Mario Kleiner

Semantic 3D motion retargeting for facial animation

We present a system for realistic facial animation that decomposes facial motion capture data into semantically meaningful motion channels based on the Facial Action Coding System. A captured performance is retargeted onto a morphable 3D face model based on a semantic correspondence between motion capture and 3D scan data. The resulting facial animation reveals a high level of realism by combining the high spatial resolution of a 3D scanner with the high temporal accuracy of motion capture data that accounts for subtle facial movements with sparse measurements.Such an animation system allows us to systematically investigate human perception of moving faces. It offers control over many aspects of the appearance of a dynamic face, while utilizing as much measured data as possible to avoid artistic biases. Using our animation system, we report results of an experiment that investigates the perceived naturalness of facial motion in a preference task. For expressions with small amounts of head motion, we find a benefit for our part-based generative animation system over an example-based approach that deforms the whole face at once.

Atypical brain activation patterns during a face-to-face joint attention game in adults with autism spectrum disorder

Joint attention behaviors include initiating ones own and responding to anothers bid for joint attention to an object, person, or topic. Joint attention abilities in autism are pervasively atypical, correlate with development of language and social abilities, and discriminate children with autism from other developmental disorders. Despite the importance of these behaviors, the neural correlates of joint attention in individuals with autism remain unclear. This paucity of data is likely due to the inherent challenge of acquiring data during a real‐time social interaction. We used a novel experimental set‐up in which participants engaged with an experimenter in an interactive face‐to‐face joint attention game during fMRI data acquisition. Both initiating and responding to joint attention behaviors were examined as well as a solo attention (SA) control condition. Participants included adults with autism spectrum disorder (ASD) (n = 13), a mean age‐ and sex‐matched neurotypical group (n = 14), and a separate group of neurotypical adults (n = 22). Significant differences were found between groups within social‐cognitive brain regions, including dorsal medial prefrontal cortex (dMPFC) and right posterior superior temporal sulcus (pSTS), during the RJA as compared to SA conditions. Region‐of‐interest analyses revealed a lack of signal differentiation between joint attention and control conditions within left pSTS and dMPFC in individuals with ASD. Within the pSTS, this lack of differentiation was characterized by reduced activation during joint attention and relative hyper‐activation during SA. These findings suggest a possible failure of developmental neural specialization within the STS and dMPFC to joint attention in ASD. Hum Brain Mapp 34:2511–2523, 2013.

Manipulating Video Sequences to Determine the Components of Conversational Facial Expressions

Communication plays a central role in everday life. During an average conversation, information is exchanged in a variety of ways, including through facial motion. Here, we employ a custom, model-based image manipulation technique to selectively “freez ” portions of a face in video recordings in order to determine the areas that are sufficient for proper recognition of nine conversational expressions. The results show that most expressions rely primarily on a single facial area to convey meaning with different expressions using different areas. The results also show that the combination of rigid head, eye, eyebrow, and mouth motions is sufficient to produce expressions that are as easy to recognize as the original, unmanipulated recordings. Finally, the results show that the manipulation technique introduced few perceptible artifacts into the altered video sequences. This fusion of psychophysics and computer graphics techniques provides not only fundamental insights into human perception and cognition, but also yields the basis for a systematic description of what needs to move in order to produce realistic, recognizable conversational facial animations.

Audiovisual interactions in binocular rivalry

When the two eyes are presented with dissimilar images, human observers report alternating percepts-a phenomenon coined binocular rivalry. These perceptual fluctuations reflect competition between the two visual inputs both at monocular and binocular processing stages. Here we investigated the influence of auditory stimulation on the temporal dynamics of binocular rivalry. In three psychophysics experiments, we investigated whether sounds that provide directionally congruent, incongruent, or non-motion information modulate the dominance periods of rivaling visual motion percepts. Visual stimuli were dichoptically presented random-dot kinematograms (RDKs) at different levels of motion coherence. The results show that directional motion sounds rather than auditory input per se influenced the temporal dynamics of binocular rivalry. In all experiments, motion sounds prolonged the dominance periods of the directionally congruent visual motion percept. In contrast, motion sounds abbreviated the suppression periods of the directionally congruent visual motion percepts only when they competed with directionally incongruent percepts. Therefore, analogous to visual contextual effects, auditory motion interacted primarily with consciously perceived visual input rather than visual input suppressed from awareness. Our findings suggest that auditory modulation of perceptual dominance times might be established in a top-down fashion by means of feedback mechanisms.

