Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:00,230 --> 00:00:07,820
In this video, I will share the easiest way to create an AI influencer, and I want to make a picture
2
00:00:07,820 --> 00:00:10,850
that is similar to Fit Athena.
3
00:00:10,850 --> 00:00:18,320
So Eternal Opus is of course here our model and I want to make something that is similar.
4
00:00:18,320 --> 00:00:25,220
Yes, I can also download some of this and use them, but I will not do that.
5
00:00:25,220 --> 00:00:32,480
Instead, I work in a Google Colab notebook and this is totally for free and it's named focus.
6
00:00:32,630 --> 00:00:36,770
All we have to do is to play this button right here.
7
00:00:36,770 --> 00:00:40,550
And the code comes from GitHub, but that is totally fine.
8
00:00:40,550 --> 00:00:46,550
We will connect a graphics card from Google and of course this is totally free.
9
00:00:47,150 --> 00:00:51,380
The first thing we want to do is install git two.
10
00:00:51,410 --> 00:00:55,940
After the installation is complete we will clone our GitHub.
11
00:00:55,940 --> 00:00:58,070
So this right here is the GitHub.
12
00:00:58,070 --> 00:01:02,660
If I go in a new tab while this is loading you can see the GitHub link.
13
00:01:02,930 --> 00:01:04,850
It's basically this right here.
14
00:01:04,850 --> 00:01:08,030
And you can say thank you to the creator of this.
15
00:01:08,030 --> 00:01:11,000
We will make pictures that are similar to this.
16
00:01:11,000 --> 00:01:13,730
Or to be honest, even better pictures.
17
00:01:13,730 --> 00:01:19,880
And this is a really, really good and nice colab notebook because of course it's totally for free,
18
00:01:19,880 --> 00:01:25,160
but it makes stunning good pictures with stunning easy prompts.
19
00:01:25,310 --> 00:01:31,880
After this is cloned, we will create something small in Python, but you don't need to know anything.
20
00:01:31,880 --> 00:01:39,230
We are connected with T4 so the free graphics card from Google, and then we simply press the link as
21
00:01:39,230 --> 00:01:40,490
soon as we get it.
22
00:01:40,490 --> 00:01:42,050
And there we have it.
23
00:01:42,050 --> 00:01:45,830
So you'll see we have downloaded some stuff from Huggingface.
24
00:01:45,830 --> 00:01:48,200
So the models come from Hugging Face.
25
00:01:48,200 --> 00:01:51,290
This is all because our GitHub repo does this.
26
00:01:51,290 --> 00:01:53,600
And then we can do the following.
27
00:01:53,600 --> 00:01:59,420
Either we can run it locally, but I want to run this on the cloud.
28
00:01:59,420 --> 00:02:02,000
So run on public URL.
29
00:02:02,000 --> 00:02:07,040
If I press this right here we get a nice little interface.
30
00:02:07,040 --> 00:02:11,510
And in this interface we can create pictures totally for free.
31
00:02:11,510 --> 00:02:14,540
And the Google GPU will work for us.
32
00:02:14,540 --> 00:02:17,060
We can use advanced settings.
33
00:02:17,060 --> 00:02:23,330
If we activate this we get this box right here and here we can use some different stuff.
34
00:02:23,330 --> 00:02:25,220
So first we have settings.
35
00:02:25,220 --> 00:02:29,240
We can use speed quality or extreme speed.
36
00:02:29,240 --> 00:02:30,800
You can try this out.
37
00:02:30,800 --> 00:02:35,240
Right now I use speed just to make the generations a bit faster.
38
00:02:35,240 --> 00:02:43,190
But of course they get better if you use quality and worse if you use extreme speed then you can use
39
00:02:43,190 --> 00:02:44,450
aspect ratios.
40
00:02:44,450 --> 00:02:50,090
Right now I simply use one by one because I think that works totally fine for me.
41
00:02:50,090 --> 00:02:55,130
You can also use nine by 17 or something for Instagram that's better suited.
42
00:02:55,130 --> 00:02:57,530
Just use something that you like.
43
00:02:57,620 --> 00:03:02,750
The number of images I leave it at two and here is something that you know.
44
00:03:02,750 --> 00:03:07,880
So the negative prompt, this is of course the stuff where we put ugly, blurry and so on.
45
00:03:07,880 --> 00:03:14,660
But to be honest, we don't have to be that exact in this notebook because we have such a good model.
46
00:03:14,660 --> 00:03:22,550
Of course, if we combine really, really good models and good prompts, we get unbelievably good results.
47
00:03:23,030 --> 00:03:26,390
Ugly, blurry, out of frame.
48
00:03:26,390 --> 00:03:27,470
For example.
49
00:03:27,470 --> 00:03:29,180
You can be creative.
