All language subtitles for DaVinci-Resolve-17-Fusion-3DCompositinginFusion

Hi, my name is Matt.

And today we're going to talk about 3D
compositing infusion within DaVinci

Resolve.

The main difference between 3D
compositing and 2D compositing

is that we'll use Fusion's 3D environment
and a specific set of tools

that are designed
to work in that environment.

When you're compositing
live action footage in a 3D environment,

you'll
almost always need a virtual 3D camera

that replicates what was filmed on the day
with a real world camera.

That virtual 3D camera will allow us to
render

objects placed in the 3D environment
in a way that matches the real world

camera in terms of perspective,
camera, motion and depth of field.

So let's go ahead and look at the footage
and get started.

So we're playing this back
and what we want to add into this shot

is a pirate ship over here
and do a couple other refinements.

The first
thing I see when I look at this shot

is that the camera's moving through space,

and that tells me that this is not going
to end up being a 3D composite

because I need to recreate
that camera motion

for any of the objects
that I'm going to place in the scene.

So let's
go ahead and get to the fusion page

and get started.

The first thing I'm going to do
is I'm going to go ahead

and just go to a single viewer here
by clicking that button

and then I'm going to rename my media node
using the function

F two button
to bring up the rename pop up,

and I'll just name this element plate.

I'm going to need a 3D camera
to really do anything.

But even before getting the 3D camera,
I need to actually define

the areas that I'm going to track
to generate that 3D camera.

So when you're looking at the scene,
you want to find the areas

that are actually static objects and
that don't have any emotion on their own.

So we don't want to track these characters
here.

We don't want to track the woman,
we don't want to track the ocean

because they're all moving on their own.

We just want to track static objects then,
and that will give the camera tracker

the information it needs
to recreate this camera virtually.

So we'll

start off by defining those areas
that we want to track.

I'm going to go in there a polygon
by dragging it

in from the toolbar.

And the first thing I'm going to go ahead

and define is these rocks here.

So I'm not really defining

the tops of these palm trees here
because they're moving a bit in the wind.

And then as I click and drag this, I'm

just creating a nice smooth curve here.

I don't want any the ocean
because it's moving on its own.

And so that should work for those frames.

And then I'll just jump
to the end of the shot

and adjust this polygon a bit

by selecting all the points.

And now that I'm closer to these rocks,
this polygon is a bit too small.

So I'm going to hold down s and click
and drag and that'll allow me to scale

without having to actually

manipulate each of those individual points
themselves.

I can actually do a similar thing
with rotation, so I'll select

just these points and then I'll hold down
T for twist and click and drag,

and that allows me to rotate these points.

I can also scale an X
if I hold down X and click

and drag for Y and click and drag.

So that works for these rough frames,

maybe a little bit less on the y axis.

And let's play
back to see how this is looking.

Okay, so we're drifting a bit
and then the camera pan.

So I want to go to that point and adjust

my polygon here.

Now that seems to be

roughly tracking how I want it to.

Maybe one more keyframe in here,

and that's going to work for what I need.

Doesn't have to be exact.

I'm just defining this rough area
that I want the camera tracker to track.

So I'm going to go ahead
and name this tool

by right clicking, then go to rename

and I'm just going to name
that Poly Rocks.

I'm going to add another Polygon tool.

Instead of dragging in from the toolbar,
I'm going to hit shift space

and then type in Polygon into my tool
search,

pop up.

Now I'm going to define the beach

a little bit.

Overlap with the rocks

right?

Just my spot in here. But

then I won't get too close to the water
because I don't want to track the water.

Okay, I'll jump to the last frame.

Just this accordingly.

Bit of scale here.

Since we're closer

and a better rotation with t click drag.

All right.

And let's see how that plays.

Oh, maybe add another keyframe

more in the middle here to retain a bit

more of this beach.

Okay, that'll work.

So we're looking pretty good here.

The only issues we have with
this is this character here runs

across one of these polygons,

and then these characters
are running across the rocks.

So I'm going to do
some additional polygons to make sure

that they're not tracked.

So let's
go ahead and start with the woman here.

And I'm going to add a polygon with shift
space to bring up the tool search.

And then I'm already preselected
on Polygon.

So I'll click add

and I'll draw a row rough

polygon spline around her.

And instead of

adding to these other shapes, I'm
going to go ahead

and change my paint mode
in the inspector to subtract.

And let's go ahead and view this tree
real quick.

You can see here
that this polygon is actually subtracting

from the other two polygons
that I originally added.

Go back to the plate

and me play forward
and add a few keyframes

for this character.

Again, I'm

going to use s click drag to scale

and a little bit of additional refinement.

You don't have to be too tight and exact

with this

since there is quite a bit of good area

that the trackers are going to have
access to.

One more keyframe
for when she's out of frame.

Okay, that should do it.

And now I just need add two more polygons
for these characters.

L name the ones that I already have done.

Okay.

Okay.

So now that we've finished the polygon
for the woman,

we've got two more polygons
to do, one for each of these pirates.

So I'm going to add

another polygon tool by adding shift space

and selecting the polygon
from the search menu.

And then for this one,
I'm going to go to the last frame

because the characters larger in frame.

And so I need to add a little bit
more detail to really define him roughly.

So I'm going to go ahead
and draw around this character,

and I want to make sure
that I capture his shadow.

So not just his body, because that shadow
is really high contrast

and the tracker
may actually track a point here

and that would be a
definitely a false positive.

We've got that.

I'm going to go ahead and play backwards

and right here, he's
basically moved pretty far off

and I'm looking
for those high points of animation.

So either where the character
is furthest away from the roto spine

or they're too close to an edge.

And that's generally
when I need to add a keyframe.

