00:00:05
[Music]
00:00:07
Good morning children How are you
00:00:11
daddy brought food for all of you
00:00:21
let's pray before eating
00:00:25
[Music]
00:00:33
this is a video with 3D animation style
00:00:36
What I made was inspired by
00:00:38
a free nutritious meal video that some time ago
00:00:41
was circulating on social media
00:00:43
Not only inspired, I even
00:00:45
intentionally made the story according
00:00:47
to the one in the original video
00:00:49
If you may have never watched
00:00:51
the original video made by the Ministry of Communication and Information
00:00:53
you can just search
00:00:55
on YouTube free nutritious meal video
00:00:57
you will definitely find it. but don't compare between
00:01:00
my creation as an ordinary citizen
00:01:02
and the creation of the Ministry of Communication and Information, it must be different
00:01:04
even though the story is similar
00:01:06
Because I made it by copying a video made by
00:01:07
this Ministry. but here I recreate it
00:01:10
the visuals are different
00:01:12
my purpose in making this video
00:01:14
just to practice how to make a 3D animation style video
00:01:16
using Ai then I can share
00:01:19
the tutorial to those of you who might
00:01:21
want to make animated film content
00:01:24
I hope this video can be useful because
00:01:26
I made it using the Dzine Ai tool
00:01:28
a multifunctional tool with many features
00:01:30
Not only can you make images and videos
00:01:32
This tool also has a custom character training feature
00:01:34
so we can make
00:01:37
consistent character of people in many scenes
00:01:40
In addition, there are many other editing features
00:01:42
that can be used, for example in
00:01:44
my previous video on how
00:01:46
to apply a hijab product photo to a model made by Ai
00:01:48
and many more features
00:01:50
in this Dzine Ai tool
00:01:52
for the Dzine Ai link I've written in
00:01:55
description column, you can just check there
00:01:56
let's start this tutorial right away
00:01:59
usually the first stage I always
00:02:01
create a story idea first with the help of chatgpt
00:02:04
now because of the video story idea
00:02:05
I adjusted it to the original free lunch video
00:02:07
then the first stage
00:02:09
I immediately asked chat PT to make a scene prompt
00:02:12
the way is like this. here
00:02:14
I immediately uploaded a free nutritious food video file
00:02:17
here I have uploaded
00:02:19
then I added a Prompt like this
00:02:21
I have attached an animated video file
00:02:23
blah blah blah blah blah and I asked ChatGPT
00:02:26
to make a Prompt the scene image
00:02:29
directly like that. Well, below
00:02:31
I will give you chatgpt information or details
00:02:35
to be considered in making
00:02:37
image prompts from the video
00:02:40
This is an example of the core scene
00:02:42
I will just give you the outline of the video
00:02:44
showing a man comes to
00:02:45
a school carrying a stack of lunch boxes
00:02:48
then goes into the
00:02:50
classroom and so on, the second one make sure
00:02:53
Prompt produces composition details
00:02:55
and the same feel as the reference video
00:02:57
then I give information too
00:02:59
to chatgpt that I will remake this video
00:03:02
with Ai text to Image tools
00:03:04
then image to video. pay attention to each image Prompt
00:03:07
contains elements of photography
00:03:09
which is cinematic as follows
00:03:11
subject, scene, composition, lighting and others
00:03:13
and so on then chatgpt answers
00:03:16
like this. he starts analyzing the video file
00:03:18
that we uploaded. here chatgpt
00:03:20
said he would analyze the frames
00:03:22
to determine the main visual elements
00:03:24
such as lighting, color, composition, and
00:03:26
others, until finally chatgpt
00:03:29
can directly give us a scene prompt
00:03:32
according to what we want. well this is the result
00:03:34
here is the Prompt, the first, the man came
00:03:37
to school bringing food then
00:03:39
entered the class greeted by students with
00:03:41
applause. the man distributed food
00:03:43
to the students. praying together before eating
00:03:46
children are happy blah blah blah
00:03:49
men end with namaste then clap their hands
00:03:51
then here I write
00:03:54
the details that I want to pour
00:03:56
in the image prompts
00:03:58
you can read this yourself, okay? Well
00:04:00
For example, for all Prompts
00:04:02
produces cartoon animation style images
00:04:05
then the Prompt that displays
00:04:08
the face of the main character is a man carrying
00:04:10
food then chagpt must consistently
00:04:13
write the character traits
00:04:15
Well here I write
00:04:17
the characteristics are roughly like this
00:04:19
age over 50 years blah blah blah blah
00:04:22
black hair, navy blue short-sleeved shirt
00:04:25
to details like where to eat
00:04:28
must be silver or chrome
00:04:30
and don't forget this is also important
00:04:32
all students wear white shirts
00:04:35
then chatgpt revises the prompts earlier
00:04:38
perfecting according to the details I added
00:04:40
like this. Well here
00:04:43
I feel these six scenes are still incomplete
00:04:46
because these six scenes
00:04:48
are just the main scenes. if we
00:04:50
only make these six scenes
00:04:52
the video will feel choppy and the story
00:04:55
it won't be coherent, right? so we need
00:04:57
add additional scenes
00:05:00
then I ask here chatgpt to make 10 Prompt
00:05:03
new additional scenes, so that the video
00:05:06
is more coherent and interesting to watch
00:05:08
Now here I give an additional example
00:05:10
like a scene showing the main character
00:05:12
arriving in front of the classroom door
00:05:15
another example here, there is a male main character
00:05:17
walking among the students
00:05:19
then there is a close up scene of the male character
00:05:23
praying. so in this process, if we want
00:05:26
add more details that we want
00:05:28
we just go back into chatgpt
00:05:30
Now then chatgpt rewrites
00:05:32
all Prompts, totaling 6 + 10
00:05:36
additional Prompts, so there are 16 Prompts in total
00:05:38
So if you feel that everything is complete
00:05:41
This Prom is in accordance
00:05:43
with our wishes, then actually
00:05:45
this is finished, we can
00:05:47
generate this Prompt into an image
00:05:48
we enter the stage next
00:05:51
namely creating a dataset in the form of a 3D cardboard image
00:05:53
the main character in this video
00:05:56
so before we start generating the scene image
00:05:58
we first create a dataset
00:06:01
in the form of a cartoon image of the main character
00:06:03
which will often appear in this video
00:06:06
for us to use as custom character training material
00:06:08
so in this video I made
00:06:10
this 3d cartoon character comes from
00:06:13
original photos, so I collected all
00:06:16
photos of the main characters that will
00:06:18
be taken in this video, you can
00:06:20
prepare where they come from
00:06:22
whether it's my own photos or if you want to take them from the internet
00:06:24
because if in Dzine
00:06:26
written We are told to
00:06:29
collect 1 to 40 image files to
00:06:32
make a data set. Well here I
00:06:35
around maybe not up to 20
00:06:37
I've collected these photos then
00:06:39
we go straight to the image to Image feature
00:06:41
in Dzine Ai. now the display
00:06:44
is like this for the Ai design tools.
00:06:46
Now in the image to Image feature, we set
00:06:48
the settings are like this. we choose 3D
00:06:50
and this is specific in Qute 3D, for style intensity
00:06:55
we choose the weak 0.4
00:06:57
structure match choose the most strict
00:07:00
then we choose activate Face match
00:07:03
and high quality, that's all,
00:07:06
then we upload the photo we want to convert
00:07:09
into a cartoon image, after the upload is complete
00:07:12
we can immediately click generate
00:07:14
and the results are like this
00:07:17
we try again for other images
00:07:19
we upload here, then
00:07:21
let's make sure the settings are all the same
00:07:23
if it's ready, we just generate it
00:07:25
the result is like this
00:07:27
Now here we just choose which one is the best
00:07:29
in our opinion
00:07:31
Now to tidy up the results of this image, we delete
00:07:34
the background part, so that it doesn't
00:07:37
disturb the training results later
00:07:39
because this background is too busy
00:07:40
we just place it on Canvas, then
00:07:42
we remove the background like this
00:07:45
when it's finished, then we can
00:07:47
directly download the results like this
00:07:50
Here are some images that I have deleted
00:07:55
the background. we continue directly
00:07:57
to the custom character training process
00:07:59
here we use the consistent character feature
00:08:02
then we make a new character. Well here
00:08:04
we can enter the name of our character
00:08:07
Let's say here, the cartoon president
00:08:09
then we upload all the images
00:08:12
that we have prepared earlier
00:08:14
we just wait for the upload process
00:08:16
it's done, we can start right away
00:08:18
train process. Well here it says the duration
00:08:21
approximately around 19 minutes. let's just wait
00:08:23
until it's done. Well, when it's finished,
00:08:25
our character results will appear
00:08:27
on this list. we can continue directly
00:08:29
generate Prompt into a scene image
00:08:31
now when we choose our character
00:08:33
he will automatically appear here
00:08:36
this is the Prompt that appears automatically
00:08:38
describing the characteristics of the character
00:08:40
Now here I do slight adjustment
00:08:42
from the character traits, to match
00:08:44
the character traits I want
00:08:46
here there is light skin, short Combed Black Hair
00:08:49
close lip smile, then
00:08:52
the character also wears a short-sleeved navy blue shirt
00:08:54
now details like that
00:08:57
we can adjust
00:08:59
according to the character we want
00:09:01
then in the Prompt column below,
00:09:04
this is used to enter our scene Prompt
00:09:07
which we generate from chatGPT
00:09:09
Now here the character has been represented
00:09:11
with the sentence cartoon president
00:09:14
we add here, He is, next
00:09:16
we go to chatGPT, for example this Prompt
00:09:19
we start from the word standing,
00:09:21
down, now for the word
00:09:23
the one before standing
00:09:25
he describes the character traits, while in Dzine earlier
00:09:27
we've already made a custom character
00:09:29
so we don't need to
00:09:31
include the character details from this GPT chat
00:09:33
because it's already represented
00:09:36
by the cartoon president's sentence earlier.