50
00:03:29,180 --> 00:03:31,130
This right here, this is the seed.
51
00:03:31,130 --> 00:03:33,800
So I think we can use a random seed.
52
00:03:33,800 --> 00:03:39,800
But of course you can also use a fixed seed if you want to be even more specific.
53
00:03:39,800 --> 00:03:42,170
I go with random right now.
54
00:03:42,200 --> 00:03:48,920
Then we have the style we can use really really a lot of different styles.
55
00:03:48,920 --> 00:03:52,940
And the styles that are used right now are the best styles.
56
00:03:52,940 --> 00:03:59,660
If we want to make really, really good looking pictures, maybe you can also try focus, photograph
57
00:03:59,660 --> 00:04:01,130
or focus masterpiece.
58
00:04:01,130 --> 00:04:03,620
But the first three works really fine.
59
00:04:04,340 --> 00:04:06,440
Then we have different models.
60
00:04:06,440 --> 00:04:13,550
Right now we work with the base model so stable diffusion XL and it's a juggernaut model and a really
61
00:04:13,550 --> 00:04:16,400
nice variation that makes enormous good pictures.
62
00:04:16,400 --> 00:04:19,100
And we have one, Laura included.
63
00:04:19,100 --> 00:04:25,340
And because we have one Laura, we make, of course, really, really good and consistent pictures,
64
00:04:25,340 --> 00:04:27,830
even if we use a different seed.
65
00:04:27,830 --> 00:04:30,380
And then we have the advanced settings.
66
00:04:30,380 --> 00:04:33,920
This right here is the guidance scale and the image sharpness.
67
00:04:33,920 --> 00:04:37,040
To be honest, just leave it how it is.
68
00:04:37,040 --> 00:04:43,310
The default settings work really, really fine, so everything we have to do right now is to give our
69
00:04:43,310 --> 00:04:44,540
prompt right here.
70
00:04:44,540 --> 00:04:48,800
And I will work with this Instagram model.
71
00:04:48,830 --> 00:04:54,230
Pink hair, beautiful and masterpiece.
72
00:04:54,230 --> 00:04:56,960
And I think that's basically it.
73
00:04:56,960 --> 00:04:58,610
And we press generate.
74
00:04:58,610 --> 00:04:59,840
And here you can see.
75
00:04:59,970 --> 00:05:05,580
We start to the views right now and I think we get a really, really nice picture.
76
00:05:05,610 --> 00:05:14,400
I also need to say, I think this model is also maybe trained on the pictures of, I think, Lopez,
77
00:05:14,400 --> 00:05:17,880
maybe it's even the Laura that's trained on I-10.
78
00:05:17,910 --> 00:05:24,030
I don't know exactly, but I just know that we have that we can create really, really good pictures
79
00:05:24,030 --> 00:05:28,830
with this model that look nearly exactly like this model.
80
00:05:28,830 --> 00:05:34,860
So if we press on it, you can see we have nearly the same pictures as a tenor.
81
00:05:35,100 --> 00:05:37,950
We have the brown eyes, the pink hair.
82
00:05:38,220 --> 00:05:41,190
She looks nearly the same, to be honest.
83
00:05:41,190 --> 00:05:47,430
And there we have our two pictures and I have to say, they are both really, really good.
84
00:05:47,430 --> 00:05:51,810
This picture, this is I think this is nearly perfect.
85
00:05:51,810 --> 00:05:56,730
So I want to download this picture because this will get important.
86
00:05:56,730 --> 00:06:00,570
We press on this button right now and then we can download it.
87
00:06:01,170 --> 00:06:03,390
And now comes the cool stuff.
88
00:06:03,390 --> 00:06:06,930
We have this right here input image.
89
00:06:06,930 --> 00:06:12,120
If I use input image I can use this picture right here as input.
90
00:06:12,120 --> 00:06:13,620
So I downloaded it.
91
00:06:13,620 --> 00:06:16,200
Now I will upload this picture.
92
00:06:16,200 --> 00:06:20,640
And now Stable Diffusion has this picture as a reference picture.
93
00:06:20,640 --> 00:06:28,920
And now I will do this right here I simply adjust the prompt just a little tiny bit Instagram model.
94
00:06:28,920 --> 00:06:29,610
Pink hair.
95
00:06:29,610 --> 00:06:30,690
Beautiful.
96
00:06:31,420 --> 00:06:39,100
Or we do this right here being here or Instagram model and the gym being here.
97
00:06:39,100 --> 00:06:40,270
Beautiful masterpiece.
98
00:06:40,270 --> 00:06:40,900
And that's it.
99
00:06:40,900 --> 00:06:47,860
So we basically changed just the environment and we will get really, really good pictures.