And as I'm adjusting this
on these additional frames,

Fusion is automatically
laying down a new keyframe

for this polygon

play forward.

And I see his arm sticks out a bit here.

So I'm going to go ahead and adjust this

and add a keyframe.

So that works for those frames

and play it back a bit further.

Yeah, I'll add one

at the beginning and then we can just see
where we land in the middle.

So I'm really scale this one down a lot

by holding down s and click dragging

and maybe scale down a bit on Y

to get the top of the polygon
a bit closer to his head

and give him a little bit more room
to the right side of him.

So let's play forward from here.

Looks pretty good.

Doesn't get too close to the edge,
so I think we're good on that character.

Go ahead and name his polygon
by right clicking in the inspector.

I'm going to rename

and I'll name him Pirate one.

Okay, That leaves us with one more pirate.

We're going to go to the last frame
so we can capture

all the detail around him,

meaning we have to

add a number of points
to actually fit this polygon to his shape.

Again, we want to

make sure we're capturing

the tip of the sword and his shadow

so the tracker doesn't lock onto

any of those high contrast areas.

It seems like another good keyframe
because cameras

move significantly further away from him.

I'm going to need to give him a little
bit of room to the right of his hand,

and I can probably tighten this up

with a bit of a wide scale.

I can even use this rotation
on a single point like this.

Sometimes I find it difficult
to grab one of these control handles

so I can just hold down T and rotate

by holding down T and click dragging.

One thing to keep in mind
when you're using those hotkeys

is that the anchor point for the rotation
is based off of where you're clicking

with the mouse.

So if I hold down T
and I click out here, I'm

actually rotating around that point,
which isn't really what I want to do.

I want to rotate around this point here.

So I hold down T and rotate from there.

Okay.

It's looking pretty good.

So let me get another keyframe here
at the very beginning

and see how it interpolate

or polygons get a little bit big.

So I'm going to go and scale it back down.

Yeah, that would work for our purposes.

I'm going to go ahead
and rename this last polygon

to match the others.

And the last thing

I'm going to do
is I'm actually going to view this polygon

tree here to see what a result is.

And you'll see I actually forgot to

change the paint mode on these polygons

for the pirates because we don't actually
want to track these pirates,

so we want them to actually poke a hole in
the other mats that we've already drawn.

So I'm going to go out

and change the paint mode on the polygon
for the second pirate to subtract.

And the same thing for the first pirate.

So now when we play back,

this whole branch, we're seeing

just the areas that we're going
to want the tracker to track.

Okay,
now that we've completed our last polygon,

let's go ahead and view this entire branch
to see what we have.

Okay?

And so what you can see here is

we've got the rocks to find
and then we have the beach defined.

So that's our rocks polygon
and our beach polygon.

And these are both overlapping.

It's okay that they overlap

because those are both areas
that we want to be tracked.

And then we have the woman's polygon
that's actually cutting out a hole

because we don't want to track her.

And then we have this pirate.

I actually didn't change the paint
mode earlier, so I can change it now

by clicking on that node
and then going to the inspector

and changing the paint mode to subtract.

And I'll
do the same thing for the second pirate.

He's also overlapping the other pirate.

That's okay,
because we're just doing a very general

definition of where these objects are.

And ultimately
the only purpose of that polygon

is to subtract from the rock polygon
and the beach polygon.

So I'll set that to subtract.

And now when we play this back,
we're left with

just the areas that we want the tracker
to track and white and the areas

that we don't want the track
to track in black.

So now that we have

that,
let's go back and view our plate footage

and we're going to add a camera tracker,
which is our 3D camera tracker.

With infusion, you can find that
in the effects library under tools

tracking and camera tracker.

So we'll drag the camera
tracker down into our main flow.

And this differs from the

other tracker to the main fusion tracker
because that's a 2D point tracker.

This camera tracker will create dozens
or hundreds of tracks

and from all of those tracks
it will derive

the camera that was used to film the scene

with its internal processing.

And so these polygons

that define the area that we want to
track will feed into the camera

trackers, track mask input,
which is the great input.

So I'll just grab the output
and connected there.

So now that we're feeding

the mat for the information
that we actually want to track, we're

ready to really dive into this camera
tracker and talk about how it works.

So I'll hide my effects library.

Now I've got the camera tracker selected
and in my inspector

there is a number of different things

that we want to set up
before we click auto track.

First thing is, I want to set

preview auto track locations.

And what that'll do is

it'll show me the points
that the tracker is going to track next.

I'm going to change my detection
threshold, and as I manipulate

that, you'll see that green dots
appear on the screen.

And actually so I can make it
a little bit easier for you to view.

I'm going to go to the three dot menu
as a viewer.

I'm going to bring up my gamma
gain controller.

And what this does
is it allows me to to adjust the image.

I'm not actually changing the image.

I'm just adjusting how it's presenting.

So now we can see all these green dots
a lot better.

So these are the tracker points

that are automatically being laid down
by the camera tracker.

You can see as I adjust my detection
threshold, those may change.

Really, the detection
threshold is controlling how much contrast

that the camera trackers
looking for to lay down a point.

Then we have minimum feature separation.

And the minimum
feature separation adjusts.

How far apart
each of these trackers needs to be.

I want to have a lot of points
to work with,

so I'm going to go ahead and reduce

the minimum feature separation
quite a bit.

Next on the camera tracker
tool is the track channel.

Right now it's set to Luma,
which is going to work for our purposes.

But let's say you had footage
that had a very noisy blue channel

you may want to select or a red
or green trail to track.

Instead,

but in this case, Luma

should work for our purposes.

So then we have track range.

Right now
we're going to track the render range.

You can also adjust that to be global,
valid.

Valid is basically the valid range
of that particular media

in or a different custom range.

