[Tony Kloiber]

Well assuming I’m right in my thinking here (I could be wrong, I have no example footage to verify my thinking) lets name the layer for clarity.

Footage of two guys in car against green screen = TwoGuys
Footage of street wising by used for filling in green screen area = Bkg
Solid layer = Bkg matte
Null layer = Well… Null

Ok you talked about rotoscoping, I was reading that as a way to improve the key. So I thought you had roto’d around the TwoGuys shot. Assuming that there isn’t a lot of movement of the camera independent of the TwoGuys shot the shape of the window should not change that much so.

Draw a mask (rotoshape) around the inside edge of the window (also assuming you can only see out one window). Disable this mask on the TwoGuys layer, you won’t use it here. Copy that mask and paste it to the Bkg matte layer.

The Bkg matte layer now has the shape of the window, but you need to add the movement of the TwoGuys shot to the Bkg matte (remember your simulated bumps).

Open the TwoGuys layer in the layer window (double click the clip) and setup a track. I’m assuming everything is the easy way here so you can us a single point tracker and get the x and y movement of the TwoGuys shot.

Pick a point on the window, some where that isn’t obscured during the whole shot and track it. There are a lot posts and links about tracking in AE if you have problems with it or it isn’t the assumed easiest track.

Once the tracking is complete and it looks like it’s matching the movement well, then apply the track to the target layer which would be your Null. People have lots of reasons to use a Null for the target of a track. At this point its kind of a habit for me.

In AE you can’t apply a track directly to a mask (rotoshape) so to achieve this you apply the mask to a solid (Bkg matte) and then the track is applied to a Null.

Parent the Bkg matte layer to the Null layer.

Now if everything was easy the solid layer should be covering the TwoGuys car window area and moving along with it.

Use the Bkg matte layer as a (wait for it) matte layer for the Bkg layer. You want the Bkg layer to be big enough that as the Bkg matte layer moves around it never reveals the edge for the Bkg layer. If you shot the Bkg layer with the same camera, lens and focal length and the scene filled the frame you’ll be fine, just watch out for the proportions of the two clips (Bkg and TwoGuys) if the relative sizes of objects in the two clips are very different you will have a hard time making a convincing composite.

To sweeten the composite you can feather the mask on the Bkg matte layer just a little (2 or 3) this will allow a simulated light wrap from the Bkg layer onto the TwoGuys layer.
Also If you had shot the Bkg layer with the TwoGuys layer (as one shot) the motion blur from the cars movement over the bumps would effect the Bkg Layer.

To get this motion blur you could use the Y axis information from your track of the TwoGuys layer and apply it to a motion blur (plug-in) on the Bkg Layer.
I’m not sure how to do it exactly but I would imagine that you might be able to write an expression that would give you the amount of blur relative to the amount of change in the Y direction. I’m just considering the vertical motion blur here to make it simpler as well as the assumption that there already is a great deal of horizontal blur from the forward motion.

TonyTony

[Tony Kloiber]

Well assuming I’m right in my thinking here (I could be wrong, I have no example footage to verify my thinking) lets name the layer for clarity.

Footage of two guys in car against green screen = TwoGuys
Footage of street wising by used for filling in green screen area = Bkg
Solid layer = Bkg matte
Null layer = Well… Null

Ok you talked about rotoscoping, I was reading that as a way to improve the key. So I thought you had roto’d around the TwoGuys shot. Assuming that there isn’t a lot of movement of the camera independent of the TwoGuys shot the shape of the window should not change that much so.

Draw a mask (rotoshape) around the inside edge of the window (also assuming you can only see out one window). Disable this mask on the TwoGuys layer, you won’t use it here. Copy that mask and paste it to the Bkg matte layer.

The Bkg matte layer now has the shape of the window, but you need to add the movement of the TwoGuys shot to the Bkg matte (remember your simulated bumps).

Open the TwoGuys layer in the layer window (double click the clip) and setup a track. I’m assuming everything is the easy way here so you can us a single point tracker and get the x and y movement of the TwoGuys shot.

Pick a point on the window, some where that isn’t obscured during the whole shot and track it. There are a lot posts and links about tracking in AE if you have problems with it or it isn’t the assumed easiest track.

Once the tracking is complete and it looks like it’s matching the movement well, then apply the track to the target layer which would be your Null. People have lots of reasons to use a Null for the target of a track. At this point its kind of a habit for me.

In AE you can’t apply a track directly to a mask (rotoshape) so to achieve this you apply the mask to a solid (Bkg matte) and then the track is applied to a Null.

Parent the Bkg matte layer to the Null layer.

