Please disable Adblockers and enable JavaScript for domain CEWebS.cs.univie.ac.at! We have NO ADS, but they may interfere with some of our course material.
AnswersQ5
- Figure 5.4/Page 84 - Tamara Munzner writes, that Bar charts only encode one attribute using a linke mark and the visual channel of spatial position. But why is that? If i use Bar charts i still have two attributes displayed. One on the y-axis which usually should display a quantity of something (in case of bar-charts), and the x-axis is showing me the category which is still a attribute, or isn't it?
- discussed in long detail in class.
- Wieso wird bei den Channel Rankings bei den ordinal / quantitativen Daten die 3. Dimension beruecksichtigt, nicht jedoch bei den kategorischen Daten? Oder spielt der Hinweis 'add depth' genau auf diesen Umstand an?
- unklar was mit "add depth" gemeint ist. 3D im allgemeinen ist keine gute Idee (mit Ausnahmen für 'spatial' data). 3D ist nur erwähnt für den 'how much' channel der sich nicht für Categorical data eignet!
- Wie ergibt sich der Faktor n im 'Law of Stevens'? Beruht diese Zahl auf einer mathematischen Hypothese oder wurde diese empirisch eruiert?
- n ist eine Variable, die für jeden channel anders ist. Das Model hier ist ein exponentielles.
- Luminance: The author says that bacause of our low accuracy in perceiving, it must be considered whether noncontiguous regions have the same luminance because of contrast effects. Therefore, the number of discriminable steps should be small, typically less than 5. What are these "discriminable steps"? Is one "step" transition from a darker color to a brighter one?
- Yes, think of a color-steps like here: https://vis4.net/blog/wp-content/uploads/2013/09/Foto-522x288.jpg
- "Zu 5.5.2: Was sind bins?" and "What are actually "bins"? […] What are other examples of them beside line with its width and length?"
- "bin" ist simply a term to express discretization of the continuous scale -- see question above.
- Im Punkt 5.5.2 wird die „Discriminability“ beschrieben und das man bei einer kleinen Anzahl von bins die Beziehungen zu einander durch einer „Linewidth“ darstellen kann. Ebenfalls wird erwähnt, dass bei einer größeren Anzahl von bins dieses Verfahren nicht dafür geeignet ist. Welche Möglichkeiten bzw. Verfahren könnte man in dem Fall anwenden?
- Entweder sollte man einen anderen Channel wählen oder … kreative Ideen entwickeln :) [z.B. die Komplexität der Vis herunter schrauben.]
- What means "A single attribute can be encoded with vertical spatial position" in the context of visual chanels?
- Hmmm -- haben wir heute in der Vorlesung hoffentlich deutlich gemacht?!
- How can a visual channel be independent of the dimensionality of the geometric primitive if it encodes the size, doesn't make much sense for dots, the shape, even more so or, the orientation, even more so still?
- Why not? I can change the size of a 2D circular dot just as well as the size of a 3D circular dot?!
- Vertical position is a visual channel, horizontal position is another, but doesn't using them both make labeling the data points a lot harder, if the individual points are of interest, defeating the purpose of adding clarity via visual channels?
- well, there are always trade-offs and Tamara is always trying to give an idea about the scalability for specific visual encodings (in Chapter 7+8).
- I cant understand how saturation/lightness/stipple density is worse than curvature and volume. The number of bins they offer doesn't seem to differ much."How much more curved is this line compared to that line" is equal if not worse to "How much more intense is this red compared to that red" in our first lecture we used 3d volume as a negative example, since 3d stuff doesn't work that well in 2d space, which most vis utilize. Why is it ranked that high? Can we get a negative example for expressiveness?
- I don't have a good one right now. I'll ask Tamara :)
- How does one avoid giving an ordering to unordered data when showing it?
- By not using "how much" channels.
