All language subtitles for nuk231-class10

hello and welcome back to Nuke 2 3 1

Nuke composity best practices

this is Victor Perez

and in this last class of the course

which is class No. 10

we are going to discuss a few different topics

let's start by the use of 4K blades in a 2K warflow

is because you know

it's still a standard

to work in a film in a 2K resolution

but there are several cameras in the market

that are now offering the possibility

of shooting in higher resolutions okay

and that's good for us

for certain task

but not for everything

because working in a 4K

unless you have an super fast computer with tons of RAM

is going to be quite heavy for processing

and at the end of the day

very proudly

the film is going to be preacted in 2K

so if you are working on a 2K workflow

but you have the possibility to go to the 4K

that is going to be very handy for setting operations

mainly when the task is to extract information

so we are going to focus

first of all

on the camera tracker

because he's where the resolution actually matters

so the first thing we need to understand is

the 4K and the downscaling

2K have to be the exact aspect ratio of the image

so in this case

the aspect ratio is 4 by 3 okay

and here the 2K is 4 by 3

so if you see the images

at a certain distance

there is no difference between them

the difference comes when you feature

the full resolution

so I'm going to get closing here

and you are going to notice very

very clear from the 2K to the 4K okay

so the precision we have to point to certain corners

or blocks of the image

is exactly four times more precise

when you are using double the resolution

because it's double

behave by double the width

so you have four times the resolution

in a 4K than the 2K

so if you have the same aspect ratio

you can use the camera tracker

with no problem in the 4K

extract the camera

and then apply the camera for the 2K warflow

that is not going to change at all the camera

even if the camera is going to be calculated

in an image that is way bigger

at the end of the day

the camera is just taking in consideration

the features you see

in relation with the habiture of the camera

so the habiture for both is going to be the same

the only change is the amount of information

you are getting through the aperture

okay so when you are working with the camera tracker

I would recommend you to go to the highest resolution

but always check

that the aspect ratio of the format

and also the pixel

in case you're using anamorphic

must be exactly the same for both images

okay of course

the precision that you have to get

any feature

is going to be way better

so the amount of error you are going to get

in the camera trucker

intensive results

I mean numbers

so the error you getting here

probably is going to be more or less the same

but the amount of effective precision you get

position in those features

in the pixels

is going to be four time more

so when you apply

the error from the 4K to the 2K plate

the error is actually 1/4 of the original error

so you are reducing everything

okay so that is the first thing

the second is going to be

very clearly the patch extraction

for Rick removal

so for Rick removal

I mean anything that you need to cover using

maybe to the projections

or just simple pre multiply patch

to be moved into the so

is exactly the same process

but the same

when you are feeding

camera projection

is way better to feed the projection

using the 4 key

okay even if

at the end of your projection set up

your scan line

render is going to be set at 2K

okay but if you feed

the camera projection

with the 4K

the precision

for the calculation of the anti allius

and the filter kick

is going to be again 1/4

from using the original 2K

so I just preparing here an example

and this can be apply

for 2D or 3D

is exactly the same principle

okay so in here

I have my 2K image

and then what I did

is just a very simple wrote on

let me close this

just to get

one of the windows mask

okay so then I move that window to the next one

just to cover this one

so you are going to see like the windows have no wipes

okay so if I put

that over using the original 2K to use the same patch

you will see that the difference between this

which is the original and this

which is the image I use for covering

is having a huge impact in the resolution

because of course

he's losing due to the filter

for the antialius of that transformation

okay here I'm using a corner ping

which is the same effect of applying a car in 3D

okay so of course

you can just change in here the filter

and that is going to create maybe a sharpening effect

but remember

sharpening is going to add some issues

due to the negative values in very dark areas

or values that are not supposed to be that contrasting

in the edges between the highlights and the dark areas

okay so using sharpening or sharpeners filters

like for instance Keys

Simon and Riffman

they are sharpeners

they are going to add certain amount of

sharpening effect

for instance

in here that is an artifact

so probably is going to look nice

but is not exactly

as the original because in the regional

there is no artifacts like in here

okay so I'm going to return to the cubic

because I have another example in here

where I was using the 4K for the patch extraction

I can hear and then I use the corner pin

to reduce that image

and to place it in there

so my background plate is always a 2K

and this is the result

okay so as you can see in here

the resolution is quite the same

because the resolution I got in the very beginning

even after the filter

is enough to recreate the same amount of anti alias

as the 2K has

okay so let me show you

the 4K against the 2K

so see there is a massive degeneration of those pixels

okay so by using the 4K

you are going to decrease by a lot

the amount of filtering

and remember that filtering

when you are using Rick removal

is one of the main issues

