Download "НЕЙРОСЕТЬ ЗАМЕНЯЕТ СЪЁМКУ ВИДЕО / Как с помощью After Effects и Stable Diffusion создать КИНО?"
%3Aformat(webp)%2Fi.ytimg.com%252Fvi_webp%252FeKVTWVH_oVo%252Fmaxresdefault.webp&w=1139&q=75)
Please wait. We're preparing links for easy ad-free video watching and downloading.
Similar videos from our catalog
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252F9Y47MKeGR9c%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi_webp%252FBrs7KUMa_5M%252Fmaxresdefault.webp&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FMUKOfGTfcEk%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FiQ3aEX83T-s%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FoEkK4ATBEmY%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FwM8tMedCv-w%252Fmaxresdefault.jpg&w=1139&q=75)
Video tags
Subtitles
Russian
00:00:01
I will show you how to create beauty in00:00:03
Austria using neural networks where they00:00:05
will do almost all the work for you and00:00:08
most of them actually use neural networks00:00:09
incorrectly. Because you00:00:11
take some Table de Fusion,00:00:12
put your footage there and get00:00:15
every time a new picture is generated00:00:18
on each frame, but for a video this is not00:00:21
super applicable Because you want, for00:00:22
example, to make some kind of monster00:00:24
to animate it Yes, so that it00:00:27
is as realistic as possible,00:00:33
so our task today with you in00:00:35
this blog is to figure out how we we can00:00:36
get a plausible result and our00:00:39
research in this area began when00:00:40
I came across this video of this I smell00:00:53
And as we can see from the title of this00:00:55
YouTube video that the person uses two00:00:57
neural networks stable defusion and sing00:00:59
Probably this is what you and I need to00:01:02
cling to for in order to understand how to00:01:04
get such a result. Therefore, look00:01:06
now, I’ll show you everything. The first thing you and I will00:01:08
need is some kind of footage,00:01:10
because with an abstract background, I decided to00:01:12
build00:01:13
such a miniature photo studio right in my backyard and right there00:01:16
I shot this large one. the frame00:01:17
that worked best with her socket00:01:19
So you're all set You've taken the footage,00:01:22
now you need to do the following:00:01:24
install stable The Fusion on your computer00:01:27
Because this will be the most convenient way00:01:28
to work, or you can use the00:01:29
online version I'll now00:01:31
show you an example of the offline version00:01:32
on the internet you can find a huge00:01:33
number of tutorials because you can00:01:35
also install it on your computer.00:01:37
In the last issue we launched a00:01:39
full-fledged mini-course on animation with a00:01:41
very cool speaker Ilya But that’s00:01:43
not all because if you want to reach a00:01:45
new level of animation and make 3D00:01:47
animation of characters in Cinema 4D, then we00:01:49
also have material for you. It is suitable for00:01:51
more advanced users. Therefore, be00:01:54
prepared in this mini-course we will not teach00:01:56
you how to model 3D characters, but we will teach you how to00:01:58
animate them. So this is your first step to00:02:00
try yourself in the role00:02:01
of motion designer and then completely00:02:03
immerse yourself in this profession and this course00:02:05
will allow you to decide00:02:07
whether you want to work in 3dho or you’re fine as is. You00:02:09
can stay in 2D. But my advice to you is00:02:10
to run and study all this quickly because00:02:12
we’re giving it to you for free,00:02:14
so revive it your 3D characters And00:02:17
be happy everyone on the link we have opened with00:02:19
you a neurosec which you have not opened you00:02:21
will need to write it correctly00:02:22
the search query the correct query00:02:23
looks like this you write What do00:02:25
you want to see and in what style for00:02:27
example a portrait of a monster00:02:32
gigger Style or a votorial cgi00:02:39
render00:02:40
Aktan00:02:42
That is, you kind of00:02:44
throw in that search engine for imagine00:02:46
as if you were googling an image on the00:02:48
Internet for a request This is the same thing00:02:52
brothers forgot, there00:02:54
was only a neural network involved here00:02:58
for you and it will specially generate a00:03:00
moment then you and I move00:03:03
here here there is you can find the algorithms to compare them00:03:05
on the Internet here, I wo00:03:06
n’t give details here, I won’t be interested further in what00:03:09
we choose with you, the width of the picture and its00:03:12
height. I want a ratio of 16 to 9,00:03:14
something like this And the next00:03:15
parameter is very important srge Scale, the00:03:18
smaller it is, the more the neural network will00:03:20
