All language subtitles for 3. Optimizing Channels

In this clip, we're going to take a look at optimizing channels.

So the first definition that we're going to review is

chroma sampling and what exactly that is.

So digital signals are often compressed to save

transmission time and reduce file size.

Since the human visual system is more sensitive to

variations in brightness or luminance than color,

video signals can be optimized by devoting more bandwidth to the luminance or Y,

than the color difference, which is the Cb,

Cr.

So we're going to take a look at the YCbCr color space,

and those are the color space that digital footage is captured,

or at least it works in.

It's a little bit different than the linear color space that

we're used to working with inside of Nuke.

The idea behind this is that we can optimize the channels,

and this works with all the other ones that we're going

to be taking a look at in these clips.

So let's take a look at optimizing some channels and take a

look at an example of this compression.

So here I have an EXR and a JPEG,

and you might have noticed in other modules and clips,

I am working primarily with EXR files and I try to stay away from JPEGs,

with the exception of just screen capture because they're much lighter.

And there's a reason for that.

The reason for that is it contains far less color information.

So let's take a look at the metadata.

So here I have a 16-bit half float image with the EXR.

So it contains thousands and thousands of different pieces of color information.

And over here I have a JPEG, which is 8-bit,

which contains drastically less information than the EXR.

So what's the benefit to both of those?

Well, as you can imagine, the EXR has more information.

Therefore, it's going to produce a better and higher quality image.

The EXR can contain values that are above one,

or color values,

I should say; and the JPEG reduces those values between zero and one.

So when we're working with images inside of Nuke,

it's highly recommended that you stick to better quality file

formats to really reduce on that image compression.

Now I've used a JPEG here just because it's an extreme

example of what you can see in compression.

And before I move on, let's take a look at that in detail.

So here's my EXR,

and I'm going to zoom into my image here and compare it to the JPEG.

And as you saw it full-screen, or at least zoomed out,

it's hard to see the difference.

But zoomed in, you could start seeing the slight variation.

It looks almost a little bit more blurred in the JPEG versus the EXR.

So EXR, and there's the JPEG.

So now let's take a look at how we can reduce those.

And I'm going to take a look at the blue channel one last

time so we can see that difference.

So there we go.

So JPEG, EXR, and again it's more blurred.

Okay, moving on.

So on to the denoise.

We're going to denoise the image.

So that JPEG does blur ever so slightly.

This is the default denoise.

Feel free to optimize it.

And then we're going to move on to the YCbCr color space.

So we've added a color space node,

and we take it from our linear color space to the YCbCr.

And let's go full color.

So this was the default image in linear.

And this is the YCbCr.

And it looks kind of funny.

And typically, we don't work in this color space,

but it can be really great for keying purposes and

for optimizing those channels.

So let's take a look at those individual channels.

So red represents the luminance in this case.

The green and the blue represent the chrominance or color channels.

Now we know that there's more compression in the chrominance,

so we're going to slightly blur those particular channels so

that the artifacting isn't as noticeable.

So with the Blur tool attached to this particular image,

so let me go full color.

Oops, went to the wrong one.

Let's go back to the YCbCr.

And I'm going to add a blur.

So here we have a blur node.

By default, the RGB is turned on.

Turn off the red,

we don't want to attach the luminance because our

eyes are the most sensitive to it, but we are going to change the size.

So keep the size to two to four.

We don't want to push it more than we have to,

just so we retain the color information or the detail in

the information and the RGB linear space.

Now why would we want to do this?

Why would we want to blur the image if essentially the RGB looks okay?

And there isn't much variation between the JPEG and the EXR.

Well, unfortunately, keyers see big differences in the chrominance,

because they're chroma keyers.

Most of them are chrominance keyers.

So they're going to pick up the jagginess or the

stepping that happens in that artifacting.

So back to where we were at.

So here's the blur.

We're going to round trip back to where we were.

So YCbCr.

And we've gone back to where we were to begin with, so let's just take a look.

So here is the denoise, and here is the round-trip color space.

So back to where we were.

But let's zoom in to see what it has done.

So before, after.

And there's small changes again.

Let's go to that blue channel.

You can see, much larger differences in the blue channel.

So now let's see what that does when we pull our keys.

So I have two keyers here, and I've pulled a keylight key.

And I'm going to shuffle out what the non-blurred image looks like.

So here's the default key.

And that is with the denoise footage, so that's where we were at,

and here is the key when we blur the chrominance.

So much better than it was before.

Drastically better.

In fact, if I turn off the noise, it'll be even more apparent.

So here is that artifacting that we were talking about,

and its reduced or at least blurred.

So there it is, so there's the artifacting.

This is the reduced artifacting.

So a little bit better.

Not great.

Pump this up to four, and it's a little bit better.

So again, before, after.

And let's take a look at the RGB.

So here is the before, and zoom in.

And the after.

So we do lose a little bit of information in the RGB,

but look at the difference in those edges.

So much better than it was before.

Remember that all images or all digital images have some sort of compression.

The JPEG example is an extreme example of this,

but this is a really great way to add to your pre-processing workflow.