As of the release of Analyzer 2.6, the ability to optionally track the Jaw and Cheeks has been added in Analyzer. This allows the user slightly more nuanced control of the various facial groups when Retargeting.
*Note - Jaw and Cheek tracking is an advanced feature of Analyzer 2.6/Retargeter 4.6. Autotrack will NOT work with the Jaw or Cheek face groups. These groups can be included in custom Tracking Models with the 'Import/Export Tracking Models' feature. Please understand the basics of Analyzer tracking before attempting to track and retarget these groups. Jaw and Cheek tracking are also not recommended for batch workflows.
*Important Note: As of the release of Analyzer 3.2, Intelligent Drag will now work with the Cheeks groups. The Jaw will not respond to Intelligent Drag, but is not necessary, as this group only contains three landmarks. Do not use Intelligent Drag with these groups if you are using Analyzer 3.1 or earlier. Using Intelligent Drag with these groups may cause Analyzer to crash.
Creating a new job with Jaw and Cheek tracking
When you create a new job in Analyzer, you will be presented with the New Job dialog. Import your video and choose your settings as you normally would. Before you click the 'Create' button, however, go to the 'Advanced' tab (see image below).
In the Advanced tab, you will see the following:
This is where you choose your Analysis Definition file. The Analysis Definition is what tells Analyzer which landmarks and face groups to use for tracking. Navigate to your Analyzer assets folder (by default, "C:Program FilesFacewareAnalyzerAssets") and select one of the three following files according to your needs:
AnalysisDefinition.xml - This is the default Analysis Definition. It includes the Eyes, Brows, and Mouth face groups.
AnalysisDefinition_withJaw.xml - Use this file is you wish to track the Jaw in addition to the default groups, but not the Cheeks. Includes the Eyes, Brows, Mouth, and Jaw face groups.
AnalysisDefinition_withJaw_withCheeks.xml - This files is for tracking both the Jaw and the Cheeks groups in addition to the default groups. Includes the Eyes, Brows, Mouth, Jaw, and Cheeks face groups.
Once you've selected your desired Analysis Definition, finish creating the job as normal.
Tracking the Jaw face group
The Jaw group includes three landmarks. These should be placed on the jawbone (or as near as is possible). See the image below for examples of proper landmark placement.
The three landmarks should ideally remain along the jawline and in the same basic 'V' shape for each training frame. Create the training frames, train & track like normal to create your results.
Notes and tips for Jaw tracking
The easiest way to keep the shape of the landmarks consistent is to create one training frame, select and copy the three landmarks, then paste them onto your next training frame. Select all of them and drag them together to the appropriate location on the frame.
Important shapes will include mouth closed, mouth wide open, mouth open with teeth touching, mouth open without teeth touching, side to side movement of the jaw. You can often find good places for Jaw training frames by looking at the ones you've created for the Mouth.
The Jaw group tends to require the most hand-fixing of all of the groups. You may be able to get a more consistent result by adding markers to the actor on the bottom of their jawline where the landmarks would go. If you use markers, do not place them too high above the jawline as the skin stretching caused by the mouth can add extraneous jaw movement.
Remember to lock the nose landmarks before working with the Jaw group.
Tracking the Cheeks face group
The Cheeks can be slightly difficult to track at first due to the lack of definitive places to put the landmarks. Despite this, the Cheeks tend to track very well once you have created a good set of training frames. The Cheeks landmarks are a bit harder and less obvious to place. Below are a few examples with some notes.
Due to the lack of definitive identifying marks on most actors' cheeks, there is a little more leeway in where the landmarks can go, but it's important to keep the general shape the same as shown in the images above. Specific notes on landmark placement are below. Please note that this is for setting up your first training frame. The actual placement will vary depending on how your actor is moving. See the above image for landmark references.
A (top landmarks) - These should be placed under the eye. The easiest way to pick a good spot is to choose a place along the actor's cheek bones. Alternatively, you can imagine a line going out to the left and right from the bridge of the actor's nose and place the landmarks along that line underneath the eyes.
