You have a couple of options to solve this. The first is easiest but may be less adaptable:
1) Make sure that the pre-comp with collapse transformations enabled has its 3D layer switch enabled as well.
2) Set only the Z scale of the pre-comp layer to 200%. This is the third number in the scale property and is only available when the layer is a 3D layer (see step 1).
The 200% value is derived from the fact that your main comp is 200% as wide as your pre-comp.
The more adaptable solution might be to:
1) Create a new camera in your pre-comp (the default camera is “50mm”).
2) Copy that camera layer.
3) Paste that camera into the main comp and then press the “Reset” button next to the Transform group header to reset the camera transforms to default.
This gives you a camera with the same zoom value as the camera in the pre-comp and sets it in the correct “default” position.
If you already have a camera animation in the pre-comp or the camera is not at the default position then this solution gets a little more complicated as you’ll have to account for the offset in the “center” position of the main comp. You could create a new 3D Null and set it to the center value of the pre-comp. In your example this would be [400, 360]. Then parent the pasted camera from the pre-comp to the null and set the null’s position to the main comp’s center point, in your example [800,360]. This should essentially ‘Reset’ the camera transforms without changing any pre-existing keyframes on the camera.
The null can be deleted after this is done.
Darby Edelen