RTbox: A device for highly accurate response time measurements

Although computer keyboards and mice are frequently used in measuring response times (RTs), the accuracy of these measurements is quite low. Specialized RT collection devices must be used to obtain more accurate measurements. However, all the existing devices have some shortcomings. We have developed and implemented a new, commercially available device, the RTbox, for highly accurate RT measurements. The RTbox has its own microprocessor and high-resolution clock. It can record the identities and timing of button events with high accuracy, unaffected by potential timing uncertainty or biases during data transmission and processing in the host computer. It stores button events until the host computer chooses to retrieve them. The asynchronous storage greatly simplifies the design of user programs. The RTbox can also receive and record external signals as triggers and can measure RTs with respect to external events. The internal clock of the RTbox can be synchronized with the computer clock, so the device can be used without external triggers. A simple USB connection is sufficient to integrate the RTbox with any standard computer and operating system.

Steady-state responses in MEG demonstrate information integration within but not across the auditory and visual senses

To form a unified percept of our environment, the human brain integrates information within and across the senses. This MEG study investigated interactions within and between sensory modalities using a frequency analysis of steady-state responses that are elicited time-locked to periodically modulated stimuli. Critically, in the frequency domain, interactions between sensory signals are indexed by crossmodulation terms (i.e. the sums and differences of the fundamental frequencies). The 3 × 2 factorial design, manipulated (1) modality: auditory, visual or audiovisual (2) steady-state modulation: the auditory and visual signals were modulated only in one sensory feature (e.g. visual gratings modulated in luminance at 6 Hz) or in two features (e.g. tones modulated in frequency at 40 Hz & amplitude at 0.2 Hz). This design enabled us to investigate crossmodulation frequencies that are elicited when two stimulus features are modulated concurrently (i) in one sensory modality or (ii) in auditory and visual modalities. In support of within-modality integration, we reliably identified crossmodulation frequencies when two stimulus features in one sensory modality were modulated at different frequencies. In contrast, no crossmodulation frequencies were identified when information needed to be combined from auditory and visual modalities. The absence of audiovisual crossmodulation frequencies suggests that the previously reported audiovisual interactions in primary sensory areas may mediate low level spatiotemporal coincidence detection that is prominent for stimulus transients but less relevant for sustained SSR responses. In conclusion, our results indicate that information in SSRs is integrated over multiple time scales within but not across sensory modalities at the primary cortical level.

Naturalistic Stimulus Structure Determines the Integration of Audiovisual Looming Signals in Binocular Rivalry

Rapid integration of biologically relevant information is crucial for the survival of an organism. Most prominently, humans should be biased to attend and respond to looming stimuli that signal approaching danger (e.g. predator) and hence require rapid action. This psychophysics study used binocular rivalry to investigate the perceptual advantage of looming (relative to receding) visual signals (i.e. looming bias) and how this bias can be influenced by concurrent auditory looming/receding stimuli and the statistical structure of the auditory and visual signals. Subjects were dichoptically presented with looming/receding visual stimuli that were paired with looming or receding sounds. The visual signals conformed to two different statistical structures: (1) a ‘simple’ random-dot kinematogram showing a starfield and (2) a “naturalistic” visual Shepard stimulus. Likewise, the looming/receding sound was (1) a simple amplitude- and frequency-modulated (AM-FM) tone or (2) a complex Shepard tone. Our results show that the perceptual looming bias (i.e. the increase in dominance times for looming versus receding percepts) is amplified by looming sounds, yet reduced and even converted into a receding bias by receding sounds. Moreover, the influence of looming/receding sounds on the visual looming bias depends on the statistical structure of both the visual and auditory signals. It is enhanced when audiovisual signals are Shepard stimuli. In conclusion, visual perception prioritizes processing of biologically significant looming stimuli especially when paired with looming auditory signals. Critically, these audiovisual interactions are amplified for statistically complex signals that are more naturalistic and known to engage neural processing at multiple levels of the cortical hierarchy.