- Im Punkt 5.5.4 Popout wird beschrieben, wie zum Beispiel ein Roter Kreis innerhalb von 20 roten Quadraten visualisiert werden kann. Bei einer größeren Menge wird erwähnt, dass das serial search Verfahren dafür besser geeignet ist. Jedoch würde ich gern wissen, wie effizient dieses Verfahren bei einer wirklich großen Menge ist in der zum Beispiel viele Rote Quadrate, Rote Rechtecke oder Kreise vorhanden sind. Dies würde ja eine Weile dauern, bis genau ein vorhandenes Objekt zwischen mehreren raus gefiltert wird. Gibt es da noch weitere Verfahren?
- "besser geeignet" ist falsch. Besser geeignet wäre eine parallele Suche, aber die ist perzeptuell nicht immer möglich.
- "The popouts are also called in text, what for do we use them actually?" and "What is a visual popout and why is it not supported from a few chanels?" and "How do channels support popout individually?"
- Popouts could be used for identifying an item we search for. Many channels support them. I don't understand the last question. Please try to be precise (and elaborate) in your questions, otherwise I cannot give you points for them.
- What happens in Visual Encoding when the datasets are too large to understand from the side of the user?
- Often times we utilize the Shneiderman Mantra -- "Overview first …"
- How is the projection interpreted, with using the Dimension Reduction Method of projection?
- I don't understand your question … please elaborate.
- "Even though i could look it up, how can I imagine a 2.05D of reality in comparison to a 3D? Even though maybe hard to image, but how would then a 3D view of reality look like?" and ""We do not really live in 3D, or even 2 1/2D, we see in 2,05D." Das verstehe ich nicht ganz."
- I hope I clarified this in the lecture today, at least there were no questions! :)
- How does expressiveness and efectiveness influence the visual encoding?
- Again, I hope today's lecture helped in clarifying this.
Questions regardings figures
- "I don't quite understand Figure 5.15 on p.101. Is it a problem that picture on the left has to much colors, and is therefore undistinguishable?" and "Wie waere ein Alternativvorschlag fuer das Problem, welches in Figur 5.15 veranschaulicht wird? Ich nehme an, eine Erweiterung der Skala mit Helligkeit und / oder Saettigung wuerde keine wirklichen Vorteile bringen, eher Nachteile. Waere es moeglich, moegliche Alternativen im Kurs zu diskutieren?" and "Figure 5.15 habe ich nicht ganz verstanden." and "fig 5.15 (chromosomes) What would be a viable/optimal solution to convey the information in a similarily concise manner?"
- To really present alternatives, one has to dive into the application and figure out what people actually want to do! Perhaps they are "searching" for something that we could incorporate an interaction where we selectively highlight one 'Mouse' after the other. Maybe there is a hierarchy within the Mouse type we can exploit, etc. etc.
- Die Abbildung 5.18 ist mir nicht ganz klar. Meiner Meinung nach sind die Abbildungen gleich groß. Es ist schwer zu erkennen, jedoch finde ich das die graue Maske in der Abbildung B keinen großen Unterschied aufzeigt.
- Ich hoffe das war dann in der Vorlesung deutlicher!?
General Questions
- "A general question: how (much) is the perceptual/cognitive system influenced by individual or cultural aspects, and is this taken into account in the field of visualization?" and "Page 88/Steven's Law - should Steven's Law be so generalized? What about people who people who are blind? Do they receive the same amount of sensation like other people? Are there differences between people with different cultural background? Or differences between different peer groups (for example Age (old vs. young people))?"
- In terms of Weber's law -- this is universal. However, some of us can train to enhance their perception in particular ways, e.g. some folks have perfect pitch, etc.
- This question might go to deep, but if I want to calculate in color space, which of these systems, like RGB, HSL and so on, are better to use? and are they encoded in image files, so that I can use them like matrices?
- Which colour space to use is an issue of the purpose/application! RGB is perfectly fine to use in rendering. HSV is better for modelling, etc.
Letzte Änderung: 20.03.2014, 17:02 | 1391 Worte