In questo video vediamo 6 trucchi o tecniche per utilizzare al meglio il nuovissimo strumento "Riempimento generativo" di Photoshop CC 2024, che da poco è stato implementato nella versione finale dell'applicazione.
Hi guys, welcome back to my channel, welcome to this new video, I'm Marco Agustoni of Graficata.com
Today we're going to see some techniques, I'll give you some tips, some interesting tricks to use the new incredible Photoshop technology called Generative Filling
that you've been hearing about for a few days, it was finally implemented also in the final version, the version of 2024
For those who don't know it yet, Generative Filling, you can also see it in this animation, but you can also read it below
allows you to add or remove contents from our images in a non-destructive way using simple text prompts
so through a description that we can add in this input field that you see here
we can describe what artificial intelligence is going to generate in our images
you see that here it generates three different cases and it does it in an extremely accurate and precise way
this is certainly the most important novelty of the last 20 years in Photoshop
and I'm sure it will literally change the way we are used to working with this application
but this is not the only novelty, you see that in this window we have a recap of the new functions
so in addition to the Generative Filling we also have the Generative Expansion that is based on the toolbar
and also in this case it exploits artificial intelligence, so you see that it starts from a square image, expands the canvas
and then it generates the missing background automatically, always through artificial intelligence
this is also a fantastic tool. In addition to this, we have an improvement regarding the Remove tool
which always allows, through AI, to remove elements automatically from our images
and then we have an improvement that also concerns the contextual activity bar
which has been further improved with new functions and new tools
so these were some of the novelties, in this video we will talk more than anything about Generative Filling
and we will see some tricks, some techniques to use it to the best.
Before getting into the lesson, I would like to show you the guidelines that Adobe has created for this Generative Filling
because obviously the company had to impose some rules, it had to put some limits on this technology
it put some limits that we will also see directly in Photoshop when we try to use this technology
which does not allow us to do anything, it had to create these limits to prevent people, users
from using this technology inappropriately
and on this page that I link below we have a sort of re-epilogue
basically of what we can do and what we cannot do through this technology
you can see that in point 1 it is written that no commercial use of the beta version is allowed
in reality point 1 is no longer valid because now this technology is out of the beta version
so all the images that we are going to generate can also be used for commercial purposes
but the most important point in my opinion is point 3 which says to be respectful and correct
and here you can read that it is forbidden to try to create, upload or share offensive, illegal or reserved content
here we have some cases that, for example, it is forbidden to upload or use pornographic material
content that incites hatred, highly offensive, violence, blood, promotion, exaltation of threat or violence
activities or illegal goods, etc.
I repeat, I leave you the link below so you can also understand what you can do and what you cannot do with this technology
but you will also see it when you try to generate some things in Photoshop that do not respect these guidelines
you will be blocked and a pop-up will appear that will tell you that you are not respecting the guidelines, ok?
Anyway, be very careful when using this technology because in point 7 it is written that
Adobe could act on your account if it detects content or behaviors that violate the present guidelines
so be careful not to mess up
Ok, having said that, I would say that we can finally enter Photoshop to see the first trick of this lesson
the first suggestion I would like to give you to best use this generative editing technology
is to pay particular attention to the text prompt, then to the description that you are going to put in the input field
Let's take a practical example right away
I'm going to create a selection, I'm going to call the generative fill
and then in this field I can go and describe what should be generated in my selection
I said describe and this is very important to understand
because many people try to give instructions in this field
by instructions I mean, change the color, modify this, modify that
do this, move the element, enlarge it, create a scene, etc, etc
it does not work with instructions, it only works with very detailed descriptions
and this is also specified by Adobe on the official website
here we can read that for optimal results it is advisable to use descriptive and adjective names
instead of messages and instructions such as fill the area, create this, create that, etc, etc
ok, so to use it best, describe what AI, artificial intelligence, will have to generate
and this also makes us understand that if we only wanted to change the color of this shirt
maybe this technology is not the most appropriate choice
because this technology will also change the shirt
I'll show you right away, if I just wanted to change the color
I could try the problem with this new technology
now I create a very fast selection without being 100% precise
I create a selection of the shirt
and I try to give an instruction, which should not be done, directly in this field
so make the shirt blue
let's see what happens
here you see, the shirt has been modified, it has been partially made blue
the problem is that it has been literally changed
you see, we have to do with another shirt, completely different
so this technology works very well, but it goes to modify the elements that we are going to select
if instead I wanted to keep the shirt completely unchanged
I would have to proceed in the old way, let's say, with the various selection tools
as I did here, creating a mask, adding a level of adjustment, etc.