okay so those two case scenarios

can be expand to any other cases

where the filter matters

okay so I left you test those okay

now we are going to work with the real lighting

because in the real lighting

we have a very similar issue

which is related

with the calculations of the anti alias

okay so I'm going to deactivate those two notes

let me just run through the setup

so you have an overview of what I have

so in here what I did is just to put together the RGBA

with the position pass and the normal pass

remember that you need

those three components for the relighting

and then in here

I converted that into 3D images

so I can have a visualization of my light even in 3D

then I got my camera

so in case I want to see the image from my camera

okay I have this image

and then of course

the light and the relight

that is having the basic material

just for creating the dynamics of the surface

that I relighting okay

so if I look at this in the real light

you will see that

is doing what is suspected to be doing

but there is a little problem

that I'm going to show you very clear in here

for instance

look at that

so we are working with a pixel precision calculation

but there is no subsembling in the pixels

so a pixel can be in a certain position

and in a certain angle

so the position is related with the position pass

and the angle of that surface in that precise pixel

is given by the normal pass

but when you have different surfaces

in one pixel and another

that is going to create this strange alias effect

okay and it's totally normal

but in the real light

you don't have any Antilias

or any filter to change that

okay so how can we add that extra filter

well we are going to create soup pixels

so what we are going to do is just to divide a pixel

into a certain amount of soup samples okay

how to do that

well it's actually very easy

it's going to increase the amount of computing

of your image

but he's not that dramatic

okay so what I'm doing here is

I'm getting this reformat scale

and I want to scale this image by a proportional double

so you can get twice the scale

or four times the scale

or eight times the scale

so we are going in a progression of 2 4 8

16:32 and I mean

when you get 16 super pixels for every pixel

that is a lot right

now I'm going to do it just with the scale of two

so I'm going to create

for every pixel

four pixels

I mean two in width

two in half

okay so I'm going to apply this reformer

but very very

very important here is that you use the filter impulse

so that way

we are not creating any kind of antialias or filtering

we are just doubling the amount of pixels okay

and that's why we are using that double

or four times

or eight times

because we are just getting double the amount of pixels

without creating intermediate pixels

I mean positions that are between two different pixels

okay so now

everything is regularly and arametically double

okay so if this is a pixel

for instance

when I increase

the reformats

now that pixel

has been divided into four pixels

but still the appearance is that the Pixel

is exactly the same

another reason why you cannot use any filter

is because you are going to create filtering also

on your Aobs

for the position

and the normal

so that is going to create a mess

in terms of the positioning of those pixels

so there is no other filter you can apply here

so impulse okay

and because we are doing double

I don't care about the width

or the center

is going to be everything at the same scale okay

so now that I have that

the relight

is going to be actually the same

so we are going to have the same problems before

but every pixel

is now for pixels

so if I create in the beginning

double the amount of pixels

now in here

I can downscale everything

just to compress all those pixels

into one single pixel

so what I'm going to do

is just to force nuke

to create a filter

so every time

is going to calculate

every single pixel in here

is going to be

calculated with the surrounding pixels

depending on the filter we are going to use

so if in here

I use double

in here at the end

I'm going to use half

if you use four times

then in here

you have to use a quarter

so is everything proportional to the original scale

okay so in here

I'm going to apply

the filter notch

which is the softest

that you have in here

so when I reduce that

and I put my beauty in here

you will see that now

what was originally

something without anti alias

is now very well alias

see of course

is increasing also

the feeling of softness

okay so that is not great

so in case you don't want to have that soft

you can't use any other

filtering here

remember that

softeners in here

you have the Mitchell

person and notch

notch is the most

then person is in the middle

and Mitchell is doing

kind of a soften

with a sharpen effect

okay so here

is not changing much

in here is increasing

but I can feel still the original pixels

and then with notch is adding that extra softness

okay of course

you can also add the soften filter in here

or any other post processing that you prefer okay

sometimes is going to be really good

that you use also the ambient occlusion

because the ambient occlusion

is not going to be processed by the relight

but is going to add that tiny edge of the curve

that is going to reduce the effect

of overlapping differences of plane

so in here if I apply that see

is going to help you a bit with those areas

and also to recreate the effect of sharpening

in terms of the original resolution of the image okay

but this of course

is going to depend on the effect

that you are aiming for

okay so in here

I'm going to leave it at full

that is the way of creating antiallias

of course you can create it in here

in the relight

because the relight has no antiallis

but you can create it even in other case scenario

like for instance

in a skyline render

you can feed a format that is twice

the hate and the width

