Jodie Davies-Thompson


Our research depends on helpful participants! Help us by taking part in experiments either in the lab in Nottingham (email: j.daviesthompson{at} or online on the links below.

<There are currently no online experiments>


Below is stimuli used in experiments. If you use them, please reference the appropriate paper.

The HVEM FIVE database (Face Identies, Viewpoints, & Emotions)
Face images used in the Face Training paper.These images were collected, created, and edited by Kim Fletcher and Jodie Davies-Thompson at the HVEM lab, University of British Columbia, Vancouver.

Original images: 12 male identities (Figure 1), 6 viewpoints (0, 10, 20, 30, 40, 50 degrees), 7 expressions (neutral, angry, disgust, fear, happy, sad, surprise) (Figure 2). This resulted in 504 original colour images.
1. Pairing: Within each set of 6 identities, face identities were paired (Pair_A1, Pair_A2, Pair_A3, Pair_B1, Pair_B2, Pair_B3). Each pair have approximate similar levels of discrimination difficulty. Pairs in Set A and Set B are also matched. Grouping and pairing of stimuli was chosen based on the performance of a group of subjects in a same/different task.
2. Editing: Images were first converted to grayscale. Moles etc, body, hair, and any residual facial hair was then removed in Photoshop CS5. For these images, download 'Starting Images Only'. Eye colour and face colour was matched between each face pair.
3. Within identity morphing: To create the different steps of facial expressions, each expression was morphed with the 'neutral' face image of the identity. For example, 0% neutral ID1 and 100% sad ID1 were morphed together to create 0% happy, 33% happy, 66% happy, 100% happy. This was repeated for each expression. Together with changes in viewpoint, this created 84 different 'starting' imagefor each identity, and 1,008 in total (Figure 2).
4. Morphing:
For the same viewpoint and expression, the two face identities within a pair were morphed together (ie. 10degrees, sad, ID1 and 10degrees, sad, ID2). These are in increments of 2.5% to create 40 images for each pair (ie. 0% ID1 / 100% ID2 , 2.5% ID1 / 97.5% ID2 ... 50% ID1 / 50% ID2 ... 97.5% ID1 / 2.5% ID2 , 100% ID1 / 0% ID2). (Figure 3). All morphing was done using FantaMorph. This resulted in 40,320 images.

5. Editing: After morphing, each face image was then edited to remove any 'ghosting'. For these images, download 'Entire Collection'.

< Please email me if you require the original color photographs >

Figure 1. The 12 different male face identities


Figure 2. The different viewpoints and expressions for each identity.
different images


Figure 3. Example images from the morph series of an identity pair
morph series