Threads

All threads

All of the threads for this selected topic

I was a bit unclear about the coplanar ratio hypothesis. The Gilchrist reading described that according to the coplanar ratio hypothesis, “the perceived lightness of the target is governed by the luminance relationship between the target and whatever regions are seen as coplanar.” In terms of the relevance of this hypothesis in art and vision, I was hoping we could clarify what it means for something to be “seen as coplanar”. In some of the demos from last weeks class it felt as though I may perceive objects as coplanar even if in reality they were not as a result of the depth, shadow, and motion cues I was receiving. Would this fake co-planarity still influence the lightness of the target?

5 replies | 5 unread

Purves et al (2001) describe an empirical theory for vision. According to this theory, it seems like people with very little visual experience (e.g. newborns) would have different visual percepts than people with a lot of visual experience. How much experience is needed to have veridical visual percepts?

1 reply | 1 unread

Why is it easy for some people to see stereograms and hard for others? Is this something that can be trained?

1 reply | 1 unread

Are there neuroimaging studies on pictorial depth?

1 reply | 1 unread

According to the reading by Purves et al., “visual percepts are based on patterns of reflex neural activity shaped entirely by the past success” and that images we see are in response to similar retinal stimulus, “rather than the physical properties of the relevant objects.” Does this mean that individuals perceive “luminance” differently due to the different exposures to previous visual stimulus? Are we then able to train what an individual can perceive with repetitive exposure?

What can we learn about the far field from pictoral space (and vice versa)? What are the next big questions that we should be asking?

Purves et, el. claim that what we "see" is based on patterns of reflex neural activity that is determined by things we have seen in the past that evoked a similar stimulus. In this case, does it mean that it is likely that people are seeing things slightly differently from others based on their visual history? If so how is it possible to design for a certain end result (as Koenderink et, al. mention) if everyone sees things differently?

Why would it be important for vision to process depth perception before lightness perception?

Without binocular stereopsis (e.g. for “prey” type animals with eyes on the sides of their head) it motion necessary to interpret depth? Would such animals perceive any depth if looking at a still scene/ with no eye movements?

It seems like most state-of-the-art computer vision neural networks focus on monocular vision. (Binocular vision is somewhat implicitly considered via image transformations in training datasets, but the resulting two images are usually fed to the model as separate training / inference instances.) This makes sense when using these algorithms to understand pictoral space of images IN our world, but would models with (simultaneous) binocular vision better “understand” the visual space OF our world? (PS - Is anyone familiar with ongoing research of models that use binocular vision?)

what factors enhance or diminish the dominance of the pictorial mode? In Figure 1 of Gauge Fields in Pictorial Space, what factors could 'guarantee' (is that even a possibility?) the viewer perceives the cube as pointing towards the viewer instead of a flat object.

Submit New Thread

Active thread

Only the replies to the one thread you selected

Select a thread on the left to view the conversation.