adhere to your request and look for00:03:22
exactly this, that is, imagine you are00:03:25
in the forest trying to find a needle in a00:03:27
haystack and you are walking with a flashlight, if00:03:29
this is the corner of this flashlight, that is, why is it narrower and00:03:33
the chance of actually finding00:03:35
a needle is less because you only have a00:03:37
narrow beam But on the other hand, you can clearly see00:03:39
the picture far away, right in front of you. And the00:03:41
higher this value, the00:03:45
wider your beam and the greater the chance of finding a needle, but00:03:48
at the same time the request will be too00:03:50
wide, you may not get it. Yes And the next00:03:52
parameter is the Seed, that is, the random number00:03:55
on which the neural network will rely,00:03:57
if it’s -1 here, it means that every00:03:59
time it will look for your search query in a new place00:04:01
in its own00:04:03
ocean of possibilities, if you put a00:04:05
specific one, it’s like in short in00:04:08
Minecraft when you create seeds for00:04:09
each individual world which is randomly00:04:12
generated if you want a directly00:04:14
defined specific world You know00:04:15
it Sid you write it And the neural network, despite the00:04:18
fact that it is random even in Minecraft00:04:19
And you know, it will generate the00:04:21
same world, so I’m sitting here, the same thing for you and me00:04:23
later in the future it will come in00:04:24
handy, for now I’ll put a minus00:04:26
One because I don’t know what I want, I00:04:28
click00:04:29
Here it generated something for you and me that’s00:04:31
great, cool, but how can we get a00:04:34
result that is as identical as possible to the00:04:35
picture that you and I took?00:04:36
Quite simply, we go to imach00:04:38
to Image tab here we write, let’s say the00:04:40
same search query, but00:04:43
we upload a picture and then we need to00:04:46
break our footage into pnge00:04:49
sequence with you, that is, we have to feed00:04:51
this neural network to another one, too, photos are not00:04:54
videos, we can’t do this with you yet00:04:55
We upload ours file author00:04:57
after that we just add a sequence,00:05:00
look for that part of the image that we00:05:03
need,00:05:05
take it, crop it here, but this00:05:08
can be done in the premiere with any other editing00:05:10
program,00:05:11
then you take it, press it with control and00:05:13
here you paste it, output it to a PNG00:05:17
sequence, leave all other settings00:05:18
as they are, okay, I got everything from00:05:21
you a bunch of pictures now00:05:22
let's go back to the neural network further Our00:05:24
task is to select a key frame for each00:05:27
specific scenario, it can be00:05:28
different. I stand here and periodically00:05:30
open and close my mouth, so I will choose00:05:31
the one that I have with my mouth open00:05:33
because then the neural network will00:05:34
process everything much easier, press00:05:36
generate to to my request, I look at what00:05:39
I got, I see that it turned out to be00:05:41
some kind of nonsense, let’s get into the settings with you,00:05:43
firstly, you need to adjust00:05:44
the size because the size should be00:05:45
identical. Well, practically here00:05:47
at the stage of the first generations, the higher the size,00:05:50
the higher the detail, but you are00:05:52
detailing you can then raise a00:05:54
third one with a neural network that will simply00:05:56
fire at it, so we’ll do it that way,00:05:58
otherwise, here we’ll spend a huge00:05:59
amount of time with you, these are the00:06:00
dimensions. I chose this algorithm:00:06:03
stop sampling, another very good parameter that00:06:05
tells you how many times the neural network00:06:07
will pass through your picture from the00:06:08
detail depends now let's make it 20 let's00:06:10
see here again we look at the00:06:12
gfx-hale I would leave it at 7 and then00:06:15
there is a very interesting parameter dinozing00:06:17
stranks it is responsible for how00:06:20
strongly the neural network will adhere to the00:06:22
original image look we set it to 0900:06:24
bam it does almost nothing at all00:06:26
changed Well, I smoothed out the edges somewhere,00:06:28
accordingly, we’re trying with you, we’re twisting00:06:30
there more by 0.42 And now we’re00:06:33
getting closer to the result that00:06:35
we want You look at what a cool shadow And00:06:37
in general, the wildest like, we’ll remember, we’ll remember00:06:40
our parameters on how to do this Yes,00:06:42
because I want to continue experimenting00:06:43
I like it I’ll click on Save since he00:06:47
saved it here and I00:06:50
would also click here sent to extras00:06:52
here I’ll have this picture then00:06:55
shoot it So for now it’s left here00:06:57
we won’t come back again in the future00:06:59
let's move it here, I00:07:02
actually really like everything, so00:07:03
then I'll take the parameters and stop, if00:07:05
now I click generate again, then you and I will00:07:08
lose this result00:07:09
because each Seed is updated00:07:11
accordingly. I want to pull this Seed00:07:13
out of here, let's see what Seed00:07:16
this picture has. here it is written Here00:07:18
in the settings you see the Sid like this I00:07:22
copy it directly I paste everything here now I00:07:25
will specifically work with this00:07:26
result that I liked and00:07:28
then I will slightly raise the dinozing00:07:31
stranks if I go up too high00:07:33
see what happens then we will you got a00:07:35
cool result, but because of this I won’t be able00:07:38
to make a believable00:07:39
moving picture later, that is, a video clip, we00:07:42
got a very suitable00:07:44
image, considering what else is involved.00:07:46
If you change the width, height,00:07:49
firstly, the higher it is, the longer00:07:51
it will take you be displayed And secondly, in00:07:53
fact, this will greatly influence the00:07:55
final result. I received very00:07:56
different pictures. Despite the fact that00:07:58
all these parameters remained the same,00:07:59
therefore, taking this into account, it is better to immediately00:08:02
set the resolution to two times lower than what00:08:04
you need to get even and already00:08:06
then we’ll do all the things we00:08:08
got what we wanted since we’re00:08:10
saving this picture. Why because in00:08:12
this file that you and I will download and00:08:13
throw somewhere in the right place,00:08:15
not only the picture will be stored, but also the00:08:17
metadata inside and you can always00:08:19
return. It is to this00:08:20
interpretation of your request that this is done00:08:22
using a neural network and we can tweak the parameters if00:08:25
necessary, but we need to raise its00:08:27
resolution right up to the ideal so that it00:08:28
matches our footage, we say to the sent00:08:30
extras moved here here00:08:32
we write 2048 by 10 24 I have this00:08:35
footage I choose with the key one I select any of00:08:37
the algorithms, click here,00:08:41
it hasn’t already generated an image of00:08:43
my resolution and then I00:08:45
save it in a special place, now I’ll00:08:49
show you which one next to00:08:50
use the next neural network00:08:52
that will move our pixels,00:08:54
namely epsing, I need to create a00:08:56
structure of folders, how does it generally work,00:08:58
it takes from you some original00:09:01
image of yours that you drew there on the00:09:03
generated one, they also put something under it with00:09:06
your video, they look at how the pixels00:09:08
moved and shifted in your video,00:09:10
and in relation to this it moves00:09:12
what you drew or on the generated00:09:14
neurosec, so here I have such00:09:16
frames This is my00:09:18
generation of the Neuro mesh, I go there my00:09:21
folder with the structure here I have00:09:23
keyframes Here is my Shot that I00:09:25
generated since it is called00:09:28
173, it is very important that it is called exactly the same00:09:31
as my original picture files,00:09:33
yes yes, I took it 73 frames,00:09:36
respectively 1.73 and there it should00:09:38
also be called 173 I take it means00:09:41
I’m pointing out since he hooked me up with these little00:09:45
touches, it means that he started00:09:48
counting them correctly here is the video I choose00:09:51
my source, well, the first picture00:09:55
I’m just stitching everything together he also picked up00:09:57
this sequence, he understands this Musk, you00:09:59
can have a sequence derived from00:10:02
masks If you want, for example, to move00:10:04
only your character, you don’t want00:10:05
it to be in the background, I’ll show you00:10:08
how to get rid of it in the author, just00:10:10
two clicks, here we go further00:10:12
here I select from the first frame to 7400:10:14
because I have so many frames in my00:10:17
video and I say that I have frames like this00:10:18
73, then I select some folder00:10:21
where all this will be output and I say00:10:23
we can see how it’s done right in00:10:25
real time Our neural network00:10:27
is completing the second frames,00:10:29
so if you and I click, we will already see00:10:32
our entire00:10:34
picture in motion, but we need to00:10:37
load it into the author now and do just that, we00:10:39
see that the network is over.00:10:41
Turn it off, open the author, drop00:10:44
our sequence here, look to00:10:46
turn it on, say import, see noise00:10:49
interpreted it incorrectly 30 frames00:10:51
set I had 25 change it create00:10:54
