a season and visuals on average
contains nine million frames a frame
and every female knows this frank video
this is a trophy this is another one
frank from netflix this year spends 8
billions to produce a type 700 stuff
original content and more
you will have to spend some marketing stuff
like 2 billion and in front of all this
I stopped lard content
the question that one does and how he does it
netflix to determine what are the
right images to show
users when they look and move in the
platforms to entice them to watch
more what is the right thumbnail thing
he laughs sting puts himself there and one by one
of cefla put the crisis size tour a
nora wait let’s put this one on
creative as they do but if you sir why
this is a key to show you in
continuation more content more I look
more rest hooked to the platform
then they wrote this article on the
technical blog of netflix that is a lot
interesting because you practically
discover that they have developed at home
a whole series of algorithms to do
two very interesting crafts to write down
and classify each frame that each
content that’s on netflix works
this one system that is called a vaffa
two things the first is annotations
the annotation means that they blend away
inside all the videos in this grinder
and he starts tagging so let’s put it this way
every frame of each video according to some
Parameters first parameter the so-called
visual metadata let’s say visual data
the algorithm that is super advanced of
computer vision to recognize what
there is inside a video goes to see
the color shows
the brightness of that video if
the blurred image or not the second one
parameter that looks and the context is there
a face inside this frame yes or no
it’s the yes or no movement on that one
sentiment analysis dirty word to say
that I can understand if
the happy or sad expression or that
that is what is the context there are objects
in the frame we can not third
parameter is the composition to take
the ideal laminating image must
understand if done with a composition
right in photography or you do not have
present when that is the rule of a
third, that you are all here on one side beautiful
reached the other space as well or
you’re in the middle of which one is composition
once they have all this data go
to classify image by image
according to a series of parameters that ad
example the main actor is present
inside the image yes or no what it is
right or there was an expression a bit
some pip unknowns so they have to
identify the right one
and crazy that it’s all an algorithm
that does this job there is not a human
which also does what it would not be
possible with all the frames as you do
analyze them all then everything is not of
other parameters and then go to see one
sort of maturity index for which second
the guidelines do not include any
images that have sex scenes of
particular violence or others that do not
respond to media guidelines and
logos that can not use texts
of brands that maybe can not
just use it once that
the algorithm is blended all stop to
sauce or at that point pass the top of the
top the cream of the cream on the ward
creative at that point the creative of
turn and chavez these here are a thick
the ten best images and on those
I’m going to work I wooed them for
good to go out with the thumbnail
perfect that then comes to bits was
issued in comparison with the title
to see the one that works the most and
when a user opens net these are there
also nice ginetta estimate and I always go
more to take the image than then
performs better ie it keeps you more
attached to see your SLR
how it is said that a picture is worth more
a thousand words in netflix cases
a miniature image is worth a beautiful one
$ 100 billion valuation
contains nine million frames a frame
and every female knows this frank video
this is a trophy this is another one
frank from netflix this year spends 8
billions to produce a type 700 stuff
original content and more
you will have to spend some marketing stuff
like 2 billion and in front of all this
I stopped lard content
the question that one does and how he does it
netflix to determine what are the
right images to show
users when they look and move in the
platforms to entice them to watch
more what is the right thumbnail thing
he laughs sting puts himself there and one by one
of cefla put the crisis size tour a
nora wait let’s put this one on
creative as they do but if you sir why
this is a key to show you in
continuation more content more I look
more rest hooked to the platform
then they wrote this article on the
technical blog of netflix that is a lot
interesting because you practically
discover that they have developed at home
a whole series of algorithms to do
two very interesting crafts to write down
and classify each frame that each
content that’s on netflix works
this one system that is called a vaffa
two things the first is annotations
the annotation means that they blend away
inside all the videos in this grinder
and he starts tagging so let’s put it this way
every frame of each video according to some
Parameters first parameter the so-called
visual metadata let’s say visual data
the algorithm that is super advanced of
computer vision to recognize what
there is inside a video goes to see
the color shows
the brightness of that video if
the blurred image or not the second one
parameter that looks and the context is there
a face inside this frame yes or no
it’s the yes or no movement on that one
sentiment analysis dirty word to say
that I can understand if
the happy or sad expression or that
that is what is the context there are objects
in the frame we can not third
parameter is the composition to take
the ideal laminating image must
understand if done with a composition
right in photography or you do not have
present when that is the rule of a
third, that you are all here on one side beautiful
reached the other space as well or
you’re in the middle of which one is composition
once they have all this data go
to classify image by image
according to a series of parameters that ad
example the main actor is present
inside the image yes or no what it is
right or there was an expression a bit
some pip unknowns so they have to
identify the right one
and crazy that it’s all an algorithm
that does this job there is not a human
which also does what it would not be
possible with all the frames as you do
analyze them all then everything is not of
other parameters and then go to see one
sort of maturity index for which second
the guidelines do not include any
images that have sex scenes of
particular violence or others that do not
respond to media guidelines and
logos that can not use texts
of brands that maybe can not
just use it once that
the algorithm is blended all stop to
sauce or at that point pass the top of the
top the cream of the cream on the ward
creative at that point the creative of
turn and chavez these here are a thick
the ten best images and on those
I’m going to work I wooed them for
good to go out with the thumbnail
perfect that then comes to bits was
issued in comparison with the title
to see the one that works the most and
when a user opens net these are there
also nice ginetta estimate and I always go
more to take the image than then
performs better ie it keeps you more
attached to see your SLR
how it is said that a picture is worth more
a thousand words in netflix cases
a miniature image is worth a beautiful one
$ 100 billion valuation
Please follow and like us: