{"id":34281,"date":"2014-02-12T17:53:27","date_gmt":"2014-02-12T17:53:27","guid":{"rendered":"http:\/\/www.labri.fr\/perso\/barla\/blog\/?p=34281"},"modified":"2014-02-12T18:14:03","modified_gmt":"2014-02-12T18:14:03","slug":"prism2-workshop","status":"publish","type":"post","link":"https:\/\/www.labri.fr\/perso\/barla\/blog\/?p=34281","title":{"rendered":"PRISM2 2013"},"content":{"rendered":"<p id=\"top\" \/>\n<h3>Hannah Smithson<\/h3>\n<ul>\n<li>Represent diffuse-to-specular spectrum as a subspace (segment) of color space. <em>But the distribution of intermediately specular points is concentrated in diffuse-to-highlight transitions, which are sparsely distributed!<\/em><\/li>\n<li>Illuminant and reflectance changes involve different types of transformations in color space.<\/li>\n<\/ul>\n<h3>Katja Dorschner<\/h3>\n<ul>\n<li>Structure-from-motion is different for diffuse and specular of course, but it can also lead to false percepts such as when the axis of rotation is improperly assigned&#8230;\u00a0<em>With restricted aperture, specular flow even leads to a non-rigid, vicous-flow-like motion. How does the &#8220;anchoring&#8221; to the occluding contour turns that percept in that of the rigid motion of a specular object?<\/em><\/li>\n<li>Specular flow seems to be strongly dependent on silhouette motion, but at the same time it does not convey the same cues to motion. <em>If specular flow indeed requires the existence of a boundary, one way to study it in isolation would be to consider objects which silhouettes are invariant w.r.t. a given <em>(partial) rotation. The &#8220;internal&#8221; shape pattern curvatures could then be made to align more or less with the direction of motion&#8230;<\/em><\/em><\/li>\n<\/ul>\n<h3>Pascal Mamassian<\/h3>\n<ul>\n<li>Simple stereo experiments (on correspondence problem) show that complex scenes (with non-fronto-parallel objects) lead to difficulties in estimating true depths (slant seems to introduce ambiguities). <em>One potential explanation is that correspondence is grounded on more differential terms: flows would then be made to correspond (i.e. align) instead of individual elements.<\/em><\/li>\n<li>Light-from-above assumption is something that can be observed in experiments, and it shows a preference for light coming from the top-left. <em>Could left-side preference be due to cultural biases? If one shows patterns with different implied direction of illumination at once, then the 2D percept dominates; could be interesting to show multiple different patterns that force perception toward an unlikely percept (as light-from-below)&#8230;<\/em><\/li>\n<li>Combining stereo cues with other structural cues thus seems not as simple as basic weak fusion models might suggest. <em>Multiple cues could in reality be fused in different ways depending on where one looks at, like close or away from silhouettes from instance.<\/em><\/li>\n<\/ul>\n<h3>Roland Fleming<\/h3>\n<ul>\n<li>Orientation fields are more reliable and likely cues than mere image intensities. <em>It seems that portions of the field that are divergence free at multiple scales are more reliable. This should be observed experimentaly, and analyzed through the computation of flow properties of the first-order rendering equation.<\/em><\/li>\n<li>Even when applying a non-monotonic intensity remapping (producing a so-called weird shading), we are able to perceive shape. <em>Such weird shading admitedly looks a bit velvety. This suggests that the rate of intensity change orthogonal to the orientation field suggests a different material of course; but it also seems to change a bit the perception of curvature magnitude, without altering its direction or sign though&#8230;<\/em><\/li>\n<\/ul>\n<h3>Steven Zucker<\/h3>\n<ul>\n<li>Hue flow seems to provide a good separation between variations due to pigmentation and those due to shading. <em>Play with the hue flow (mod pi or 2pi?) and measure its _curl and divergence at different scales!<\/em><\/li>\n<li>The preference for smooth shading flows might be obtained through regularization across orientation columns. <em>Do we have any idea of the type of regularization going on there? Do these columns represent luminance of surface orientation (or something in-between)?<\/em><\/li>\n<li>Each configuration of shading flow corresponds to a family of potential surfaces. <em>Are there configurations that lead to broader\/thinner families? How to deal with faceted\/sharp objects (e.g., man-made, rocks)<\/em><\/li>\n<li>Shading and hue flow often provide different orientations that should make their separation possible. <em>But both flows seem to get aligned at occluding contours on physical objects. How exactly?<\/em><\/li>\n<\/ul>\n<h3>Sylvia Pont<\/h3>\n<ul>\n<li>Textiles produce split lobes that yield very uncommon appearances, which are due to their specific, regular macro\/microscopic structure. <em>Are these lobes perceived as properties of the material, or simply as if there was a different lighting environment (as with perfectly grooved surfaces)?\u00a0 How is this perceived in complex scenes?<\/em><\/li>\n<li>One goal of material design is to use projection to modify appearance directly in the hands of designers. Hence the need to investigate image mixing techniques to control lighting and material appearance. <em>This is a process that is actually pretty similar to what is done at the compositing stage. It raises the questions of how individual integral components of the rendering equation could be re-combined together to yield a different appearance, without creating weird effects or involuntary masking.<\/em><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Hannah Smithson Represent diffuse-to-specular spectrum as a subspace (segment) of color space. But the distribution of intermediately specular points is concentrated in diffuse-to-highlight transitions, which are sparsely distributed! Illuminant and reflectance changes involve different types of transformations in color space. Katja Dorschner Structure-from-motion is different for diffuse and specular of course, but it can also &#8230; <a title=\"PRISM2 2013\" class=\"read-more\" href=\"https:\/\/www.labri.fr\/perso\/barla\/blog\/?p=34281\" aria-label=\"Read more about PRISM2 2013\">Read more<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[81],"tags":[],"class_list":["post-34281","post","type-post","status-publish","format-standard","hentry","category-talks"],"_links":{"self":[{"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/posts\/34281","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=34281"}],"version-history":[{"count":2,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/posts\/34281\/revisions"}],"predecessor-version":[{"id":34411,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/posts\/34281\/revisions\/34411"}],"wp:attachment":[{"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=34281"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=34281"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=34281"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}