the composition ready Now you see the foundation00:10:58
is moving I don’t like it I create a double00:11:00
I say there here for example in this frame00:11:03
Free Frame we draw a soft mask00:11:07
I shade and you get the most amazing00:11:10
result00:11:12
[music]00:11:16
here bro So if you have fun working00:11:18
neural networks You want even more information00:11:20
You have a like and there’s also a comment below and00:11:22
maybe you’ll even take a full00:11:25
course on neural networks from us, too, let us00:11:27
know about it. For us, this is very important. Let’s do00:11:30
moreDescription:
👉 Ссылка на бесплатный мини-курс https://sabatovsky.com/characteranimationfree Привет, творец. На связи Слава Хохлов. Основатель Хохлов Сабатовский. Тут ты найдешь всю информацию, что бы научиться создавать киношные видео и построить профессию. ► Переходи в мой телеграмм канал https://t.me/+hhNn1iT40Hs5YmRi ► Хочешь получить творческую профессию? Записывайся по спец. условиям в мою школу https://sabatovsky.com/ ► Нужно заказать видео или найти исполнителя? Переходи на сайт моего продакшна https://sabatovsky.com/production ► Хочешь попасть к нам в команду? Заполняй анкету https://docs.google.com/forms/d/e/1FAIpQLSfpqDwNntb12TB55VmFdcY59N70bGH41m7IkM_iGAzq3wyUAQ/viewform?usp=send_form ► Хочешь стать партнером нашего проекта? Заполняй анкету https://docs.google.com/forms/d/e/1FAIpQLSfpqDwNntb12TB55VmFdcY59N70bGH41m7IkM_iGAzq3wyUAQ/viewform?usp=send_form Для студентов - ссылка на тех. поддержку https://t.me/khslearn_bot Production Music courtesy of Epidemic Sound
Preparing download options
* — If the video is playing in a new tab, go to it, then right-click on the video and select "Save video as..."
** — Link intended for online playback in specialized players
Questions about downloading video
How can I download "НЕЙРОСЕТЬ ЗАМЕНЯЕТ СЪЁМКУ ВИДЕО / Как с помощью After Effects и Stable Diffusion создать КИНО?" video?
http://unidownloader.com/ website is the best way to download a video or a separate audio track if you want to do without installing programs and extensions.
The UDL Helper extension is a convenient button that is seamlessly integrated into YouTube, Instagram and OK.ru sites for fast content download.
UDL Client program (for Windows) is the most powerful solution that supports more than 900 websites, social networks and video hosting sites, as well as any video quality that is available in the source.
UDL Lite is a really convenient way to access a website from your mobile device. With its help, you can easily download videos directly to your smartphone.
Which format of "НЕЙРОСЕТЬ ЗАМЕНЯЕТ СЪЁМКУ ВИДЕО / Как с помощью After Effects и Stable Diffusion создать КИНО?" video should I choose?
The best quality formats are FullHD (1080p), 2K (1440p), 4K (2160p) and 8K (4320p). The higher the resolution of your screen, the higher the video quality should be. However, there are other factors to consider: download speed, amount of free space, and device performance during playback.
Why does my computer freeze when loading a "НЕЙРОСЕТЬ ЗАМЕНЯЕТ СЪЁМКУ ВИДЕО / Как с помощью After Effects и Stable Diffusion создать КИНО?" video?
The browser/computer should not freeze completely! If this happens, please report it with a link to the video. Sometimes videos cannot be downloaded directly in a suitable format, so we have added the ability to convert the file to the desired format. In some cases, this process may actively use computer resources.
How can I download "НЕЙРОСЕТЬ ЗАМЕНЯЕТ СЪЁМКУ ВИДЕО / Как с помощью After Effects и Stable Diffusion создать КИНО?" video to my phone?
You can download a video to your smartphone using the website or the PWA application UDL Lite. It is also possible to send a download link via QR code using the UDL Helper extension.
How can I download an audio track (music) to MP3 "НЕЙРОСЕТЬ ЗАМЕНЯЕТ СЪЁМКУ ВИДЕО / Как с помощью After Effects и Stable Diffusion создать КИНО?"?
The most convenient way is to use the UDL Client program, which supports converting video to MP3 format. In some cases, MP3 can also be downloaded through the UDL Helper extension.
How can I save a frame from a video "НЕЙРОСЕТЬ ЗАМЕНЯЕТ СЪЁМКУ ВИДЕО / Как с помощью After Effects и Stable Diffusion создать КИНО?"?
This feature is available in the UDL Helper extension. Make sure that "Show the video snapshot button" is checked in the settings. A camera icon should appear in the lower right corner of the player to the left of the "Settings" icon. When you click on it, the current frame from the video will be saved to your computer in JPEG format.
What's the price of all this stuff?
It costs nothing. Our services are absolutely free for all users. There are no PRO subscriptions, no restrictions on the number or maximum length of downloaded videos.