All language subtitles for 11. Overview of pgBokeh

in this lesson

we're going to learn how to use PG bouquet

to create realistic depth field

some of you may be aware of PG Bouquet and what it is

is it's a plug in that does depth the field or to focus

and it's real claim to fame

and why it's popular with studios

is that it supports deep

and as far as I'm aware

it's still the only

depth of field plug in that supports deep

and it's also extremely physically accurate

unlike most tools

things like defocus

or Z defocus

or any of the other techniques you can use

to basically just blur an image

it uses algorithms that are designed to actually

simulate the way that light travels through a lens

and the kind of optical artifacts

that you would actually produce

so that you end up with a result that's very

photographic

very optically realistic

it tends to be a little slow and in certain cases

it can actually be a little bit unstable

but if you're looking for accuracy and quality

especially if you can take the time to use it

afford the uh

the computation time

the results can be really

really beautiful

so let's dive into it

alright so we've got our

our shot again

so let's start with making the PG Boken node

so we've got a bunch of inputs

we'll take a look

input kernel camera and deep input

then here additionally we have mats

if you wanted to start putting mats in

mats would be just the same thing as masks

in a normal nuke note

so I kind of walk you through this

input is going to be your regular input

your beauty

you know your beauty pass

your primary image uh

something that is important to note is that

if your deep data is also in your primary input pipe

it can accept it through this just fine

uh otherwise you'd have to plug it into the deep input

if there was just deep by itself

uh the camera

the camera is something that the uh

the camera simulation aspect

of this plugin is really driven by a nuke camera

and I'll I'll explain how that works

and then the kernel

if you're not familiar with how image kernels work

basically when you think about

uh bouquet the

the actual optical effect

you'll get things like um

uh the shape of the Iris in a defocused photograph

well the shape of that Iris is what's referred to

in image processing terms as the kernel

and to be more technical

it would be

if you were to try to recreate that in image processing

which we are

in this case

it would be referred to as a convolution kernel

because convolution is the type of

blurring that we're doing

we're actually blurring it based on

the shape of an image

not just averaging pixels together

based on how a normal blur node

like a gaustian blur

just takes neighboring pixels

and that averages them together

in the case of a convolusion operation

you're actually blurring the image based on a shape

and that shape would be the kernel input

we're not going to use a kernel in this case

but I just thought I'd explain it

in case you wanted to try and match something

cause if you have a very specific look

in a project that you're working on

and you can see very clearly as a certain shape

when you have the focused image and

and you want uh

the focused areas of your visual effect shots

to match that exactly

then you would definitely try to use a kernel shape

based on a reference image from a

actual out of focus area

from the real lens

let's start off by plugging this guy in

input actually set this to

if we take a quick look

to which we still have deep data

yes we do

the problem here is I am not merging to deep data

so let's do all

quick tip is

if you do two brackets in the label value space

and then the name of any knob

any knob at all

in this case we want also merge

so let's just grab the deep

so it'll be more specific in the label just deep

if you just hover the

the cursor over any knob it will give you the name

B box meta input

and this one is also underscore merge

so also underscore merge

so inside a brackets value space

also underscore merge

and the result of course

as we get deep in the label here automatically

and if we change this it changes

so this is a little bit of a

nuke TCL or tickle expression that I

I use all the time because I like to label things

because now I know that at a glance

this is not just RGBA

it's also RGBA deep

so that I can plug in the input here

because as deep data requires deep input

and we need to now make the camera

go ahead and load in just like last time

read in our Olympic camera yes

okay so one thing that we're gonna do

after we hook up this camera

is that we've read in these values

actually gonna uncheck read from file

so now we're not reading from file anymore

cause we have read from file checked

you can see that these knobs are grayed out

we can't adjust these at all

and we're gonna want to

is we're gonna want to manipulate things

in order to control the plug in

let's take a look at the result the plugin is giving us

okay so this case these default settings

not really seeing anything happening

the first reason why is depth channel

the fault is depth dot Z

we don't wanna use the depth channel

we're gonna use deep so choose deep dot front

okay so now it's starting to render this frame

we still don't see much of an issue happening

so let's switch this up

integrate front pack separately

Wanna check this for accuracy

let's go to lens

okay first things first

so now you can see it's still in the default mode

where it's basically just faking it

what we wanna do is check real world lens simulation

we run to the frame

immediately we see a huge difference

so the camera film back

that we're rendering with here is full frame

full frame still just 36 millimeters

horizontal by 24 millimeters vertical

so let's go ahead

set this to custom

36 24 to match the camera

these aren't linked unfortunately

so now what we're getting

with an f stop value of 1.4 is really

really extreme depth of field

this is not something that I would normally pick

let's back off a little bit to something like maybe F4

that's a little bit more manageable

let's take a look at what maybe F8 looks like

okay and then F16

so if you're familiar with photography

have you ever done any photography

you'll know that if you are using a real camera

this is the result you'd expect to get

this is how a real lens would behave

and that's really one of my favorite parts about Peachy

Bouquet is that you can get very

realistic results using realistic methods

so you can apply what you know about photography

directly to what you know about compositing

let's just go back down to F4

what I was doing there in case

case you didn't know

is that you can do little operations division

multiplication right in a knob

and when you press enter it will do the math

and the result will actually become the new value

we have now reached the end of the lesson

we Learned how to use Peragon Labs Bouquet plugin

to create a realistic depth of field

in the next lesson

we will learn how to set the focal point in PG bouquet