00:09:39
here we just copy it, then we
00:09:41
paste it here. Now here you can see
00:09:43
He is standing blah blah blah blah
00:09:45
until finished, then the settings
00:09:48
we choose 16 to 9. That's it.
00:09:51
then we directly generate
00:09:53
and the result is like this
00:09:54
we just repeat these steps
00:09:57
to generate all the scene images
00:09:59
then we go to
00:10:01
the process to convert this image into a video
00:10:04
here I use Runway
00:10:06
Now for this first scene
00:10:08
The prompt is roughly like this
00:10:10
tracking shot, the man walks forward, the camera
00:10:12
follows from behind. then we generate
00:10:14
and the result is like this
00:10:17
Now let's try for another scene
00:10:19
like this. here I use
00:10:21
Prompt: Static shot the man is greeting, natural
00:10:24
movement. we generate and the result
00:10:26
like this. and this is for the video generate result
00:10:30
for other scenes
00:10:32
here I have included the prompt
00:10:35
friends, believe it or not
00:10:37
most of the Prompts I use for
00:10:40
convert this image in Runway, that's almost all of it
00:10:43
using the natural Movement Prompt
00:10:45
with this Prompt alone I can
00:10:47
when I use a more detailed Prompt
00:10:49
sometimes the results are not better
00:10:50
There are only a few scenes
00:10:53
that need a More Detailed Prompt
00:10:56
For example, He is Walking
00:10:58
natural Movement
00:10:59
we enter the stage of making a lipsync video
00:11:01
using Runway act one to
00:11:04
get lipsync results with character voices
00:11:06
as we want
00:11:08
which in this video I want to use
00:11:10
a voice similar to Mr. Prabowo.
00:11:12
the way I do it is by taking a voice sample from a YouTube video
00:11:14
that shows him giving a speech
00:11:16
or a greeting. then I process
00:11:19
the audio sample file to the voice cloning tool
00:11:22
here I use a free tool
00:11:24
it's called vocloner. the method is like this
00:11:27
here we open vocloner, then
00:11:29
we enter the name of the voice character
00:11:32
now in this section, we will upload the audio sample
00:11:34
here I will play a little
00:11:36
for the audio sample that I took
00:11:39
from YouTube. like this. "with this step
00:11:41
which will come into effect in March"
00:11:44
well something like that. that's the original voice
00:11:46
after we upload we immediately process submit
00:11:48
Now when the cloning process is complete
00:11:51
then the name of the voice character
00:11:53
will appear here
00:11:55
then we continue to the text to speech process
00:11:57
changing the text into voice
00:12:00
according to the character we have created
00:12:01
now in this section, we enter the text that
00:12:04
we want to change into character voice here
00:12:06
then the format is MP3. then we
00:12:09
generate speech. the result is like this
00:12:12
"Good morning children, How are you,
00:12:15
father brought food for all of you
00:12:18
let's pray before eating"
00:12:20
Well, I think this is pretty good, if for
00:12:23
the color of the voice is very similar
00:12:25
although indeed the intonation is a bit stiff
00:12:28
then the next stage
00:12:30
we prepare Driving video for us to use
00:12:32
in Runway act one feature. so here
00:12:35
I've prepared it, I'll record it
00:12:37
close up of the face acting to say
00:12:40
the words we generated earlier
00:12:42
the result is like this. footage recording
00:12:44
that I took. "good morning kids
00:12:47
how are you" so this video I act
00:12:50
saying the words we made earlier
00:12:51
we try to be as similar as possible
00:12:54
and then this video footage
00:12:56
we put it into capcut
00:12:58
at the same time as the character's voice that we made earlier
00:13:00
we make the lip movements
00:13:02
according to the voice that we made earlier
00:13:04
so it's something like this
00:13:06
after we arrange it, just adjust it
00:13:08
the result is like this. "good morning kids
00:13:11
how are you" and this is Driving video
00:13:14
