Alright, we’re back. The next big point: weird things happen at equiluminance. The main thesis is that depth perception is colorblind. Quick demo (neurobiologist using OpenGL – that this blows my mind says we have a long way to go still) adjusting luminance of rotating points shows that three-dimensional perception needs luminance contrast.
Now we’ve started using the red-blue glasses, so I can’t type as fast… Point: impressionists and later artists use colors as symbols, and luminance as shape cues. More programmatic demos! Equiluminance kills your motion perception.
‘The “what” system can see color, the “where” system cannot’. The equiluminant text is making everyone in the audience sick – you can hear the groaning. Kitaoka’s illusions, more motion sickness. Interestingly, looking at the far projection screens makes the effect far weaker, so there’s definitely a particular sweet spot for the scale.
Cool example of a Duccio painting with Penrose’s triangle, and Magritte’s inverted occlusion tricks. Ceci n’est pas une depth buffer – Magritte is awesome. Afterimage color effect demos – more groans. Signal processing tidbit: RCA used to transmit 4MHz of luminance, and only 2MHz of color.
Now into object recognition. We have cells for eye and mouth recognition. Not that surprising, but amazing nevertheless. The audience is being setup for a object-recognition visual hack by looking at this side-by-side image of Clinton and Bush for a long time, and then having a “object afterimage”. I’ll try to hack an animated gif with this effect – it was pretty awesome (HERE). She’s now showing an example of a drawing by an autistic child – it is really realistic, but fails to capture the “important” parts. Children, on the other hand, draw the “heads with legs” just fine. This blew my mind: transcranial magnetic stimulation of the brain makes everyone more artistic – knocking down the high-level object recognition part.
Now for the final part of the talk: stereopsis. Closing an eye helps shut down stereopsis. Have people looked at using brightness to help stereo projection? This could be a cool paper. She’s now rushing through the dyslexia relation to object recognition and drawing, since we’re running late. Many more artists have strabismus than the general population. Rembrandt is consistenty cross-eyed in his self-paintings, and this finding got published in the NEJM!
Overall, a talk in the tradition of (just) slightly off-topic Vis keynotes. I love it.