All language subtitles for 2. De-noise vs. De-graining Green Screen footage

af Afrikaans
ak Akan
sq Albanian
am Amharic
hy Armenian
az Azerbaijani
eu Basque
be Belarusian
bem Bemba
bn Bengali
bh Bihari
bs Bosnian
br Breton
bg Bulgarian
km Cambodian
ca Catalan
ceb Cebuano
chr Cherokee
ny Chichewa
zh-CN Chinese (Simplified)
zh-TW Chinese (Traditional)
co Corsican
hr Croatian
cs Czech
da Danish
nl Dutch
en English
eo Esperanto
et Estonian
ee Ewe
fo Faroese
tl Filipino
fi Finnish
fr French
fy Frisian
gaa Ga
gl Galician
ka Georgian
de German
el Greek
gn Guarani
gu Gujarati
ht Haitian Creole
ha Hausa
haw Hawaiian
iw Hebrew
hi Hindi
hmn Hmong
hu Hungarian
is Icelandic
ig Igbo
id Indonesian
ia Interlingua
ga Irish
it Italian
ja Japanese
jw Javanese
kn Kannada
kk Kazakh
rw Kinyarwanda
rn Kirundi
kg Kongo
ko Korean
kri Krio (Sierra Leone)
ku Kurdish
ckb Kurdish (Soranรฎ)
ky Kyrgyz
lo Laothian
la Latin
lv Latvian
ln Lingala
lt Lithuanian
loz Lozi
lg Luganda
ach Luo
lb Luxembourgish
mk Macedonian
mg Malagasy
ms Malay
ml Malayalam
mt Maltese
mi Maori
mr Marathi
mfe Mauritian Creole
mo Moldavian
mn Mongolian
my Myanmar (Burmese)
sr-ME Montenegrin
ne Nepali
pcm Nigerian Pidgin
nso Northern Sotho
no Norwegian
nn Norwegian (Nynorsk)
oc Occitan
or Oriya
om Oromo
ps Pashto
fa Persian
pl Polish
pt-BR Portuguese (Brazil)
pt Portuguese (Portugal)
pa Punjabi
qu Quechua
ro Romanian
rm Romansh
nyn Runyakitara
ru Russian
sm Samoan
gd Scots Gaelic
sr Serbian
sh Serbo-Croatian
st Sesotho
tn Setswana
crs Seychellois Creole
sn Shona
sd Sindhi
si Sinhalese
sk Slovak
sl Slovenian
so Somali
es Spanish
es-419 Spanish (Latin American)
su Sundanese
sw Swahili
sv Swedish
tg Tajik
ta Tamil
tt Tatar
te Telugu
th Thai
ti Tigrinya
to Tonga
lua Tshiluba
tum Tumbuka
tr Turkish
tk Turkmen
tw Twi
ug Uighur
uk Ukrainian
ur Urdu
uz Uzbek
vi Vietnamese
cy Welsh
wo Wolof
xh Xhosa
yi Yiddish
yo Yoruba
zu Zulu

Original subtitles

In this clip,

we are going to take a look at denoise versus degrain for green screen footage.

In previous clips and modules,

we looked at ways to analyze the footage to determine the areas of interest.

In this module,

we will review ways to denoise the plate as a really great way

to preprocess the image before pulling keys.

Before we begin, we're going to take a look at a couple of different definitions.

The first one is the definition of grain.

So what exactly is grain?

Well,

film grain or granularity is the random optical texture of

process photographic film due to the presence of small

particles of metallic silver, or dye clouds,

developed from silver halide that have received enough photons.

Or in other words,

basically grain is found in film and it is caused by

small pieces of metallic silver.

Note different types of film stock may vary in size and

amounts of this particular silver.

So then what is the definition of digital noise?

Well, most noise in the image sensor is thermal or heat noise.

It's created by electrons being thrown around by vibrations.

Image noise is an undesired byproduct of image capture

that adds random and extra information.

The original meaning of noise was and still remains an unwanted signal.

So in other words, noise can be found in both film and digital footage.

And if we take a look at what are some of those causes,

one is image compression, filters and plug-ins.

So now that we have an idea of denoise basics,

let's a look at how to denoise the plate.

So back to our script.

We're going to scroll down to our first piece of footage.

So I have hooked up the following tools.

Each one of these tools is hooked up to this piece of blue screen footage.

Now before we really begin, I do want to make a note that often,

the words denoise and degrain are used interchangeably.

This is technically incorrect,

but a lot of us that worked with film stock previously do

sometimes mistakenly use that incorrect term.

So in most cases, we are referring to digital footage.

And if you do hear the words used interchangeably in this clips and modules,

I am referring to the digital footage that we are using

for these particular clips and modules.

So let's take a look.

So the first one is that I've hooked it up to a blur node.

