1/1
MF3D R1 Animations
Version 3 2020-03-01, 18:09
Version 2 2020-02-13, 03:14
Version 1 2019-06-12, 14:06
media
posted on 2020-03-01, 18:09 authored by Aidan MurphyAidan Murphy, David A. LeopoldThe Animation subset of the MF3D R1 stimulus set consists of a series of high quality animated video clips (3840 x 2160 resolution at 60 fps encoded in H.264 perceptually-lossless .mp4 file format). The dynamics of the main facial action sequences are based on estimated time courses extracted from video footage of real Rhesus monkeys' faces, but these animations represent just a fraction of the combinations of facial dynamics that are possible to generate using the 3D avatar model.
The accompanying spreadsheet (.csv file) lists the filenames and parameters of all videos in this set. Since the first and last frames of each main action sequence video for a given head azimuth angle ('Haz') are identical, multiple action sequences can be edited together to generate continuous novel combinations of actions (see MF3D_AnimationDemo.mp4). For this purpose, we provide a Python script for automated video editing using the open-source software Blender (www.blender.org) in the MF3D GitHub repository: https://github.com/MonkeyGone2Heaven/MF3D-Tools
For more details on how and why the stimuli were generated, please see the associated paper: Murphy AP & Leopold DA (2019) https://doi.org/10.1016/j.jneumeth.2019.06.001
