I’ve recently been playing with a bit of software called FaceGen, and it’s basically awesome. As you might expect it’s a piece of 3D modelling software which is specialised in producing human face stimuli. You can either start off with a randomly-generated example, or upload your own (or someone else’s) picture, which the software can then extrapolate and model in 3D. The 3D model can then be modified to your heart’s content along various parameters – age, sex, race, emotional expression etc. etc. It really is an awesomely powerful piece of software, and pretty easy to use too, with the interface mostly based around a set of sliders for manipulating the various dimensions of the stimulus.
Here’s a brief video which gives an overview of some of the features:
The full version of FaceGen costs $299, but there is a free demo version that you can play with, available here.
Face stimuli have always posed a problem for researchers, and historically there have basically been two choices. The first is to use a standardised face-set such as the Ekman faces:
These stimuli have the benefit of being naturalistic, i.e. they are of real people, but several significant drawbacks. These face sets are often idiosyncratic in various ways and may not have all the facial expressions you might need, for all the picture subjects. In addition, they’re often not well-balanced in terms of race, age, sex etc. In particular the ‘classic’ Ekman face set is looking very dated and frankly, pretty ghastly, these days. Another more recent example would be the NimStim face set.
The second option is to use schematic or computer-generated faces such as these from this paper:
These have the big benefit of precise experimental controllability, but the obvious drawback that they aren’t very naturalistic at all.
The FaceGen software seems to offer the best of both worlds, in that you can create an almost infinite variety of precisely-produced images and easily control for confounding factors like age and race, while at the same time the stimuli it produces are pretty naturalistic – particularly so, if you import ‘real’ pictures and then modify them. I’m currently setting up an experiment which will use some face stimuli and I’m almost certainly going to use stimuli produced using FaceGen.
Computing for Psychologists 
FaceGen – 3D modelling software for faces
Jun 9
Posted by Matt Wall
I’ve recently been playing with a bit of software called FaceGen, and it’s basically awesome. As you might expect it’s a piece of 3D modelling software which is specialised in producing human face stimuli. You can either start off with a randomly-generated example, or upload your own (or someone else’s) picture, which the software can then extrapolate and model in 3D. The 3D model can then be modified to your heart’s content along various parameters – age, sex, race, emotional expression etc. etc. It really is an awesomely powerful piece of software, and pretty easy to use too, with the interface mostly based around a set of sliders for manipulating the various dimensions of the stimulus.
Here’s a brief video which gives an overview of some of the features:
The full version of FaceGen costs $299, but there is a free demo version that you can play with, available here.
Face stimuli have always posed a problem for researchers, and historically there have basically been two choices. The first is to use a standardised face-set such as the Ekman faces:
These stimuli have the benefit of being naturalistic, i.e. they are of real people, but several significant drawbacks. These face sets are often idiosyncratic in various ways and may not have all the facial expressions you might need, for all the picture subjects. In addition, they’re often not well-balanced in terms of race, age, sex etc. In particular the ‘classic’ Ekman face set is looking very dated and frankly, pretty ghastly, these days. Another more recent example would be the NimStim face set.
The second option is to use schematic or computer-generated faces such as these from this paper:
These have the big benefit of precise experimental controllability, but the obvious drawback that they aren’t very naturalistic at all.
The FaceGen software seems to offer the best of both worlds, in that you can create an almost infinite variety of precisely-produced images and easily control for confounding factors like age and race, while at the same time the stimuli it produces are pretty naturalistic – particularly so, if you import ‘real’ pictures and then modify them. I’m currently setting up an experiment which will use some face stimuli and I’m almost certainly going to use stimuli produced using FaceGen.
For more on faces in psychology research see my previous post on face-morphing. TTFN!
Posted in Commentary, Cool new tech, Experimental techniques, Software
5 Comments
Tags: 3D modelling, age, emotion, face perception, FaceGen, faces, race, sex, stimuli