We're going to leave it to render range.

And lastly, we're

going to turn on bidirectional tracking.

By default, bidirectional tracking is off.

And what that means is that
the tracker will track all these points

while the clip is playing forward.

BI directional tracking will actually try

to extend all of those previous tracks
backwards in time.

So this generally gives you a longer track
which will ultimately

give you a better score.

So I generally like to have this option

turned on.

Lastly, we can talk about gutter size.

Basically that defines how close

these trackers are allowed
to get to the edge of frame.

You can get false tracks if they get
too close to the edge of frame.

So if you need to,
you can adjust this parameter

to pull some of these trackers
back from the edge of frame.

I'm going to go ahead.

Just hit reset
because this value should work fine,

and I'm going to go ahead and

reset my gain control
and turn off that widget.

And now that we've set up
all these parameters,

we're ready to click Auto track.

So what that does is plays

the clip forward and the auto tracker
will follow all those features.

And you can see here
we've got our track points

with blue lines
that define the motion of that point.

And if you hover over any of these points,
you'll get some metadata about that.

What the number of the tracker is,
what the error

is, etc., etc..

So you also notice
all these trackers are tracking

just the regions we defined in white
with all of these polygons here.

So that's exactly
what we were looking to do.

Now that we have some good

track points, we can go to the next tab
in the camera tracker tool.

So we're going to go to camera.

And here what you want to try to do
is add in all the information

that you know about the real world
production camera.

So in this case,
I know this shot was captured

with the Blackmagic
Ursa production, 4K at 69.

So I'll select that.

And I also know
the focal length is about 13 millimeters.

So I'll just that there.

And so that basically gives

the camera tracker
a hint to what the original camera was,

which helps the camera tracker
solve for the real world camera faster.

So if you ever want to check
to see if your source footage

has any of the camera information
embedded into it,

you can click on the media
in for that particular shot,

view it
and then go to your sub view button here.

And so
in that sub view, there's a dropdown

which gives you
a number of different options

and if you select metadata,
this will display

all of the metadata that's associated
with that source footage.

In this particular case,
we don't really have

any useful information,
but often there's camera information

like focal length,
ISO time code and other important things.

So I can expand this box by just clicking,
dragging the edge,

and I can move around within the box
by center

clicking and dragging around.

I already know all this information
and I already know the camera information,

so I'll just go ahead and hide
the sub view and let's come back

to the camera tracker.

Now that we've filled in all of the camera
information

that we know, we can go to the next
tab, which is the solve tab

first thing you

want to do on the solve
tab is click solve.

The progress bar may get to 98%
multiple times.

Just give it time
and eventually that solved will complete.

So in this case,
the solve is now complete.

And you can look in this area
for information about the solve.

Most importantly, you want to look
at the average solve error.

And in this case we've got point
three pixels solve error.

What that means is that for every point

there is an average error rate of about
a little bit under a half a pixel,

and that's a pretty good solve rate
You wouldn't want to solve error.

That was over one.

And generally the higher the resolution,
the larger the solve error you can accept.

So for HD footage, anything under

five is generally considered usable.

If you do need to refine this all further

to get a better average solver,
there's a few ways to do that.

You can find points manually
that you think might actually be an error.

So in this case, I'm
going to delete this point

because it's on the
edge between this foreground

palm and this background palm.

So it may be a false positive.

So delete that.

And in fact, these palm fronds

move throughout the shot.

So I'm going to select all these points

and then delete them in the inspector

and you can play through the shot
and see if any additional points pop up.

I see a few that are.

So I'm going to go ahead
and go to those frames

and delete a few of those tracks manually.

You can also find some read trackers
and generally

those are trackers
that have a larger solver.

That track in particular

is moving up the edge of that rock.

And so what's probably happening
there is there's a false pattern that it's

finding between this foreground rock
and this background rock.

So I'm going to
definitely delete that point.

There's a couple other points here
that are red.

Let's see how those play.

Those
actually don't look like errors to me

except for this one on the bottom.

This point on the bottom here
ends up tracking

the splash of the waves.

So I'm going to delete that one.

Okay.

So that's one way to refine points

and we're going to hit
command F to go full screen.

Another way to refine

and reduce the number of bad points

is to use your track
filtering in the inspector.

So if I adjust my track length,

this is finding any point
that has a minimum

track of less than the number of frames
that are defined here.

Same thing for a track
error and maximum solve error.

So I can adjust these parameters
to identify the

the tracked points
that don't fit these parameters.

Select track satisfying filters,
and then I can delete all those tracks.

Now that I've deleted some of these points
that are maybe less high quality,

I'll go ahead and hit solve again
to see if I get a better average solver.

So our solver just went from point
three to point 25,

which is great.

The lower

the solver,
the better result we're going to get.

Now that we've completed the solve,

we can move on to the export tab.

So the export tab
you use to actually generate

your 3D camera and set up a 3D scene.

Before we hit this export button though,
we want to go into the 3D scene transform.

So what the 3D transform will allow us
to do is adjust

where the origin of the scene
is, what the orientation

of the scene is, and what
the overall scale of the scene is.

So to get started with these parameters,

you need to change it
from aligned to unaligned,

and that allows you
to change the 3D scene transform.

So the first thing I want to do

is set the orientation
to the ground plane.

So I'm going to select a series of points
that are on the ground plane

and then I will set those from selection.

So now the X, Y and Z rotation now

are going to match these points
in setting our ground plane.

Now, if you don't see a ground plane,
let's say

we just saw this rock wall or a building,
you could set your selection

to x y plane, which will allow you to set
your orientation based off of a wall.

Instead of a ground plane.

But in this case, the ground plane works
fine for us.

One thing I need to do
now is set my origin.

