FaceBuilder

FaceBuilder can be used to create 3D models of human faces and heads with a few photos. The models can be later used for geometry tracking with GeoTracker or facial tracking with FaceTracker in Nuke and of course, they can be exported for further modifications in any other 3D software.

To create a model connect bg input to a sequence with photos, create and connect camera node with sensor and focal length settings matching the real camera settings and then start pinning the model pressing Center Geo button on the toolbar.

Inputs

Knobs

Toolbar

Name Label Description
center_geo center geo Place the object in 3D space so it will be visible in the frame
focal_length_view Current focal length
reset_to_neutral reset shape Reset the shape to the default state. The facial expression, position and all pins on all keyframes will be left intact.
reset_to_neutral_1 reset expression Reset the facial expression to neutral in the current keyframe
remove_pins remove pins Removes selected pins if there are any, or all pins if nothing is selected
pin_unpin pin/unpin Pin or unpin selected pins, applied to all pins if nothing is selected
spring_pins_back spring pins back Pins will be sticking to the surface of the object after dragging
estimate_focal_length estimate focal length Focal length estimation happens when you drag pins. Estimation starts once you have at least 4 pins. It’s automatically turned off after switching frames. It’s also not available if you chose to get the focal length from a connected Camera node.
meta_keyframes Keyframes toolbar

FaceBuilder Tab

Model

Name Label Description
non_neutral_expressions non-neutral expressions Helps when you want to match the face which is not in its neutral state, for example smiling, shouting, etc. Note that the preciseness of the model is better with neutral expressions.
copy_expression copy Copy current facial expression to clipboard
paste_expression paste Paste facial expression from clipboard to current frame. If there’s no keyframe, it’ll be created.
selected_uv_set uv The UV used for the model
model_type Choose a desired topology
auto_rigidity auto rigidity Automatic value prediction usually works best, but if you feel that the model is too rigid or too soft, you can switch it off and change the value manually.
model_rigidity rigidity High values make the model more rigid. For example if you need the model to be softer try to lower the value a bit.

Camera

Name Label Description
focal_length_mode

Focal length settings

  • focal length from Camera node: use the focal length set in the Camera node connected to the ‘cam’ input. Don’t forget to setup the proper film back dimensions in the Camera node. Focal length estimation can’t be used in this mode.
  • fixed focal length: use fixed focal length specified in the field (below or on the Results tab). One focal length is used for all keyframes. The film back dimensions will be taken from a default Camera node, if you want to override them — connect a Camera node with the desired film back dimensions, the focal length will not be taken from that node. Focal length estimation can be used with this mode.
  • varying focal length (zoom): use varying focal length specified in the field below. Focal length is stored within keyframes, it can be different for each keyframe. The film back dimensions will be taken from a default Camera node, if you want to override them — connect a Camera node with the desired film back dimensions, the focal length will not be taken from that node. Focal length estimation can be used with this mode.
focal_length focal length Current focal length

Render

Name Label Description
lit_wireframe lit wireframe The shape of the lit wireframe will be more apparent due to darkening of the furthest regions.
adaptive_opacity adaptive opacity Opacity will be reduced for dense regions of the wireframe in 2D viewer, so the footage visibility will suffer less.
back_face_culling back-face culling Don’t render back-faced polygons in 2D viewer
textured_wireframe textured wireframe A texture will be used to color the wireframe. The default texture highlights the facial parts. You can connect your own texture to the ‘tex’ input.
pins_color normal pins Normal pin color
residuals_color Color of pin residual lines
unpinned_pins_color unpinned pins Color of unpinned pins
selected_pins_color selected pins Color of selected pins
wireframe_color wireframe Wireframe color

Results Tab

Name Label Description
translate Geometry translation
rotate Geometry rotation
global_fixed_rx rx Lock X-axis rotation
global_fixed_ry ry Lock Y-axis rotation
global_fixed_rz rz Lock Z-axis rotation
global_fixed_tx tx Lock X-axis translation
global_fixed_ty ty Lock Y-axis translation
global_fixed_tz tz Lock Z-axis translation
unbreak_rotation unbreak rotation Fix rotation making it continious and removing 360 degrees jumps
scale scale Geometry input scale. All operations are performed with the scaled geometry. The scale is not restored on the output.
copy_expression copy Copy current facial expression to clipboard
paste_expression paste Paste facial expression from clipboard to current frame. If there’s no keyframe, it’ll be created.
output_transformed_geo output transformed geometry When on, the output geometry will be translated and rotated in accordance with the 2D position of the model in keyframes relative to the camera. You may need it if you want to project the texture with Project3D or TextureBuilder.

Export

Name Label Description
export_mode

Results export mode

  • TransformGeo: create a TransformGeo node containing the positions of the head in 3D relative to the input camera. If you are using the default camera or the custom focal length mode the new Camera node will also be created.
  • FaceTracker: create a FaceTracker node containing the positions of the head in 3D relative to the input camera. If you are using non- neutral expressions, they’re also copied (but not linked).
  • Axis: create an Axis node containing the positions of the head in 3D relative to the input camera. If you are using the default camera or the custom focal length mode the new Camera node will also be created.
  • Camera: create a Camera node. Only makes sense if you’re using the default camera or the focal length estimation. Doesn’t contain any movement data.
  • Translated Camera: create a Camera node containing the positions of the camera as if the head was static and the camera moved
export export Create a node of the selected type
link_output link output The exported node will be linked to GeoTracker, so when you change someting in GeoTracker the exported node will be updated. Switch it off if you need to embed all data into the exported node, to use in another script for example.
cam_rotate Inverted rotation. When coupled with inverted translation can be used as the rotation of a geometry-tracked camera.
cam_translate Inverted translation. When coupled with inverted rotation can be used as the translation of a geometry-tracked camera.

Last update: Fri Aug 7 09:49:48 2020