so when you get the original camera projection

and then you apply the lens distortion

remember that the lens distortion is a note

that is actually creating a huge amount of filtering

okay so if you create your CG

coming from the skyline render

in double the resolution

and then you apply the lens extortion

and then you apply the original format

by using another reformat

then we are talking exactly

the same thing as we did in here

so I'm using a patch

that is having way more resolution than the original

so then when you manipulate that image

and then you start losing quality because of the filter

that is something natural

then that feature is going to be reduced by a lot

because of the resolution issue

okay so when you have more

you're going to lose less

of course this is just twice

but you can increase that

and use four times

eight times

honestly if you use more than eight times

is going to be a very

very long process

and is not going to add that much amount of quality

okay and now that we are in the relighting setup

let me just tell you that

from the relight

you get this

okay so just

just get a light over the geometry

seal the geometry

let me remove this for a sec

when you add that

to the original image

you are going to lose

all the original qualities of the light

and you're going to lose quite a lot of information

even if I'm doing that supply

so something that I like to do

is to do a pre fusion

okay so what I'm doing is mixing the lighting

with the original plate

so that is going to create this intermediate state

that is going to react

to the original patterns of texture

of the image

okay so here

you can see

that is going to be emphasized in certain areas

and not much

inserting other areas

so now when you have this pre process

you can apply that in here

and that is the final result

so you're not going to lose the original lighting

but you are just adding those extra

lighting effects

okay so that is double fusion

and another example of double fusion

is when you want to apply to a certain image

a little patch

with another operation

that is quite popular

for instance this

um what if I want to multiply that

so the white

is going to be completely pre multiply

and the rest

is just going to be using the texture

so I'm going to use multiply

but that is going to create a little problem

so see I'm preserving

the areas of the text

or even in here

see that the text

or the pattern

is still there

that's very nice

but I'm losing

actually the rest of the image is

because I'm multiplying these

that is black outside

and nothing

just the white in there

how come we fix that

well we are going to limit

the action of my merge

just inside the alpha

so he's going to have the merge

inside the alpha

and outside the alpha

is just going to be the image

from the B side of the input

okay so I'm going to use

for doing this clear

and in this case I'm going to put in there

so what's the option for that well

alpha masking

alpha masking is going to apply this operation

just within the alpha

outside is going to be the being put

and here you are

so see I'm still using the pattern from the background

of course now is not integrated

and is not my intention to integrate that

but just to show you how to use that fusion

and actually

this is exactly the same as doing this other thing

so you can have an idea of what I'm doing

so if I remove the alpha masking

I'm actually doing a double fusion

the double fusion is this

so I'm getting that that is primultify

because this image has an alpha

and now what I'm doing is just putting this

over that

so here you are

is exactly the same result

but let's note in here

so you are going to limit using the alpha masking

of course you need to get an alpha in this image

and that is going to delimitate

not only the transparency

but also the area of effect of the operation

that remember

for a multiply

usually you don't need an alpha

but if you don't have the alpha

that is just going to prevent the alpha masking to work

okay so that is for the double fusion

inside one single note

in here in the relight

we have double fusion with two different notes okay

and then the last thing we are going to discuss

is how to change several notes

with just one single operation

because sometimes there are certain tasks

that requires that many notes have the same value

for instance

when you are creating an axis from a point

in a camera tracker note

so when you want to use a cone in there

you want that axis to be created without any rotation

for instance

or just for the sake of doing an example in here

I'm going to use different

blow notes okay

so I just put in here the receipt

so you just need to copy and paste

but I want you to understand

what's the process that is being executing here

so for that again

I'm going to call the

script editor

okay so let's delete that

I'm going to clear

and now I'm going to copy and paste

the contents of there

so I put this note for you

so you understand

what needs to be accomplished before to apply this

so I'm going to select my notes

let me open the properties of one to show you something

and I'm going to select my notes

remember in here you get the knob

what's the knob you want to change

and the knob

can be revealed by hovering with the cursor

and then that name in bold

that you get inside that pop up window

is the name of the knob we want to change

so in case of the size is the same as the label

but not all of them is the same as the label

so I'm going to change the size

so in here I'm going to write

size then what's the value you want to put

let's say that in this case

I want to change all my blowers to have a value of 100

in case you are using a number

usually you don't need to get they quote

okay actually

if you put the quote

is going to be intended to be an string

so a chain of text

so it's not a number

is just a text

so 13 knobs are expecting you to put a No.