So I'm going to go ahead
and go to the last frame,

select a few of these points
and set my origin to here

by saying set from selection.

Now I'm ready to actually export
and create my 3D scene.

I do that by clicking the export
button here.

You'll see in the flow
a number of nodes automatically get added

and I'll just move them to organize things
a bit better.

So the camera tracker has created

all these nodes for us
and basically set up our 3D scene.

What we have here is the camera 3D node,
which is our virtual 3D camera

that matches the camera that was used
to actually film this footage originally.

We have a point cloud

and a ground plane that we just define.

Those are all feeding into a merged
3D node,

which is a specific node
used in the 3D environment to combine

3D objects together.

And then we have a camera tracker
renderer node.

And so what this will do is
it will render all of the 3D objects

using this 3D camera.

So when
we place a 3D object into this scene,

it will match the live action footage
perfectly.

So let's go ahead and view the 3D scene
by clicking on Merge

3D and pressing two.

So you'll see I go into Fusion's
3D environment

and as I drag around

I'm using alt click to drag around.

I can see my ground plane's a little bit
slanted here, which actually is fine

because the beach is a bit
slanted as well.

So I'm going to be okay with that.

I've got my virtual 3D camera

that you can see here
and all these points.

This is the point cloud and these points

match the tracker points
from the 3D camera tracker.

So let's go ahead and play this through.

You can

see that this camera's
moving through the 3D space,

recreating the camera
that was used to capture this footage.

Originally.

All right.

Now that we have our 3D camera,
we can actually place objects

in this 3D scene for final compositing.

So let me go back to the original footage

and do a little bit of organization
in this flow.

I'm going to go ahead and just move
this camera tracker off to the side

and connect this plate to my media out.

I'm not going to actually need
this camera tracker

for the remainder of this project,
so I'm just going to move it

up to the side here.

If I ever do need to modify anything, I
can go back to this camera tracker later.

Okay.

So in this scene,
I actually want to put a pirate ship here

a couple of hundred yards
off of the coast.

So let's go to our media pool
and I'm going to grab that pirate ship

and drag it in.

First thing I'm going to do is name

this element by

changing it in the inspector.

And now I want to place

this pirate ship in that 3D scene.

So what I need to do
is actually put this image

onto a 3D object
so I can place it in the scene.

So I do that
by going to the effects library

and then under my 3D menu,

I can scroll down to find my 3D section

and in there I have an object called
Image playing 3D, so I can go ahead

and add that into my flow
and connect the ship to the image plane.

So what this will

do is it will map
the image onto this plane,

which I then can merge
into the rest of my 3D scene

by dragging the output
and dropping it over Merge 3D.

And let's look at our three scene
and see what we have.

So if I can't see
where this image point is

because I'm too far zoomed out,
I can click on it and then in the viewer

I can click F
and that will zoom in just to that object.

And now I can rotate around and I see

my ship has been placed

right where the camera's that

I want the ship to be placed
very far in the distance.

So what I need to do
is push it further back in the scene.

And what I'm going to do actually
now is I'm going

to switch my view to a top view.

So I right click in the viewer,
I go to camera

and then I select up from here.

I'm going to zoom out by holding command
and then scroll wheel

and then translate

by clicking my scroll wheel

and dragging.

And so I can see all my points.

This is my beach, these are my rocks.

So I want this image plane
to be pretty far out in the distance.

And if I'm not sure
which object is the image plane,

I can click off of that node
and then click back on it.

Now I can see where it is.

So I would say these rocks are probably

50 or 60 yards from the camera,
so I really want this ship

to be pretty darn far away,
maybe two or 300 yards away.

So I'm going to move it roughly over here.

So what's

happening here is
that the camera has a far clipping plane,

which this ship is beyond,
So that's why I can't see it.

So what we need to do is actually adjust
this far clipping plane.

So I'm going to select camera

and I'm going to turn off the lock

that's automatically on it,
and then I'm going to adjust my far plane.

So I've just added our two 250 or

so when to reset that lock

and now lets VR render.

So I can see here that even though
I've scaled up the ship, I can't see it.

And what's happening here is my image
plane is blocking the ship

and also my far clipping
plane is blocking the ship.

So the far clipping plane is

where the 3D environment
will basically end.

As far as this camera is concerned,

they won't render anything
that's beyond that point.

So to just that, I'm going to go to Camera
3D when a disabled or my lock

and then I'm going to
adjust my far clipping plane

somewhere like 50.

And so that's beyond the position
where I put the ship.

And then I'm also going to adjust
my image plane

as well by going to the image tab

and adjusting the depth of that.

So now nothing should be blocking my ship.

So I'll go ahead and set this lock again
because I don't want to accidentally

change this camera
and let's view the tracker.

Okay?

Now we've got a ship, it's showing up
and we can see it.

So I've roughly placed it at the distance
I want.

I can also change
where it's positioned on X and Y.

So let's go ahead and adjust that a bit.

You know,

when it roughly there,
I think the ship looks a little bit big,

so I'm just going to
make it a bit smaller, maybe

maybe 15 or so.

Okay.

Now that I have that, let's
go and play this back and see how it looks

in context or make my screen a bit bigger.

So you can see that
the ship is tracking nicely in the sea

and it's staying where it should size

looks pretty good.

Make it a little bit smaller
and move it a little bit to the right

because at the last frame here,

this character is crossing over it
and I'd rather not rotoscope him.

So I'm just going to go ahead
and adjust my X translation

a bit, scaled down a bit.

And just to recap a little bit,
I'm adjusting the image plane directly.

I'm not doing anything to the ship.

So this is just changing the position
of that 3D object.

I'm leaving the Z position alone
because I liked how far I put it

and then I'm going
to adjust my y position.