Now if everything was easy the solid layer should be covering the TwoGuys car window area and moving along with it.

Use the Bkg matte layer as a (wait for it) matte layer for the Bkg layer. You want the Bkg layer to be big enough that as the Bkg matte layer moves around it never reveals the edge for the Bkg layer. If you shot the Bkg layer with the same camera, lens and focal length and the scene filled the frame you’ll be fine, just watch out for the proportions of the two clips (Bkg and TwoGuys) if the relative sizes of objects in the two clips are very different you will have a hard time making a convincing composite.

To sweeten the composite you can feather the mask on the Bkg matte layer just a little (2 or 3) this will allow a simulated light wrap from the Bkg layer onto the TwoGuys layer.
Also If you had shot the Bkg layer with the TwoGuys layer (as one shot) the motion blur from the cars movement over the bumps would effect the Bkg Layer.

To get this motion blur you could use the Y axis information from your track of the TwoGuys layer and apply it to a motion blur (plug-in) on the Bkg Layer.
I’m not sure how to do it exactly but I would imagine that you might be able to write an expression that would give you the amount of blur relative to the amount of change in the Y direction. I’m just considering the vertical motion blur here to make it simpler as well as the assumption that there already is a great deal of horizontal blur from the forward motion.

TonyTony

[Tony Kloiber]

You of course know you can download it and demo it to see how you feel about?

TonyTony

[Tony Kloiber]

Is the camera (appear to be) in the car with your guys? If so then the camera bumps along with the simulated car bumps? In that case the window footage would appear to be more stable relative to the car window. If the car bumps up then you would see more of the outside space at the top of your background layer (scene outside the car) and then the opposite when the car bumps down. There will be some motion blurring in the direction of the car bump motion on you background layer as well.

Use the window roto on a solid layer, use that layer as a matte for the background footage. Track the motion of the window layer and apply it to a null, parent the solid layer to the null. To use the y variations in the track as a variable for a vector blur on the background footage you might have to poke around a bit to figure that one out.

TonyTony

[Tony Kloiber]

Is the camera (appear to be) in the car with your guys? If so then the camera bumps along with the simulated car bumps? In that case the window footage would appear to be more stable relative to the car window. If the car bumps up then you would see more of the outside space at the top of your background layer (scene outside the car) and then the opposite when the car bumps down. There will be some motion blurring in the direction of the car bump motion on you background layer as well.

Use the window roto on a solid layer, use that layer as a matte for the background footage. Track the motion of the window layer and apply it to a null, parent the solid layer to the null. To use the y variations in the track as a variable for a vector blur on the background footage you might have to poke around a bit to figure that one out.

TonyTony

[Tony Kloiber]

OK,
The short term helpful stuff first….
A pixel is not a pixel. They come in differing aspect ratios. 16×9 will be anamorphic (taller and thiner for those look at things from a relative perceptive) a greater ratio H to W. You can shoot and key a 4×3 clip and then put it into a 16×9 comp but pay attention to the pixel aspect ratios.

Also the new P2 camera doesn’t narrow it down enough for me but, if you mean the HVX-200, the camera can shoot in many formats including DVCProHD which will give you a lot more pixels to work with.

The long term….
[Alex Kuzelicki] ” I noticed that the 16:9 setting of our practice camera actually made getting good shots a lot harder than the 4:3 setting. For example, someone flying up and out of frame really quickly is much easier to catch in 4:3 because you have so much more of a frame to work with. “

Practice camera? If you not using the same camera and lens your going to seeing something different than the practice camera.

“More frame to work with” is a function of focal length, imaging area and distance to subject if the 16×9 frame is to tight move back. All things being equal a 16×9 frame will have more room side to side and the same top to bottom.

TonyTony

[Tony Kloiber]

[ZebSophia] “720 x 480 (1.20), Separating (Lower)
Millions of Colors
DV/DVCPRO – NTSC
48.000kHz / 16 bit / Stereo
“

Well there is the info you need to know. Set your AE comp to those settings (or use the preset) and animate away when you come to the point of rendering make sure your output and render templates (Edit>templates) are set to these settings as well.

If what your doing is going to be layerd over footage that is in FCP use a setting in the output template that has an alpha channel (i.e. preset -losses with alpha).

TonyTony

[Tony Kloiber]

What version you using? For me (AE7) I command d the comp and it gets the same name with a number added to it (comp > comp2) and it is right below the comp I duplicated. My project window is also sorting by name.

TonyTony

[Tony Kloiber]

Once you have gone and set your tabs as you like save the workspace.
Window>Workspace>New Workspace

TonyTony

[Tony Kloiber]

Good idea Dave

TonyTony