so I recommend, in the input field, go and write and not give instructions
if you have to make these variations in color, size, move objects
still proceed manually, because this technology is not yet ready to do this type of work
maybe it will be in the future, in the next months or years
the same goes for the enlargement or reduction of the elements
so here I can go and select the dog, for example, to reduce it
I could try to take advantage of artificial intelligence
maybe I'll expand the selection
I could say, write, I don't know, reduce the dog
but I already know it doesn't work
this is to show you what not to do with this technology
let's wait a few seconds
you see that, among other things, here we have some very interesting suggestions
you see that he is not going to reduce the size of the dog
but he is going to replace it, because he is going to interpret only the word dog
and among other things, an interesting thing that is also written here
is that from this definitive version of artificial intelligence
we can write our prompt in over 100 languages
this is interesting
the only problem is that the requests, as written here, in a language other than English
are translated into English by the translation engine
so we can go and write in Italian
he will then translate into English
so be careful, because translations are not always precise
so our prompt will not always be interpreted correctly
I would continue to write English prompts
obviously, guys, then you see what is most convenient for you
let's see now, guys, a second trick to be able to best use the generative filling function
we have seen that in order to recall this function
it is necessary to start from a selection
so the selections are a bit the fulcrum of this new tool
and it is very important to understand that the shape that we assign
and also the dimension that we assign to our selections
will determine the shape and dimension of the object that will be generated
so if I wanted, for example, to add a hat here, in this area
I could do it by assigning immediately to my selection
the shape that the hat will have
so if I wanted to add, for example, a triangular hat
I could start from a triangular selection
now I'm using the Lazzo tool
click on Generative Filling and simply write hat
you will see that the software takes account of our selection
and will generate a hat of that shape
ok, you see, it is not really triangular, but it remained within the selection
you see, this yes
ok, then obviously we can generate other variants
but it remained within the selection and kept more or less the same shape
I can go back and, I don't know, go and insert a cylindrical hat
so create a rectangular selection, you see
with the Generative Filling tool, hat
and it should respect the shape by generating a cylinder
you see, one, two, this is still triangular
and three
ok, let's do a third test
obviously I could also describe better the hat that I want to get
for example, let's do a fairly wide selection
let's write pirate hat
I expect now that I'm going to create a huge hat
you see, a pirate hat that always enters our selection
so it is very important, I repeat, the shape and the starting dimension of our selection
mark this and remember it
always talking about selections
I discovered that there is a little trick to remove objects, subjects
always exploiting this technology in an automatic and precise way
so let's work on this image
and suppose we want to delete the subject on the left
so first of all we have to select it
in this case I will use the tool Select Object
I will click, you see, on the subject that is automatically identified
and I try, through the generative filling, without writing anything
because the input field is optional
so in reality we could also leave it empty and let the application do the work
I try to click on Generate and let's see what happens
ok, we can see that it was obtained, I don't know what, something strange
here too, different variants, you see
the subject was fundamentally modified
it was not our intention to modify it
our goal was to remove it completely
so let's go back and see how to do it
always exploiting this technology
so I always start with a selection
I now expand it slightly
so Expand Selection
I can also go to the Selection menu, Modify, Expand
here I expand it, let's make it 15 pixels
without exaggerating, ok
and after that I click again on Generative Filling and Generate
let's see now if something changes
so, we have seen that there has been an improvement
a little girl has been added here
some palettes have been added here
while here a person is running
but the car has already been rebuilt here behind, etc.