So it's right on the edge, all these waves

and let's see how that plays back.

Cool.

Looking good.

And on this last frame here,
this guy doesn't appear to be

crossing over, which is perfect.

So we've placed our ship

in this 3D environment
and we're rendering that

3D environment a rendering that ships
through that 3D environment.

And it's matching the camera's
original motion because we did the 3D

camera track
that generated our virtual 3D camera here.

So the next thing we need to do in this
composition is

we need to improve the blending
of this ship.

Right now, it's obviously way too dark.

We could match the ship to the scene
with a color corrector node,

but there's a 3D node
that can also help us achieve that.

Look, we're looking for,
and that's called ambient light,

so we can find ambient light

under the effects library,
under the 3D tools and then light.

So I drag that in.

This is going to basically create
a light source

that is going to illuminate
any of the 3D objects.

It's not directional, it's just an ambient
light that's in the environment.

So what I want to do is I want to connect
that to my merged 3D node so that it's

part of this 3D scene and I actually want

to turn on my lighting in my renderer.

So I select my renderer, they go under
lighting and I want to enable lighting

so I can see immediately
the shift becomes pretty dark.

We can adjust that
by adjusting our ambient light.

So I'm going to go ahead and click
my ambient light and adjust my intensity.

It's getting pretty
close, but I'm going to go beyond one.

Let me just jump it
up to, let's say eight.

It's too bright,
can pull it back from there

and that's looking reasonable.

Now I'm just relighting it a bit.

But one thing about this ship,
if we just view that asset on its own,

is it looks pretty red

and warm, the ship looks pretty warm.

So when we're viewing
the actual environment, it's

pretty far in the distance.

It should maybe look a bit bluer.

So what I'm going to do
is instead of just lighting this scene

with a white light,
I'm going to change the tint a bit

and I'm going to drag this eyedropper
over the sky here.

And so that's going to give us basically
a blue light to help illuminate this ship.

Okay, So this is a before

and after, before and after.

It just tends to blend a little bit
better with the background here.

So the ship is in the scene.

It's tinted just about right.

The black levels are too dark for this.

So I'm going to add a color corrector
this time

actually after the camera tracker.

So going to do that by bringing up

my tool search

and finding color corrector.

And now I can adjust my black point
by adjusting the lift.

But you'll notice
that the entire scene is lifting.

So I view this color corrector.

You'll see that it's actually bringing up

what was transparent for the color
corrector tool

you need to enable the pre divide and post

multiply option under the options tab.

What this will do is it will make sure
that your color correction

is only applied
to the areas defined in the Alpha Channel.

So now that we've done that, let's view
the merge node

and we can adjust our color corrector.

So when to bring up

black levels,

this is roughly matching
the black levels of these rocks,

which is grays.

And I'm going to adjust my gamma

to get a little bit less contrast
because the ship is so far away.

I need to bring down my left
slightly as well.

And so now I feel like I'm

I'm generally recreating the proper look

because there's a lot of atmosphere
between me and that ship.

It would be a bit blue,
a bit less contrasty.

And so this is looking a little bit
more realistic.

So let's play back those results.

That's looking pretty good.

One thing I do notice, though,
is that when I'm playing this back, the

ship is very sharp.

Even during this
this camera motion in these camera pans.

What I need is to actually have
some motion blur apply to this ship.

And I can do that
actually directly in the camera renderer.

So if I go to the renderer for the scene,

I can go to the settings tab

and enable motion blur.

So let me actually before I enable that,

I'm going to go to a frame
where there should be motion blur.

Maybe you're right there
where there's a pretty good pan,

so I'll enable motion Blur on the render.

See, now we get some motion blur
that seems like it'll probably match

pretty well.

I'm going to adjust the quality
up to maybe four.

And what the quality does is

it will draw additional

versions of the ship.

So if I draw this all the way down
to one, there's no more motion blur.

If I lift it up to two, there's actually
two images superimposed on each other.

You can kind of see that up here.

I get two additional images
around the original, so I want to

maybe bring that up to four

so I get a nice smooth motion blur.

The faster the motion, the higher
the quality you need to go.

I get also adjust my my shutter angle
if I know this is a 90 degree shutter

or a 360 shutter, I'll leave it at 180

because that's pretty standard.

So now when we view this playing back,

we'll get a much better result
because we have the blur baked in.

So let's

go ahead and view our results
and then we're going to jump

to the edit page
and watch them full screen.

We jump to the edit page.

I'm going to let it

cache here for a minute
and then I'll shoot that playback command

F to go full screen.

So that's looking pretty good.

There's a few additional things
we want to do to this scene and that will

illuminate
some other parts of 3D compositing.

So the first one,
if I go forward a bit in the timeline,

is we have some of the crew's
footprints here.

So a very common task

in 3D compositing is painting out

objects and doing cleanup,
but in the 3D environment.

So in this particular case,
I want to paint out this footprint,

but I need to have it
actually mapped to the proper surface and

and track properly
so it matches perfectly.

I could maybe do that with a point
tracker or plane or tracker,

but often that's pretty difficult
and it can be even more difficult

if it's a more complex shape like a barrel

or the corner of a wall
or something like that.

So in this case, we're going to go ahead
and paint this footprint out

on this particular

frame, and then we're going to place
that paint work on on a 3D object

that will
then track perfectly into the scene.

So the first thing we want to do for this

kind of paint work is
we want to find a frame.

This area is biggest in frame
and it's also

pretty much in focus

with no motion blur or depth of field.

This frame should work pretty well for us,

and I'm going to start out this process

by using the Time stretcher tool,

which you can find in the pop up menu,

or you can go to the effects library

and miscellaneous,

which has the time structure tool.

So drag that into my flow.

I'll connect it to the original plate.

And what the time structure tool does
is it allows you to re

time things where allows you to freeze
frame one of the sources.

So in this case,
I want to hold this frame on frame 71.

So I'm going to go ahead and enter
71 into the source time.

Now, if I view this time structure

and I playback, all I see is frame 71.

That's exactly what I want.

So a bit of organization, I'm

going to go ahead and move
some of my nodes around,

clean things up

and to make room for this time stretcher,

which I'm going to

put down here.

So now I want to paint this one
individual frame.

So I'm going to add a paint node
from the effects library,

the paint bin,
and then I'm going to drag paint in.

So there's
several options to this paint tool.

There is multi stroke
clone, multi stroke and stroke.

I'm going to use stroke

because that lays down a paint stroke
that will apply across all frames.

I'm going to change my apply mode

from color to clone

and then I'm going to adjust
my brush controls as I need to.

I'm fine with this size,
but I could make it a bigger

brush and I could adjust
the softness of that brush.

But I'm going to go ahead
and just leave it at default.

I want to view this tool,

so I'm going to go ahead and load it
into the second viewer.

So using the clone tool is very simple.

I've got this X in the middle
and I've got a circle around

the outside that defines my brush area.

Now I want to clone from this area
and paint over here.

So I'm going to hover over this part of
the image press alt and then left click.

And that's going to lay down this X
which defines my source point.

Now I'm going to move the brush over

and as soon as I
click and start dragging this

X for the source is going to lock
to that point and start

painting the area from here

over to where my brush is located.

So as I do that,
I can now start to remove that footprint.

And if I want to

change the position of this offset
for the source and the destination

of the paint, I can do alt click again

to select a new area to paint.

And now effectively remove that foot.

So now that I've done that,

I'm going
to actually just create a rough polygon.

So I can just cut out this region

and I can use that
to place in the 3D scene.

So to do that, I'm going to add a polygon

with my tool search menu

and draw around the rough area

where I laid down those paint strokes.

I mean, low down the viewer,

and I'm going to soften this edge a bit
to just smooth the transition

from the paint work
I've done to the original footage

and to actually combine
this alpha with this paint work.

I'm going to use a matte control node,

which I'm going to find through my quick
search menu, and that's going

to allow me to combine this RG information
with the alpha from the polygon.

So I'm going to connect that
to the foreground input of Mac control,

view that tool and then change my mode
from none to combined Alpha.

So that's basically copying
the alpha from the polygon

into the flow
with the RGV data from the paint.

Now I can still see
some of the background there.

That's because I have an alpha,
but I haven't pre multiplied the image,

so I have to select post multiply image.

Now I have just the paint work and
this patch of ground that I can work with.

So one thing to note
here is actually I'm on the edge of frame

and if I'm going to actually place this
on a 3D object,

I don't want to have a hard edge
like that.

So I'm going to actually
adjust my polygon back a bit

like that

and I'm
going to reduce the softness a bit.

So that's a little bit more than I need.

So now I've got a nice soft patch here

that I can have placed
in the 3D environment.

So to actually place this patch
into the 3D environment,

we're going to use a technique
called reproduction.

So what that does is
we will project this patch

through the camera
on this particular frame

and it will project
onto a piece of geometry

just like
you would have a movie projector.

So I'll get a little bit

more into the detail of why that works
and show you why that works.

But just as an overview, that's the rough
technique we're going to use.

So to set that up,

we want to actually

project through our camera, our 3D camera,
but only on this frame.

I'm going to copy this 3D camera

and paste it up here,

and I'm actually going to unlock that

copy of the camera
and I'm going to rename it

and I'm to rename it to Cam

f r 71 for frame 71.

I don't want this camera
to be moving at all

because this is from frame 71.

I want that camera to stay put

because I want my projector to stay put.

So I'm going to go to transform

and I can see that
there's animations in this transform

because there's all these red diamonds
and I can see it in the timeline here

to actually freeze this camera
on this frame.

I want to right click translation
and then remove all

which is remove
all key frames from the translate group.

And what that will do is it
will keep the parameters for that frame,

but remove all the rest of the parameters
and I'll do the same thing for rotation.

Right now I've got a static

camera
that I can use to project this image

onto a piece of geometry,
so I'll go ahead and drag this down

here and I'll connect the Mac control

output to the camera as an image input

and a few other settings
I want to adjust is

I need to go to this projection tab.

This is a new tab
that just popped up for a disconnect.

This camera again, I only have three tabs,
but when I connect an image

I'll get additional tabs to work with.

So I want to go to the projection tab
and I want to enable

camera projections, turns
this virtual camera into a projector,

and I want

to change our projection mode to texture.

So what that means
is that it will take this texture input

and directly map
that onto a piece of geometry

without letting the scenes light color

that texture at all.

So now I can connect this camera
to my Merge

3D to bring it into the scene.

And if we view

the Merge 3D, so we're viewing the scene,
we can see

I've got my patch here
and it's projected far in the distance.

Not exactly where I want it,
but it is now in the scene.

What I need to do now is actually place
a piece of geometry

for this projector to project on to.

So what we're going to do is we're going
to change our view from this perspective

view by right clicking in the viewer
and going to camera.

And then we want to view actually
through the original 3D camera.

So now we can see the scene as it was.

But let me do this.

I'm going to go ahead and connect my plate
back to our main

3D camera so we can see what's going on.

So you can see I have all these points
on the distant rocks

and the near ground plane
and they're a little bit too big.

So I'm going to go in
and click on the Point cloud.

I'm going to adjust the size down
so I can see a little bit better

what's going on.

Okay.

And for the time being,
I'm going to go ahead and deactivate

or I'm
going to disconnect this projection.