And if I go to the blue channel,

because that's where the noise is the most present,

so here we go, here is the blue channel,

and you can see that I have blurred the image but retained much detail.

In fact, if I try to sharpen it again, it looks even worse.

So clearly this is not the way to denoise the plate,

but this is the very basics of how a plate is denoised.

So we are trying to blur the image just enough to rid of that noise,

but not too much in order to retain the detail in the plate.

Now there are far superior workflows,

and we're going to take a look at those now.

So we're going to skip over these examples.

You can take a look at them in the script outside of this particular clip,

but let's take a look at FrameBlend.

So it is hooked up again to our image.

So let's take a look at our image, and I am referencing frame eight.

There's a reason for that, and that is that our character's hands start to move.

FrameBlend is best used for removing snow and rain and maybe

the odd dust or scratch in the footage.

It does not work well with images or cameras that are moving.

So we definitely want to have a locked plate and very little

movement other than that snow or the rain.

Let's see why.

So again, I mentioned that on frame eight there is some movement in the hand.

We're going to see here that we have a ghosting effect

and that is because the hand is moving.

So not only does it do not a great job with removing the noise,

but we also have doubling of our moving image.

So not a great idea for removing that noise.

But rain, snow, definitely a great workflow.

Let's move on to the next one, and that is the degrain simple.

So back to our footage and the degrain simple is exactly that, it's very simple.

We are basically blurring the individual channels.

The blue channel has been blurred more here just because

it tends to have more noise in it.

Now unfortunately,

we are still losing the details that we love to see when denoising the plate.

So let's take a look.

So here's the degrain simple,

and if I toggle back and forth from what we had before to the degrain simple,

it is doing a better job.

We're removing some of that noise,

but look at the artifacting that we are introducing in the plate.

So we're not retaining the nice detail in our image.

Instead,

we are adding elements that is going to work against us when pulling that key.

So I would try to stay away from the degrain simple.

It is a very fast tool.

So if you just need something as quick references,

something like that, feel free to use it.

But there are, by far, more superior tools to denoise the plate.

So speaking of denoise, let's go to the Denoise tool.

So back to our original image and I'm going to show you a

quick denoise that has already been pre-done.

So here is the denoise.

What a difference.

So much better than the other examples that we looked at so far.

And here is the original image and here is the denoised plate.

So much better detail in our plate.

In our edges, no doubling.

And we have removed the noise.

Now sometimes, the default values when we pull the denoise operation,

blur the plate too much or too little.

I would describe it as a guess is what it's trying to do with the automation.

So we're going to take a look at how to pull that denoise from

scratch and then how to modify the settings so that we can

retain as much detail as possible.

So scroll in on down to our brand-new Denoise tool.

Again, it's hooked up to our plate.

So here's our original plate.

I have taken the light and color slump up just a

little bit so we can see those details.

So let's zoom out,

go back to the regular lut and double-click on the Denoise tool.

Now this is a brand-new Denoise tool and it's telling us

that we have to move the analysis region.

And that's because it doesn't understand what it's supposed to look at.

So we have to tell it what to look at in order to

determine how much to denoise the plate.

So there's a couple of different things that you want to look out for.

The first one is an even surface.

So denoising over her hair probably won't work well.

It doesn't like patterns or uneven surfaces.

So something like the blue screen up here or the back of

the chair will be great areas to sample.

Now personally, I like to sample in darker areas,

and that's just because the grain is a lot more visible in those areas.

So again, I'm going to go and increase my lut so I can see more details,

and drag the square up to the back of the car seat.

So there we go.

The car seat or head rest in this case.

Let's take a look at the grain and look at the results.

So again, great first results with the automation.

It's looking pretty darn good overall, which is fantastic.

Let's take a look at what happens if I make this area too small.

It doesn't like that.

So it has to be at least 80 by 80 pixels.

I don't like to make it much bigger than it has to be to do that denoise.

So let's go look at how to refine those particular settings.

So I have a tool that has already been optimized.

So here I have the Denoise tool, and you will notice that it is outputting noise.

Previously it outputted the result.

There's a couple of other settings that I have revised here as well,

so we're going to look at all of them and figure out what they do.

So here is this tool here, and it has sampled it around the same area,

on the back of the car seat, or the head rest,

and the first thing is source digital.

So I do want to make sure that it is set to digital, not film.

We have filmed on digital, not film.

So sometimes you can switch between the two if it gives you a better result,

but I would leave it at digital for digital footage.

Next we have a noise model.

I would leave that at the default.

Profile, would also leave that at the default of constant.

And now we're at the output.

So the default is results, result would be the denoised plate.

And if I set it to noise, it will isolate the noise that was in the image.

So what would I use this for?

Well, sometimes we want to regrain or I guess add noise back to the plate,

and this allows us to have a denoise and isolate the

noise separately from the plate.

So let's take a look at that.