100
00:06:48,340 --> 00:06:49,930
And there we have it.
101
00:06:49,930 --> 00:06:53,890
These are two pictures that are nearly perfect in my mind.
102
00:06:53,890 --> 00:06:59,800
Maybe the eyes are not completely perfect, but I think they are really, really good.
103
00:06:59,800 --> 00:07:03,970
And of course I have used right here another resolution.
104
00:07:03,970 --> 00:07:13,450
You can also go one by one again and maybe say instead of in the gym, um, on the beach, or we go
105
00:07:13,450 --> 00:07:19,180
again with this resolution because I think for like Instagram Stories this would be perfect.
106
00:07:19,180 --> 00:07:22,780
So we generate the last one maybe on the beach.
107
00:07:22,780 --> 00:07:28,990
And if you find a seat that is really, really nice, just remember you can always use the seat.
108
00:07:29,230 --> 00:07:34,960
And now she is at the beach and also these two pictures are, in my mind at least, nearly perfect.
109
00:07:34,960 --> 00:07:37,720
Maybe the eyes are not that perfect.
110
00:07:37,720 --> 00:07:40,570
This picture right here, this is even better.
111
00:07:40,570 --> 00:07:44,770
And of course you can always do the face swaps and so on.
112
00:07:44,770 --> 00:07:48,430
And you can also always use a seat if you like.
113
00:07:48,430 --> 00:07:54,520
So if you think this picture is really nice, just copy this seat and let the prompt run again.
114
00:07:54,520 --> 00:08:01,690
But with a little bit of a different prompt and you will get even more consistent pictures.
115
00:08:01,930 --> 00:08:08,830
And that's basically it in my mind that this is really the easiest way to use stable diffusion Excel
116
00:08:08,830 --> 00:08:10,180
in a colab notebook.
117
00:08:10,180 --> 00:08:15,430
And of course, you don't have to make like young Instagram models or whatever.
118
00:08:15,430 --> 00:08:18,070
You can also make something like.
119
00:08:20,230 --> 00:08:26,020
Beautiful woman, 50 years old masterpiece in the garden.
120
00:08:26,020 --> 00:08:28,090
So you can really make whatever you want.
121
00:08:28,090 --> 00:08:31,240
Because this Google Colab notebook is really, really nice.
122
00:08:31,240 --> 00:08:35,800
You don't have to make j-10a, so simply be a bit creative.
123
00:08:36,400 --> 00:08:39,820
And there we have our two pictures of the older woman.
124
00:08:39,820 --> 00:08:43,600
So you can see we make pictures for whatever we like.
125
00:08:43,600 --> 00:08:46,900
And I think this is like nearly perfect.
126
00:08:46,900 --> 00:08:49,210
This is like from a photograph.
127
00:08:49,210 --> 00:08:56,170
You can always include better prompts if you like, but I just wanted to show you the easiest way to
128
00:08:56,170 --> 00:08:58,480
make pictures in stable diffusion.
129
00:08:58,750 --> 00:09:02,350
And the cool part is it's completely for free.
130
00:09:02,350 --> 00:09:07,300
So please, please try this out and have fun by trying this out.
12332
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.
Afrikaans
Akan
Albanian
Amharic
Armenian
Azerbaijani
Basque
Belarusian
Bemba
Bengali
Bihari
Bosnian
Breton
Bulgarian
Cambodian
Catalan
Cebuano
Cherokee
Chichewa
Chinese (Simplified)
Chinese (Traditional)
Corsican
Croatian
Czech
Danish
Dutch
English
Esperanto
Estonian
Faroese
Finnish
Galician
Georgian
German
Greek
Guarani
Haitian Creole
Hausa
Hawaiian
Hebrew
Hmong
Hungarian
Icelandic
Indonesian
Irish
Italian
Japanese
Kazakh
Kinyarwanda
Kirundi
Kongo
Korean
Krio (Sierra Leone)
Kurdish
Kyrgyz
Laothian
Latin
Latvian
Lithuanian
Luganda
Luo
Luxembourgish
Macedonian
Malagasy
Malay
Maltese
Maori
Mauritian Creole
Moldavian
Mongolian
Myanmar (Burmese)
Montenegrin
Nepali
Norwegian
Pashto
Persian
Polish
Portuguese (Brazil)
Portuguese (Portugal)
Quechua
Romanian
Romansh
Russian
Samoan
Scots Gaelic
Serbian
Sesotho
Setswana
Seychellois Creole
Shona
Sindhi
Sinhalese
Slovak
Slovenian
Somali
Spanish
Swedish
Tajik
Thai
Tigrinya
Tonga
Turkish
Turkmen
Ukrainian
Uzbek
Vietnamese
Welsh
Wolof