13 others are expecting you to put a text

so when it's a text

you put the text inside quotes

when is a number without quotes

okay so that is my target value

and then in here is just as very simple operation

which is a loop

a 4 in loop

and this operation is saying

for every element in the list

that I create by selecting my note

so every note in this selection

I'm going to take the knob call size

and I'm going to change its value

to the value I put in here

which is 100

so if I select those notes

and now I execute that

by doing that

as you can see

this note is now 100

and not only that

also this and this

and this and this

so when you have to change any knob

inside any amount of notes

you just go and select all of them

be careful because in here

I just create something very genetical

and very easy for you

but you just need to be careful

that all the notes you selected

should be the same class of notes

I mean if you are changing blurs

you select only blurs

if you are changing color correct

you just change color correct

the same for grades

but don't mix them

because for instance

in the color correct

you don't have any size knob

so if you apply this

and you ask nuke to go to every single note

and change the size

he's going to return an arrow

because there is no size in the car correct

so just pay attention to that

if you want to create a more articulate

python formula

you can but then you need to write

and for writing Python

I will suggest you to understand

all the fundamentals of the code in relation with nuke

as I mentioned in another class

I already publish a course for Python for

new compositors

that is here at FX PhD

okay so this is the last topic

regarding techniques for best practices in terms of

composting okay

but some of you were asking me what tools

or gizmos or plugins I like

well honestly

I prefer to create my own

but when you find something that is really

really good

of course you are free to use it um

instead of just giving you a list of my top 10

or whatever

I'm just going to recommend you

to go to the main place for finding the information

which is Wikipedia

of course in Wikipedia

there is many

many people collaborating everyday

and sharing their scripts

tips articles

and every piece of information that is really

really good

but they are two names of people that I really like

and I suggest you to follow

because they are always publishing

very interesting things

of course they are many others

but my favorites are

Frank Ruther

and Ivam Buzquets

they are amazing artists and very

very good at creating gizmos and

plugins and everything

so I'm going to give you a few examples

of things that I created myself

for instance

and things that others created

okay but you know

it's impossible to

to give you a list of what's the best

because depending on what you need

you have many options okay

so the first I'm going to give you is

you know in here

I paid a lot of attention

to present your scripts

and the note graph

in a correct way

okay so Frank

has published this aligned notes tool

that is helping you to get

everything in a good shape

quite easily

and you can even program a hot key for

aligning the notes

so he's going to increase your

speed when you are

putting order

into your nongraph okay

very very recommended

then from Frank

also we have this nice tool

which is a very

very nice tool

for analysing values

and understanding

the relation between the curves

of the color

okay so this is very nice

for compositing

to understand something

like contrast

frequency or

differences between channels

okay I use it a lot

then from myself

you have three tools

that I consider

the most important

which is preset backdrop

which is in here

to create the backdrops

with the icons

I already shared that with you

but in case you want

the updates

or also you can visit this page

on Wikipedia

and remember

all the links

that I'm using here

are in the class files

for this class

okay then I have an edge mat

and this is something I did

with the exact same principle

of the Luma key

you know we have our next

detector in nuke

that is very good for

finding the

edge of the alpha

when you need to

isolate the edges

just to work

with that portion of the image

but is not that precise

and is not offering you much

versatility to work

so I create

my own tool

for creating

the edge okay

so based on

any channel

is just going to display the edge

and you can

work with that edge

internal external

and many other options okay

even masking

this is something that I use a lot

then the vis late

this is mainly for pipeline

he's going to help you to keep the track

of your renders

and even to present for the client

so he's going to give you a frame

with all the information

and if you want

is going to give you born ins with the overlay

some format aspects

and many other options

of course if you are handy with Python

you can just ultimate the process of slating the plate

so everything is going to be

not only beautiful and useful

but also fast

but this is more towards the composting

technical director role

because there are many lines of

called just to create automatic slating okay

and also of course

the naming convention

that every shot should respect for getting this okay

but even if you are working in a Manuel fashion

just for presenting

your shot is actually very good

okay then in terms of fair party plugins you know

I consider that everything that needs to be done

can be done just with the use of all notes

that you have in here

so I'm not a big fan of plugins

but there is one special plugin that I use a lot

which is the optical flares from Video Co Pilot

I think this is all for this course

so I hope you really enjoy

and you learn new things

and it helps you to improve

the quality of your compositing

and the quality of your life

when you are working as a composer

it has been a pleasure

to have you on board of this course

and I hope to see you in the future

in other course still

if you have any comments or questions

publish them in the forums

so until the next time

this has been as always

Victor Press for FX PhD

thanks for watching