B (top middle) - Placing these landmarks is the easiest. Simply put them alongside the outside of the nostrils as shown above. For subsequent training frames, do the same as the actor moves his/her nose.
C (bottom middle) - These landmarks go along the nasolabial folds (often called "laugh lines" or "smile lines"). Place the landmarks about halfway between landmarks B and D along the lines. If the actor does not have prominent laugh lines, simply follow an imaginary line from the sides of the nose to the corners of the mouth.
D (bottom) - The lowest landmarks rest along the nasolabial folds in line with the mouth. If your actor has prominent visible folds, draw an imaginary line out from the corners of the mouth. Where the mouth meets the folds, place the landmark. If you cannot see laugh lines in your shot, you can simply place the landmarks a short distance out from the corners of the mouth.
Notes and tips for Cheek tracking
As with all of the face groups, it's important to have consistent landmark placement on the Cheeks. As there tend to not be many distinguishing features in the correct areas, it's important to go back and forth between training frames to make sure each one has the correct placement. Use the CTRL + LEFT/RIGHT ARROW KEY shortcut to quickly skip back and forth between training frames.
We highly recommend using simple markers on the face when shooting if you know that you will be tracking the cheeks. This will greatly aid the user and the software in knowing definitively where to put the landmarks, particularly when those areas of the face move around. If shooting over multiple sessions, try to keep the marker placement consistent. An example of simple ink marker placement is below.
.png?Policy=eyJTdGF0ZW1lbnQiOlt7IlJlc291cmNlIjoiaHR0cHM6Ly9kemY4dnF2MjRlcWhnLmNsb3VkZnJvbnQubmV0L3VzZXJmaWxlcy8zMDQ0LzQ2MTIvY2tmaW5kZXIvaW1hZ2VzL3F1L2ltYWdlKDEpLnBuZyIsIkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MTc3NTc0MDc0MH19fV19&Signature=SEtaLFAO6zkUZX45RaknjF9qUCevGLYbrC9Xes4eLZCnV6XVrKshrbgyvJe~ELTCrITdK6DVtTllRgv3bz9qIPL26T0FrAAjiliLYLcM-RYGSxQVo-H2dLVY-iZNxQmJYQAp47BHZVohulnY70gXDbaYEZfLEnCaRm7HbqvtTbj2XYxqo33qu3NPx1gepDJcB1XeAl0fCTGn5omiejfyOPg6PCLAfE~2zlKrOG-zC6EfoiAbETO46YrG1KyjgNftbdLx0cTls7rrACfdaoAbrVqxVvQFsVlWK4QnJWKPtsTSjplVhtHb-12zQK8vxqi2mA9F0w7CWwMmylfv1Q7m9A__&Key-Pair-Id=K2TK3EG287XSFC)
.png?Policy=eyJTdGF0ZW1lbnQiOlt7IlJlc291cmNlIjoiaHR0cHM6Ly9kemY4dnF2MjRlcWhnLmNsb3VkZnJvbnQubmV0L3VzZXJmaWxlcy8zMDQ0LzQ2MTIvY2tmaW5kZXIvaW1hZ2VzL3F1L2ltYWdlKDMpLnBuZyIsIkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MTc3NTc0MDc0MH19fV19&Signature=Hme3JCEAB~R5lZ56riIWkDrdqcGl-xPOj6nQ5EJ4ipOYySY9DmsAGrX1EwAUCs5DKnRJkC6cdxtUbfKaAEh~v3CUoZcHCFHgrrR54KySZB~KkwAVSvKxtioKPcQClItBrZ55mAccmNvEHRPeZ8lb2foonPhazAOPxwEqGw~9ou3Gg8PuvqscuUoYtCHNHrWVPNoIBZVFgcKgomHj-K5rtLx6NU3~Qv9cdgo5VduvfAh0RhZ113-s0D732pWeZecA-JSXINgN11ldwtjFivUg7k-q1pve7mvr-yKqY9WHgiAFsaMvsqCiZJ475zhUsXM~MioPIiuzoApuAIoemOMXFQ__&Key-Pair-Id=K2TK3EG287XSFC)