So here's the denoise with the noise separated,

and a denoise with the denoised plate.

So no noise in the plate.

And from a previous clip we went through,

the term round tripping, which means to bring it back to the original state,

so I've used, I have plus operators.

So I've plussed the noise over the top of the denoised plate,

and it reintroduces the noise back.

So let's just see if that's round tripped.

So before, that's before we denoised it; and after, which it has.

So you can isolate that noise.

I like to go back to the original plate so you'll

understand when we go into the despill process,

but this is also a great alternative workflow.

So back to the denoised image.

So we'll go back to the default one,

and let's take a look at the denoise amounts.

So the denoise amount is an estimated amount.

It's what the tool thinks that it needs to set that particular

slider to in order to denoise the plate enough.

Sometimes it blurs the image too much.

So you can bring it down just a little bit.

If you bring it too far, then it adds grain back.

So it hops of sorts.

So it looks like at least.

52 might be good for us, somewhere around.

52 and.

65.

I think it's probably a great setting for this particular image.

And now let's go on to the roll off.

So what is the roll off?

It will determine how soft the denoise will be.

And so one is the default.

You can bring that down, at least I think we can, let's see.

Not really making a difference, so I would leave that at the default of one.

Now let's take a look at the smoothness.

We're going to zoom into the plate a little bit further to see what that does.

So it defaults to one, and you only need to modify if,

for some reason, the denoise amount up here is not doing its job.

So it's not smoothing out the plate or blurring the

plate enough to get rid of that noise.

So if I bring the smoothness down,

you can see it starts to introduce more information to the plate.

And if I increase it, it pops.

So the default value was one, oops, strong one.

Let's take a look at that again.

So the default value was one.

And again, if I increase it, it almost softens it too much.

So we, again, don't want to go past where it kind of changes,

where we saw right there, so I'm going to change it to one,

which is a really nice setting.

So for the most part, we don't need to change those.

We can go back and modify them once we go through all the different

selections if we're not getting enough detail in our hair.

Sharpen.

I would not change sharpen.

The reason for that is it will add too much detail and actually ruin our image.

So here's sharpen.

It looks like it's making it more contrasty,

which you would think that we would like.

But let's go back to the color.

So here's the color, and let's turn off sharpen.

And it's actually adding more detail than we would like.

So there's my original image, there's my denoise,

but if I increase the sharpen, it's making it too contrasty.

This is actually destroying our image.

So leave that at zero.

If you do need it, just add a little bit.

It's better to add a little bit rather than too much

and go overboard and ruin the image.

Next slider is going to be the luminance blend and the chrominance blend.

So just a quick reminder, luminance is the brightness versus the darkness,

and chrominance represents color.

So the blend adds back the luminance and the chrominance.

In most cases, I would not change those settings.

But if you feel that you will improve your denoise,

by all means, please modify them until you get the right setting.

Down the bottom is tune channels,

and we have the luminance gain and the chrominance gain.

So those are going to control the values in the luminance and the chrominance.

So let's take a look at those down there.

So back to our blue channel.

Now I'm going to increase the gain.

And similar to what happened before, if we add too much, it will blur it.

If we add zero, we add back some of that noise.

So default of one is probably good,

but now I'm going to go into the chrominance gain,

see how we can add more detail back to the hair.

So bring it down.

Too far adds that noise back until it pops.

There we go.

Takes a little bit of time to update.

This is a heavy tool.

So here, I'm optimizing it.

Let's take a look at what the default settings are.

So we'll default back this one to, leave it with five.

And one for the denoise amount.

So here is before, and here is after.

So you'll notice that we still have no noise or very little noise.

We have nice detail back in our hair.

So it doesn't make a huge difference in the RGB.

However, what's important is those individual channels.

So I will stick with the one with a little bit more detail.

It is very subtle.

So if you do want to take the time, it is worth it to modify those settings,

but they will be very minimal in order to get slightly better results.

Now the last thing that I want take to look at was the Neat Video.

So this is a third-party plug-in.

It comes from neatvideo.

com.

So www.

neatvideo.

com.

And you can have a trial yourself, download it, see what you think.

A lot of studios really enjoy it and get some really great results,

although I will see the Denoise tool from Nuke 10

is working pretty great as well.

So I won't go through how to use the tool,

because they do have different videos on their website.

However, let's take a look at the results.

So it may take a while to load.

There we go.

Just wait for it to process.

And then our optimized one.

So slight differences.

Very little.

We can continue to tweak those values.

But again,

it's a great alternative to using the denoise if we can't

get a great result out of that tool.

Now I will note that it is worthwhile to pre-comp out a Denoise tool.

It can get quite heavy,

and we don't want it to constantly be processing when

we're doing our keying process.

So just write it out to an image sequence,

bring that image or new image sequence when you are pulling your keys.

Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.