How believable are real faces? Towards a perceptual basis for conversational animation

Regardless of whether the humans involved are virtual or real, well-developed conversational skills are a necessity. The synthesis of interface agents that are not only understandable but also believable can be greatly aided by knowledge of which facial motions are perceptually necessary and sufficient for clear and believable conversational facial expressions. Here, we recorded several core conversational expressions (agreement, disagreement, happiness, sadness, thinking, and confusion) from several individuals, and then psychophysically determined the perceptual ambiguity and believability of the expressions. The results show that people can identify these expressions quite well, although there are some systematic patterns of confusion. People were also very confident of their identifications and found the expressions to be rather believable. The specific pattern of confusions and confidence ratings have strong implications for conversational animation. Finally, the present results provide the information necessary to begin a more fine-grained analysis of the core components of these expressions.

Using facial texture manipulation to study facial motion perception

For the creation of manipulated video footage, these extracted “texturemap movies” can now be automatically altered by standard image processing and video editing techniques in various ways like e.g., freezing, replacing or filtering parts of the texture, mixing texture parts from different sources or changing order and timing of “texture frames”. All manipulations to facial regions need to be defined only once on a reference head and texture for a specific experiment and can then be applied automatically to different recordings of different actors, because all head models are in correspondence to each other, saving a lot of manual setup work.

Investigating idiosyncratic facial dynamics with motion retargeting

Newer technology allows for more realistic virtual environments by providing visual image quality that is very similar to that in the real world, this includes adding in virtual self-animated avatars [Slater et al, 2010 PLoS ONE 5(5); Sanchez-Vives et al, 2010 PLoS ONE 5(4)]. To investigate the influence of relative size changes between the visual environment and the visual body, we immersed participants into a full cue virtual environment where they viewed a self-animated avatar from behind and at the same eye-height as the avatar. We systematically manipulated the size of the avatar and the size of the virtual room (which included familiar objects). Both before and after exposure to the virtual room and body, participants performed an action-based measurement and made verbal estimates about the size of self and the world. Additionally we measured their subjective sense of body ownership. The results indicate that the size of the self-representing avatar can change how the user perceives and interacts within the virtual environment. These results have implications for scientists interested in visual space perception and also could potentially be useful for creating positive visual illusions (ie the feeling of being in a more spacious room).Two experiments assessed the development of children’s part and configural (part-relational) processing in object recognition during adolescence. In total 280 school children aged 7–16 and 56 adults were tested in 3AFC tasks to judge the correct appearance of upright and inverted presented familiar animals, artifacts, and newly learned multi-part objects, which had been manipulated either in terms of individual parts or part relations. Manipulation of part relations was constrained to either metric (animals and artifacts) or categorical (multi-part objects) changes. For animals and artifacts, even the youngest children were close to adult levels for the correct recognition of an individual part change. By contrast, it was not until aged 11–12 that they achieved similar levels of performance with regard to altered metric part relations. For the newly-learned multipart objects, performance for categorical part-specific and part-relational changes was equivalent throughout the tested age range for upright presented stimuli. The results provide converging evidence, with studies of face recognition, for a surprisingly late consolidation of configural-metric relative to part-based object recognition.According to the functional approach to the perception of spatial layout, angular optic variables that indicate extents are scaled to the body and its action capabilities [cf Proffitt, 2006 Perspectives on Psychological Science 1(2) 110–122]. For example, reachable extents are perceived as a proportion of the maximum extent to which one can reach, and the apparent sizes of graspable objects are perceived as a proportion of the maximum extent that one can grasp (Linkenauger et al, 2009 Journal of Experimental Psychology: Human Perceptiion and Performance; 2010 Psychological Science). Therefore, apparent sizes and distances should be influenced by changing scaling aspects of the body. To test this notion, we immersed participants into a full cue virtual environment. Participants’ head, arm and hand movements were tracked and mapped onto a first-person, self-representing avatar in real time. We manipulated the participants’ visual information about their body by changing aspects of the self-avatar (hand size and arm length). Perceptual verbal and action judgments of the sizes and shapes of virtual objects’ (spheres and cubes) varied as a function of the hand/arm scaling factor. These findings provide support for a body-based approach to perception and highlight the impact of self-avatars’ bodily dimensions for users’ perceptions of space in virtual environments.