All right.

There we go.

So now I can see the footprint here.

So what I want to do is
I want to select a couple of these points

and I want to actually

align a shape to this

actual set of points. So I can do that.

I can align the shape to these points
by selecting them and then right clicking

go to point cloud,

and then I can create a shape

aligned shape or an image plane.

And in this case, since
I'm going to be projecting the image,

I'm going to go ahead
and just create an image plane.

All right?

And you can kind of
see that there is a little sliver.

I'm going to go ahead
and go back to perspective view

and we can see

our image plane was created up here

merged into our point cloud.

I'm actually going to separate
this image plane

by shift click and then dragging

and I'll delete that
extra merge and I'll put this

image plane back into the scene.

I can't quite see it,
so I'm going to f find

where the image plane is that now
I can see where it's at.

All right.

So I want to check the orientation
and the size of this.

It's a bit tilted back for my liking,

so I'm going to go ahead
and turn on this rotate widget

and just fine tune
the orientation of this plane

so that it's aligned

with the other points in the

3D environment.

Okay.

So that should roughly work.

And I'm going to go back to the original
3D camera

by right clicking, going to camera
and then viewing through the 3D camera.

Now I can see that
that plane is a little bit

better
oriented to look like it's on the beach.

I'm going
to go ahead and increase that size

by going to the transform tab

and scaling it up

so that way we have a nice large surface

for this projection to land on.

So the last thing we need to do
is actually change this

from a regular 3D object
to what's called a catcher.

So we're going to go to the effects
library tools, 3D texture

and then catcher.

And so basically this catcher texture

is designed to catch image projections.

So I'm going to go ahead and drag that in
and attach that to my image plane.

And now instead of seeing this image
plane, we will just see the images

that are projected onto that image point
because of this catcher.

So I can go back down to

this group of nodes,
which is where I set up that patch

and that projection, that camera
to project

that patch and reconnect it to the scene.

And now projected that patch
onto this piece of geometry.

If I go ahead and
go back to perspective mode,

we can see that now that patch is mapped

onto that piece of geometry
which aligns with this ground plane.

So what that gives us

is that when we play back this scene,

you can see this camera here,
this is the projection camera.

And so it's always static

projecting that piece of texture
onto that piece of geometry.

All the while the main 3D camera is moving

through our 3D environment
and capturing that patch

as if it were stuck
to that spot in the environment.

So if we view that through our renderer,

we won't see the footprint anymore.

That footprint
has been completely painted out.

And actually
let me go ahead and disconnect

our original plate from the main camera.

So we just see
the objects that we've added.

Oh, and one thing I forgot to turn off.

So right here you're seeing the
projection, and right here you're seeing

something.

You're seeing the actual image projection.

So by default, this projection camera

also has enable image point enabled.

So I want to disable that.

And now that disappears and I'm just left
with the projection from that camera.

So as I play through,

I'm going to go ahead and pause on
this is frame 71

from which the patch originally came from,

and it is mapped to that scene

as if it was actually
in this 3D environment.

So if we look at that merged over the top,

we'll see.

We're getting a pretty darn good result.

Now this patch is too bright.

What's happening here is

all of our renders
are going through this color corrector.

So if I disable this color corrector,
the patch actually looks correct.

So what I need to do is constrain
this color corrector to just the ship.

So I'm going to create
a very simple rectangle,

move that over the ship

and this will give me a mask

that will constrain the color
corrector tool.

So if I view the rectangle on the viewer,
it's just a very simple

mask.

So now if I pull it back,

I'm going to get a much better result

where the ship looks right
in Patch looks right.

And actually I look at this
a little bit closer.

I think this patch is a bit red early on

and then it matches
better later in the shot.

So what I'm going to do
is I'm going to actually add

another color corrector,
may organize my flow a bit better here.

I'm going to add another color corrector

just for that patch,

and I'll do that

after the color corrector for the ship.

And just keep things clear.

Let me rename that color corrector patch

and I'll rename
the original color corrector

as color corrector ship.

And I also want to do a rectangle
to constrain

this color corrector to just

wear the patches.

And like before, we need to enable

pre divide and post
multiply on the color corrector

and I'm going to go ahead
and lay down a keyframe

for gain and for color on this frame.

When this patch looks good
and I'll go back to the first frame

and adjust things a bit on the wheel

here to get this color
a little bit more yellow.

And I'm going to boost the gain slightly

to get something

that's a little closer match.

Okay.

Those are before and after.

That's looking pretty good.

Now we can play through.

Gets a bit red here still,

so I'll go ahead and

just that.

There we go.

So I've got a third keyframe

and that seems to blend pretty nice over

time.

Okay, so that's looking pretty good.

I'm going to go to the edit page
so we can play this back full screen

and take one more look.

So as we're playing back,

it's looking pretty darn good.

One last thing that I notice that I think

probably should be addressed is

this is an American flag
and this pirate ship.

So a pirate ship is not really
a pirate ship unless it has a pirate flag.

So let's go in there that in real quick

and we can actually make
that pirate flag wave in the wind a bit.

So back in our fusion composition,

I'll go to my media pool

to find my pirate flag,

drop it into the scene,

and I'll just work directly

on this pirate flag
for a couple of minutes.

I may rename it

the flag and let's view that.

So it's going to be pretty far
in the distance.

I think I can get away with all this color
that's already built in.

But what I want to do
is actually change the color

over time as if it's catching the light
a bit differently in the breeze.

So I'm going to add
a tool called Fast Noise

so you can find the fast noise
tool under effects library generator

and then fast noise.

And what this will do
is it'll create a noise texture

that's static at first,
and I'm going to go ahead and change it

so it animates a bit
by adjusting the seas rate

and just my contrast
and my brightness a bit.

