'Export Landmark Positions' Feature
Export Landmark Positions is a new feature in Analyzer PRO that allows you to export the position of every landmark (facial tracking point), from every frame of video, to an XML file. The exported 'Feature' file, will contain the Job's video meta data such as time-code, frame rate, etc., as well as the details for the position of each landmark across the entire video.
This marks the first time that Analyzer can export data that can be used outside of Faceware Retargeter.
How to use 'Export Landmark Positions'
With Analyzer 2.3+ installed, follow this simple tutorial to use the 'Export Landmark Positions' feature.
Open a Job in Analyzer PRO that has been tracked with either AutoTrack or the Pro Tracking features.
Click [File -> Export Landmark Positions]
Browse to a directory and enter a file-name for your Analyzer Feature file. Click 'Save'.
Using an Analyzer Feature File
The new Feature files give you a new avenue in which to make use of facial data from Analyzer. While it's only half of the amount of data that you'd get if you were using 'Parameterize' and creating animation in Retargeter, it can still be very useful data. You can drive characters, shaders, or even use the data in more complex ways such as scan matching.
Note: Faceware Technologies does not currently provide any scripts or tools for you to use the data from the Feature file. You'll need to create them yourself or use pre-existing tools.
Feature File Contents
Opening a Feature file in a text editor will show you a very large amount of data. This is because it's written the position of every facial landmark for every single frame of your video. This could be tens of thousands of lines or more if you're exporting data from a long video. Be prepared for this file to be large in size and sometimes take a little time to open in your text editor.
<META> & <VIDEO>
At the top of the file in the <meta> section you'll see the <video> tag. This tag keeps track of some useful information about your video. When integrating the Feature file into your pipeline, this data can be used to ensure you are using facial data in the right locations, with the right assets, and everything lines up with your project's time-code.
There will be a <frame> tag for every frame of video. Each <frame> tag contains the <markup_groups> and <landmarks> tags described below.
<MARKUP_GROUPS> & <LANDMARKS>
For each Face Group in Analyzer (currently 'Mouth, 'Eyes', and 'Brows'), you'll see listed all of the landmark positions in that Group. Remember, the (3) 'Nose' landmarks are present in all Face Groups. Each <landmark> tag contains the <texCoord> and <position> tags described below.
<TEXCOORD> & <POSITION>
The <texCoord> tag contains the U & V values for each Landmark. The values are normalized from 0 to 1 and the origin (0,0) is the TOP LEFT of the image.
Note: Because the origin is at the TOP, positive values will move down in the V space. For example, 0.25 is 25% of the way down the image from the top.
The <position> tag is left alone and is meant to be used within your pipeline for various use cases such as if you were mapping a Landmark coordinate to an object or movement in 3D space.
Using 'Export Landmark Positions' in AnalyzerBatch
AnalyzerBatch.exe contains the command, 'ExportLandmarkPositions'. To run it, follow the instructions below.
Open Windows Command Prompt
Change directory to 'C:/Program Files/Faceware/Analyzer' or add this directory to the Windows PATH variable.
Run the following command: 'AnalyzerBatch ExportLandmarkPositions <FwtFile> <XmlFile>'
e.g. AnalyzerBatch ExportLandmarkPositions "C:/MyFiles/AnalyzerJob_012.fwt" "C:/MyFiles/AnalyzerFeatureFile_012.xml"
Have questions or feedback about this feature?
Email us anytime at firstname.lastname@example.org or create a ticket here in our Knowledge Base.