for the second scene
00:13:17
"let's pray before eating"
00:13:19
Now after we adjust the lip movements
00:13:22
with the voice of the character we have created,
00:13:24
the result is like this. "let's pray before eating"
00:13:26
if it's okay, we just export this video
00:13:28
then we save it
00:13:31
then we bring the finished video
00:13:33
to the act one feature on Runway
00:13:36
at the top here, where we upload Driving video
00:13:39
that we have made earlier
00:13:41
"let's pray before eating" then in the bottom section here
00:13:44
we upload for the character video
00:13:47
that we have made
00:13:48
we wait for the upload to finish
00:13:50
then we immediately generate
00:13:52
and the result is like this. "let's pray before eating"
00:13:55
and for the second lipsync scene
00:13:57
the result is like this.
00:14:00
"good morning kids, how are you" before we go into
00:14:03
video editing in capcut
00:14:05
Actually there is a stage of determining the background music
00:14:08
but here I have often taught my friends
00:14:10
where to get
00:14:12
free music easily
00:14:14
or my friends can make it themselves through
00:14:16
song making platform using Ai
00:14:19
So I think friends
00:14:20
there is no difficulty in getting music
00:14:22
this can be taken from anywhere according to
00:14:25
needs. Okay so I continue
00:14:27
go straight to the video editing stage in capcut
00:14:29
and here is
00:14:31
the capcut PC display, guys
00:14:33
you can see here, I will explain from the beginning
00:14:35
in the first scene, here is text
00:14:38
friends can design their own text
00:14:40
there are many choices of effects and fonts in capcut
00:14:43
plus animation to make it more interesting
00:14:45
here there are footsteps in all parts
00:14:48
When walking. there is the sound of footsteps
00:14:51
well when starting to enter
00:14:53
into the class, there is an additional sound effect
00:14:56
like the sound of a crowd in a room
00:14:59
like that. Well in this scene
00:15:01
there are also additional sounds like children's voices, yes
00:15:04
we have often heard this sound
00:15:06
in the scene that shows food
00:15:08
we can give it here, there is an effect
00:15:11
friends can use it in this effect section
00:15:13
there are many choices of what kind of effect you want
00:15:15
plus there are sound effects
00:15:18
Sparkling sound. continue.
00:15:20
because this is still in class so still
00:15:22
use sound effects like crowd noise
00:15:25
in this section, there is also text
00:15:27
but in Bubble form
00:15:29
friends can use this in the Bubble menu
00:15:30
here. continue.
00:15:32
until the last scene there is also writing
00:15:35
that we can set according to what we want
00:15:38
it's like that
00:15:40
again for how to edit videos
00:15:42
depends on our taste and
00:15:44
our respective creativity
00:15:52
That's all for today's video tutorial
00:15:55
I've explained
00:15:57
from start to finish
00:16:00
how to make 3D animation videos using the Ai tool
00:16:02
I hope this video can help you
00:16:05
who want to create animated video content
00:16:06
because with the help of tools
00:16:09
like Dzine Ai, making videos with
00:16:11
3D animation style is easier
00:16:13
want to make various stories
00:16:15
no matter how creative, all can be made with the help of Ai
00:16:17
and what's even more special especially
00:16:20
with the presence of tools like Dzine Ai
00:16:22
now make more consistent characters
00:16:24
in every scene becomes easier
00:16:26
if for content purposes on social media
00:16:28
this tool seems to be more than enough
00:16:30
because the editing features are also very complete
00:16:33
press the button like if you like
00:16:35
with this video and
00:16:37
if you have any questions, don't hesitate to comment
00:16:40
If you want to ask more intensely
00:16:42
you can join our discussion group on telegram
00:16:44
the link has been written in the description column
00:16:46
see you in
00:16:50
[Music]
00:16:57
next video
