Compositing: An introduction to PFTrack

For our compositing assignment Alec recommended that we use or at least look into using PFtrack from Pixel Farm.

Considering I’ve never used the software before I decided to look up some tutorials recommended for beginners of the matchmoving and came across this really nice series of lessons from Digital Tutors: Your first Day In PFtrack


First Three lessons cover the interface of Pftrack: Such as setting up our project files, setting up preferences suited to your hardware specifications and understanding the node base Tree system

Summarised notes:


Track save automatically to a specified directory on your PC however you cannot work within PFtrack without a valid project.

Tree View: Allows you to build node based trees with our pipeline.

Canvas: Displays our project and workflow on screen such as showing the trackers in our scene.

  • Displays our 3D space with our solved 2D information

Project Window: Allows us to set up our project.

Navigator: Allows us to navigate to ours file paths, from which we can import our footage.

If we click into the media administrator we see the interface changes slightly. As seen below:


We still have our Tree view where we can insert our nodes, however as you can see we are now presented with two new windows with the interface.

  • The Media Bins and the Navigator

Navigator: Allows us to navigate through our systems file directories in order to obtain the desired file to be imported into our project.

Media Bin: Where we can insert the footage from the navigator that we require for our project to be edited within PFtrack.

Setting up our  Program Preferences

Within PFtrack, we can edit our preferences by going into the little ‘GEAR’ Icon in the top right hand side of the screen, as shown below:


And it within preferences that we can set up particular parts of PFtrack to better suit the specs of our PC. Such as Cache size, Tracker options,Export settings and even the scene units, where we can set up the scene within our project to match the units of our imported footage.

I modified my cache size by increasing the limit from 2GB to 6GB giving me more space to work with.

With our footage loaded in PFtrack we can start to edit it in the media admin menu. 

Here we can de-interlace our footage (if the footage is interlaced) and modify the camera pre-sets PFtrack has automatically given us.

Modifying you’re RAW footage in this way can be beneficial as it is capable of ‘cleaning’ your footage within PFTrack.

Whenever you have imported the footage required for your project, I a good Idea to cache the footage with your scene.


You can do this by  selecting the small ‘C’ symbol located nest to the timeline on the lower left of the screen.

Doing this will load the footage information, (frame rate,etc) into the cache memory of PFTrack. As a result the percentage value of the cache used will increase. Regardless, caching your footage will allowing it to play smoothly within PFtrack without any buffering or delays taking place.

Personal Note

when recording footage, try not get a lot of lens distortion or blur in it.

When it comes to tracking the scene it may cause difficulties, because trackers have an easier time linking onto defined and contrasting details within the footage. Distortion and blur affects this greatly.

However we can fix any distortion or blur by creating an ‘Image Manipulator Node’ and applying it to our footage within our tree view.

Image Manipulator.jpg

Settings available through the Image Manipulator Node

Its in the Image Manipulator Node that we can minimise the effects of distortion on our footage, by editing  features such as the pixel density and sharpness, gamma, colour channels, maker enhancements and contrast.

I particularly like this feature because it allows you to detect which colour channel is causing the most distortion within the footage, which means you can make specific adjustments to only that colour channel as opposed to the footage entirely which could cause problems later on.

In addition to the Image Manipulator Node you can also use the Undistort Node to help clean up your footage.

The undistort Node allows us to lay out lines that defines to the software where certain lines within the footage are currently at and where they should be.


Trackers enable use to supply our PC’s with enough X and Y axis information allowing it the figure out the 3D scene within our footage.

When tracking we have two options, Auto track or user track.

Despite being different nodes within PFTrack, both allow us to edit the parameters within the node that defines how fast or accurate we want to track to be.

So for example, under ‘Search mode’ when can change the setting from Best Speed to Best Accuracy, making this change would mean our track would take longer but we would get a more accurate result in the end. If our footage contains a lot of camera based rotation,scale or skew changes we can enable these deformers under the Nodes settings. we can also increase or decrease the failure threshold of the track, meaning if any trackers go beyond the failure threshold, they will fail. This is useful as it results in a more accurate track in the end.

Automatic Tracking

The software analyses the footage provided and determines on its own accord where trackers should be placed based on contrasting details.

User Tracking

Allows the user to determine precisely where a tracker is placed within the footage based on their own preference.

In addition to using user tracks, we are provided with a really helpful feature called the track Window.

track window

The track window shows us a close up image in real-time of where the tracker will exist within the footage. This can be used for making more precise and specific tracker placements.

Key Term: A soft track doesn’t stay true to the footage its tracking, we can clean this in our footage by trimming the error information provided with the track’s information. (seen below)


Camera Solver

The camera solver node will produce the most accurate camera animation within PFTrack based off the 3D scene information along with the most accurate solution of the 3D space obtained from the footage.

Within the Camera Solver node we can smooth out the movement of the camera by changing its Translation and rotation attributes to Med, Smooth.

We then click the Solve all button located to the bottom right of the Camera Solver attribute.

Orientating The Scene

Orienting the scene allows the us to determine the axis orientation and scale of the scene. By adjusting its rotation,translate or scale attributes.

To orientate a scene we select Marquee under the Orient scene node attributes, and with a tracker selected, hit Set origin   (highlighted below in the yellow)

marquee_set origin.jpg

This causes our scene to orientate itself around the selected tracker.

When adjusting the orientation of the scene, try to line up the horizon line of the orient scene node to the horizon line of the footage.

Exporting Our Scene

With our footage altered and tracked to meet our specific requirements we can export our footage say that it may be used in other software applications such a MAYA or NUKE.

To do this we simply add and Export Node to our footage in the tree editor.

Here we can choose from a variety of different file formats ranging from Afters Effects files to NUKE python files.

In my personal case I’d export the footage as a Autodesk MAYA 2011 (Ascii) file.

Setting Up Our MAYA scene:

Before we start working with our tracked footage in Maya, we need to ensure that the workspace units of our maya scene matches the workspace units of our PFtrack project, for example if  the  units within our workspace in PFtrack  were set to Metres then we should also change the  the  units of our Maya scene to metres also. This will ensure that both our tracked information matches the scale of our maya scene.


Matching Scene Units

We may also want to change our camera settings within maya to match the setting of our tracked camera.

Alec gave us some good advice regarding how to go about  setting up or adjusting the focal length of of tracked camera within Maya.

“Setup a camera with the same focal length (check the camera make for the multiplier to use. E.g. 35mm on a cannon 600D has a multiplier of 1.6 so actual focal length is 35 x 1.6 = 56mm”

Cheers Alec!

Editing the Near Clip and Far clip

Near Clip: If we have an object that’s distance value is lower than the near clip’s value then the camera will be blind to it

Far Clip: If we have an object that’s distance value is larger than the far clips value the camera will either clip it or not see it.

The values you set for each depends on the scale of the geometry with the maya scene.

Really enjoyed watching the tutorials provided by Digital Tutors, personally, as a beginner to the compositing process I thought each stage was explained really well, taking us through the basics of the interface to the slightly more complicated tasks of creating a decent piece of tracked footage. thinks its a good place to start for the team when compositing their footage.

Can’t wait to start!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s