:09:52,639 --> 00:09:55,199 Leave it sometimes you find you have to 192 00:09:55,455 --> 00:09:56,223 Is it to run 193 00:09:57,247 --> 00:09:58,015 Samples 194 00:09:58,271 --> 00:10:00,831 What's again in order to get the upscale to work properly 195 00:10:01,087 --> 00:10:03,903 But we can run it and you can see just how quick it is 196 00:10:04,159 --> 00:10:05,695 Inside of comfy you 197 00:10:08,255 --> 00:10:09,791 A single prompt 198 00:10:11,327 --> 00:10:14,655 And then we run the second prompt the enlargement Happening Here 199 00:10:15,167 --> 00:10:16,447 It's extremely quick 200 00:10:16,703 --> 00:10:19,519 I've got all sorts of equipment running in the background so 201 00:10:20,031 --> 00:10:22,591 Gives you an idea of just how efficient coffee UI is 202 00:10:23,359 --> 00:10:28,479 And we should soon have an image although the vae may stop for a while cuz I think my 203 00:10:28,991 --> 00:10:32,831 This is a very large image size and that can sometimes delay 204 00:10:34,879 --> 00:10:35,647 But no it's 205 00:10:37,183 --> 00:10:38,463 There we have our girls 206 00:10:38,719 --> 00:10:41,791 In the roads looking kind of serious and all sorts of worship 207 00:10:42,559 --> 00:10:43,583 Stop happening there 208 00:10:43,839 --> 00:10:46,911 So that is the I think I probably prefer the original one here 209 00:10:49,983 --> 00:10:52,031 It's a little bit technical 210 00:10:52,287 --> 00:10:56,127 Include all the stuff that you need so you'll find the pose image 15149

Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.