I go back and try again
or I can click again on Generate
I noticed that by expanding a selection even slightly
the program understands better that we want to delete the object in question
so when you find yourself having to delete elements
expand the selection and always try with Photoshop's Generative function
here we have this element
I could go and select it this way
and I could look again through the Generative Filling
to remove it, let's see if it works
or I can also use the other tools
anyway, this is a great result
we can see that he has perfectly rebuilt the pixel part that was behind the subject
you see that now he has removed practically everything
we have the usual three versions
but I would say that this is a great result
so we had to do two steps
but it is important to understand that when you expand the selection slightly
the program understands that we want to delete the figure
I'll show you another example
using this image
I'm always going to select the subject
I expand the selection by 15 pixels
in fact let's do something more
let's do 30 pixels
and I try to delete the dog
ok, you see, it was very precise
let's learn now another really cool trick
which allows you to manage the different intensities of the elements
that we are going to generate thanks to artificial intelligence
let's start with this image
and let's try to add a turtle right here under water
let's take a selection tool
like the elliptical selection tool
let's create a kind of circle
and then let's write
sea turtle
let's see the result
ok, this result
we have generated our turtle in three different versions
the problem, let's see if you can understand it
obviously the turtle has colors that do not reflect
those that could be the colors of a turtle underwater
the colors are too intense
it almost seems that the element has been stuck
we have to understand how to reduce the intensity of our turtle
and we can do it through a technique
so I go back, CTRL-CMD-Z
and deselect with CTRL-CMD-D
at this point I go to the fast Photoshop mask mode
with the key Q on the keyboard
I make sure now that
as far as the foreground color is concerned
a white has been set
while as far as the background color is concerned
the black, this is very important
you can access the fast mask
even by clicking on these icons above
you see, it makes us understand that
we are currently in this mode
at this point we use the combination
CTRL-CMD-Backspace
to go and color our image red
be careful because we are in the fast mask
ok, so actually we have not colored our image
this is just a particular visualization
that allows us to understand what we are going to select
so now we take the brush
with the letter B on the keyboard
double click on the foreground color selector
and we are going to modify the intensity of this white
here we have shades, saturation and brightness
we lower the brightness from 100% to 40%
obtaining a gray
this is very important
so be sure to pay attention to the color you get
let's go now to brush
maybe let's increase the flow
so you can see better
let's go to brush
more or less to create the area
where we want to insert the turtle
ok, more or less like this
let's exit the fast mask
with the letter Q on the keyboard
ok, we should see if the selection
is not visible, but do not worry
because it is actually present
fill in the generative
and let's type again
sea turtle
here it is guys, the result
which is completely different
compared to the previous one
now it is really realistic
you see what are the colors
of our turtle
that blend perfectly
with the starting image
version 1, version 2, version 3
obviously we can test
different intensities
and also different animals
if we wanted to insert a dolphin
we could do it easily
so this is the trick
of the intensity that few people know
let's see now a trick
to blend literally two images
we will start from these two images
we will try to insert the turtle
right here in the center of this vegetation
so I'm going to take
the image with the turtle
I'm going to select the subject
I click on select subject
I leave everything to Photoshop
I just limit myself to make a copy
CTRL or CMD C
and a paste in the other image
you see that the turtle is too big
so I'm going to scale it a little
and I'm going to position it
in the center of my canvas
so it is clear that the turtle
has been taken from another image
also because it has completely different colors
let's try now to understand
how to blend it using artificial intelligence
let's start from a filter
the harmonization filter
to make the colors
consistent with the background image
I go to filter
neural filters
this is a fantastic filter
it's called harmonization
and it allows to make
the colors of the various levels
consistent and homogeneous
I go to select the reference level
which is the background
and you see how the image of the tiger
has already changed
I click on OK
and I start to fix the colors
at this point I try to use
artificial intelligence
to better blend the image of the tiger
with the background
so I take the level 1
CTRL or CMD
and I click on the thumbnail
then I go to selection
modify, border
I create a border
so I expand and contrast the selection
I create a border of 100 pixels
let's see what happens
ok perfect, I would say that 100 pixels is fine
now I click on fill
I do not insert anything in the input field
and I simply click on generate
ok, you see that already
I have now three versions, they are not all precise
I can obviously continue to generate
new variants
but you see that the tiger has now been
completely merged with the image
below, let's try to generate
other variants
ok, in this case they are less precise
I would say that the best, in my opinion, is the first
look compared to the
initial version
so
and so, you see obviously
artificial intelligence has also changed
the ears and the tiger
but it blends it very well
with the background image
and this is certainly a technique that will come in handy
when you have to merge more levels
automatically
now let's see the last technique
the last trick
this time I'm going to open
not a photo but a video
we know that Photoshop can also manage
short videos, here we have the timeline
in the lower bar, you see this parrot
that simply turns its head
I could take advantage of artificial intelligence
or rather the cutter with
generative expansion to go
to modify this image or rather
to make it vertical, so
I'm going to create the missing part below
I'm going to select the cutter
I select
generative expansion and at this point
I expand the canvas
in the lower part
I click on send, you see that
the image has been expanded, but in reality
this is a video, so I can click on play
and I have actually converted a video
from horizontal to vertical
in a few seconds
so know that you can
take advantage of this technology
not only on images but also
on videos, I still expect
that in the coming months this
technology will also be inserted in Adobe Premiere
Ok guys, I gave you some
advice, some techniques to
use Photoshop's generative expansion
better, let me know below in the comments
if you know others of these techniques
of these tricks
leave a nice like if you liked this video
subscribe to the channel
and see you in the next lesson