I'm bringing the brightness down because
I want some areas to be totally black.

I want some areas to be negative and

yeah, let's see what we're looking at here

right?

There we go.

Now I've got some negative values
and just had to adjust

that contrast up a bit.

So I'm going to modify this flag

by using a tool called

Channel Boolean.

And let me show you where that is
in the effects library

color bin.

Now drag that in.

I'll connect the flag to the background
and I'll connect the

fast noise to the foreground,

and I'm going to change the

operation from copy to add.

Send some of our fast
noise values or negative values,

which if I have my mouse
and then you look at the bottom portion

of the screen here, you'll see
negative values and then you'll see

values that are above zero
and some values that are just about zero.

What's going to happen
is I'm going to add a negative,

which will darken part of the flag,

and I'll add values that are positive,
which will brighten it up.

So if I go to the flag
and then we look at the channel Booleans,

now we can see some areas are brighter,
some areas are darker.

I may hide my effects library.

I do see a hard line here
and what's happening is my fast

noise
isn't the same resolution as my flag.

So I need a change that real quick.

So I'm going to click fast noise,
go to image, turn off auto resolution,

and then change my Y-axis resolution

to be the same height as my flag.

And now

I'm going
to go ahead and just change my alpha.

Instead of making some of the flag
transparent.

I'm just going to keep my background
Alpha.

Okay.

So now this operation is just

affecting
the red, green and blue channels,

so it's getting a little bit too bright.

So I'm going to adjust some of these fast
noise parameters

to get a little bit better results

where we go.

So let's play that through.

Now we're seeing some lighting changes.

Doesn't look amazing at this point,
but it will work

later on, especially given how far away
the flag is going to be.

So the next thing we need to do
is actually place this

flag on a 3D object
placed in the 3D scene.

So we'll go ahead and
do that with image playing

and then we'll add that image plane

into our 3D scene
by connecting it to the Merge 3D node.

And let's go ahead and view the 3D scene

again.

If you can't actually find the object
in the 3D environment,

you can just select the tool
and then hit F

and then it'll bring you
right to that tool.

Well, I know that
I want this flag to be on the ship.

So what I'm going to do
is I'm going to go to my ship object here.

I'm going to pin that.

And what that'll do
is that'll keep this node visible.

Even when I another node.

So I'll select the image,
play node for my flag

and now I can go to the transform tab
and I can match some

of these values to get this

flag into roughly the same position
as the ship.

So I'm going to go ahead
and put in a value of 14.51

and -35 oh.

The misma negative.

And now for F

looks like that flag is right

roughly where the ship is at.

So now I just need to use
my on screen controls here

to put this flag right
where the American flag is

and change my scale
and all the rest of it.

So I hit my rotation widget there,

match that up

and again,

I'm using alt center click

drag to orbit around and just center

click drag to translate around.

Okay,

we're roughly in the right position,
but my flag is definitely too big,

so let's scale that down

on our image plane so they should be

big enough to cover the American
flag but not too big.

Okay, we're looking pretty good.

I don't want to intersect it, so I'm
just going to pull this flag a bit off.

That's looking pretty good.

The one thing we're missing
is this piece of geometry is just flat.

What we can actually do is
we can manipulate this geometry

with a node called the displaced 3D node.

So let's get that from the effects
library,

3D tools

and displays 3D.

So again, that's under the effects
library 3D and then the 3D tools,

it'll be displays 3D, and I'll drop that
in the flow test before,

oh, just after the image plane.

And I'm going to actually connect my input

to the fast noise.

And so this fast noise texture will
actually displace this geometry in depth.

So let me do that. You

can see right now very spiky.

We don't want it that spiky.

What we can do
is go to the image plane here

controls tab and let me unpin

that image there in image plane

and I can increase the subdivisions
and that's going to really smooth out

me zoom in there so we can see a better

that'll smooth out

the displacement.

So now we get much softer curves

still the displacement a bit too extreme.

So I'm going to bring down the scale

and now we're starting to have a flag
that if I play this,

be waving in the wind,
you know, very roughly something

that would be effective
because it's so far in the distance.

One last refinement I'm going to do is

this flag is actually going
to be attached to this pole.

So we don't want it waving at this far
edge here.

So I'm going to adjust the fast noise
so there's no effect on this left edge.

So I'm going to view my fast noise.

And to do that, I'm
just going to add a simple rectangle

and connect that to the mask

input of the fast noise,

adjust my size

so it's not affecting
any of the regions to the right

and it's just cropping off a bit here

and I'm going to go ahead and soften
this rectangle.

So essentially what's happening

is we're going from zero effect.

We just view one of the channels, we're
going from zero effect

up to having the full effect
by about here.

So the effect on the 3D object

is there's

no distortion
at the very edge of the flag here.

And then there is more at the end.

And actually I can probably go a little
bit more extreme with the soft edge

that's going to make my box bigger

so I don't affect any of the other edges

and let's take a look at that again.

Yep. So that's giving me roughly
the effect I want.

Now let's view it in the final composite.

So now we've got
the image comped in there.

We can see our flag in the distance here.

It's catching a bit of light there,
but we can see it's still skull

and crossbones.

And let's go to the edit page
to actually play this back.

And now we have our flag
flapping in the wind

and I think it's selling pretty well.

And now

we've got the final playback
with the flag flapping in the breeze.

And just as a quick recap, this is 3D
compositing with fusion within resolve,

where we use the 3D camera tracker
to create a virtual 3D camera

so that we could place our pirate ship
into that 3D scene

and also do some cleanup
to remove the crew members footprints

and add a flag at the end
that's flapping in the breeze.

So thank you for joining
me and see you next time.