CG Jam

Two point tracking in Blender

We learned last time how to grab movement information from our clip, today i will show you how to use rotational movement from your footage to accurately place layers into your videos.

download the files.

Two points to Fake camera movement.

so, now you have the basics of using two point tracking from the video, lets make a slightly different use of the two tracks and use them to steal movement from a video clip and apply it to a static image to make it appear like a hand held film. this will also apply if you want to add camera movement to film that was shot on a tripod.

first things first then, track the clip in exactly the same way that i showed in the video

then in the right hand panel (press n to bring this up if its not already available) scroll down to 2D stabilisation.

Tick the check box this will open up a few options. first set your anchor frame to the first frame of your clip this will set a reference for the stabilisation. next we want to press the rotation and scale buttons so that they are both selected.

below that are two layer panels one for location data and the other for rotation.  select just the first marker then press the little plus button. this will add the selected track to the location layer. you only really need one track point for location, adding more will just average out the location data and wont give any real noticeable benefits.

Next shift select the other track and press the little plus by the “tracks for rotation” panel. this will add both selected tracks.

Thats it for tracking lets go straight to the compositor.

Putting it together.


open your image, whatever you choose. though it is probably best if it is a similar type to the video clip. so a street scene maybe.  now choose ADD> DISTORT> STABILISE 2D.

and connect the image into the stabilise 2d node. on the stabilise 2d node there is a video clip data block , click the icon and choose the video clip from the menu. 

now you’ll notice that your image has jumped in the frame and there are ugly black edges around it, so we need another type of distort node, again click ADD>DISTORT>SCALE.

Now increase the size of the image by adjusting the X and Y values until the black lines disappear from the edges.

Ok, so far so good, but generally the actual size of images are at a different aspect ratio to standard HD video. so to fix that we need another scale node. either add a new one or select the one you already have and press shift D to duplicate it. change the method from Relative to Render size then connect the output into the composite output node and your ready to render.

That’s the simple set up, if you wanted to place a 3d object or replace the sky in this image also  remember, this is originating from a static image so you just need to mix your layers together before the 2D stabilise node, then this node will do all the work animating all the layers together at the same time.

(image: Flagler street by Mark Averette, Wikipedia Commons CC Attribution 3.0)

Share this: