Histograms talk updated

The version that’s now online is fairly close to a solid draft of the slides. Let’s see how the talk goes on tomorrow’s test run over lunch. (If you’re around and looking for free pizza, you know where to go!)

VP candidates experience: see for yourself

In the argument about Sarah Palin’s experience, we see a lot of rhetoric, a lot of name-calling, but very little real data. Lawrence Lessig produced a very nice 12 minute video describing the experience of all VP candidates. A commenter then created a spreadsheet in Google spreadsheet with the combined experience of all presidents and VP candidates Joe Biden and Sarah Palin. I decided to brush up on my Javascript and create an interactive plot using Google’s fancy new visualization API. All credit goes to mgmcmahon over at blip.tv for putting all the data together.

So, instead of simply arguing about whether Sarah Palin’s experience is comparable to other VP candidates or not, let’s first look at the data. Click here to see the plot (wordpress does not allow iframes). Here’s an example:

Sorry for the ugly colors - 9 categories is a lot to cram into RGB and still be both nice-looking and informative. Also, either the legend is huge or the names become illegible (I can’t control that, thank you very much, Google). Click on the checkboxes to include or exclude roles you want to consider or disregard. Instead of trying to convince you one way or another, I’ll just suggest a few different things. Try removing all legislative experience, then leaving only legislative experience. Now do the same with executive experience. What can you see?

Talk at Vis 2008

I want to try something different for this year’s Vis. I’ll be giving a talk on one of our papers, Revisiting Histograms and Isosurface Statistics. Instead of waiting until after the conference to post the slides, I will constantly post updates of the slides as I work on them. This way, you can both have an idea of what I will be talking about and let me know if I’m missing something you would like to see there.

So, if you have read the paper, are interested in the talk slides, and have any particular request, send me an email!

Vis 2008 Papers: Continuous Scatterplots

The list of accepted papers for Vis 2008 has been out for a while now, but I only now have time to talk a little bit about some of the papers I read. I’ll start with Bachthaler and Weiskopf’s Continuous Scatterplots, a really cool paper that tackles a problem we also have been looking at for a while now.

The problem setting is as follows: in visualization, we are typically interested in computing the joint histogram of a dataset. If you think of a simple dataset as a mapping , where k is the number of features, then what we want to visualize is a histogram where we pick two of the dimensions of the data space R^k and look at the joint distribution. This is the well-known scatterplot. Scatterplots work pretty well for sparse data, but for a dense sampling of volume data, they tend to look like that.

What Bachthaler and Weiskopf noticed is that we should not sample the data at all: we can compute a continuous function that is the limit of dense sampling. Even better, they show that the function can be computed with a traditional scivis technique: projected tetrahedra with an emission-only additive optical model. This means it can be very quicklycomputed on a GPU. The result looks much, much better.

Their main mathematical result is the need to carefully compute the tetrahedron density in PT: if the values in data space are close together, the tetrahedron should be “dense”, because, intuitively, there’s a lot of domain space packed into a small part of the data space (ie, a comparatively large part of in it). Their argument involves mass conservation and changes of densities in mappings between these different spaces. This turns out to be essentially the same argument that we make in our own paper, but we have a totally different application. It’s somewhat surprising that the same idea pops up for these two different places (and that it had not been previously noticed in the literature). I think it’s because this density change problem only happens in mappings between continuous spaces: it’s one of those tricky places where probability density functions are fundamentally different from probability mass functions (unless you’re a measure theorist and thinks of pmf’s as Dirac deltas).

I also think it’s kind of unfortunate that Bachthaler and Weiskopf’s paper ended up at the “Information Visualization” section (kind of like our paper from last year at the “Navigating in Parameter Space” section). However, having a “Information Visualization” section in the Visualization conference, to me, is just evidence for the silliness about the Vis and InfoVis dichotomy. Why is it there in the first place? Aren’t our shared interests stronger than our differences? And what are our differences, really?

First and last post about politics

This. Stop if you’re offended by completely justified profanity.

I feel left out

These look awesome:

Vibram Fivefingers

Too bad I can’t wear them because of my webbed toes (ewww, webbed
toes!)

… Okay you southpaws, I now get it.

Open publishing, good. Bentham Open Journals, not so much

Open publishing is a fantastic idea. However, this move by Bentham Science makes them look like Manutius Publishing. I just got an email from them that starts like this:

“Dear Dr. Scheidegger:
Bentham Open is one of the …”

Not only have they gotten the doctor thing just completely wrong, but I have been asked to join the board because of my “eminent contributions in the field of virtual reality”. However, later in the email, they write: “Should you be interested in being and Editorial Board Member, then please could you send us your CV and list of publications for consideration [sic]“. Wait, wasn’t I selected because of my eminent contributions? And do they pick virtual reality grad students to be on the board of their vis journals?

I can recognize at least one name from the current board. If it’s true (there are no affiliations and only abbreviated names, so I can’t be sure), that’s unfortunate. Also, stop calling your model Open Access. Pay-to-publish is NOT open access. I agree with Mihai Patrascu’s feeling: from first impressions, this is a case of a creative publisher trying to trick us into giving them money. I have to wonder how many people they unsuccessfully spammed before they got to grad students…

The brain’s native architecture

Marcelo Herreshoff over at Overcoming Bias claims that visualizing math is easier than symbol manipulation because “visualizing things is part of the brain’s native architecture”. It’s great to see other people making this point, but then the question is: are we (the vis people) out of a job as soon as we convince the rest of the world of this? Reaching over to the data producers at the other side of the fence is important: it gets our work noticed, and it really does help them deal with all the data. However, it’s sort of obvious we should be visualizing data. The important question is: how do we do it well?

What’s exiting is scivis, I think, is finding ways that make the visualizations obvious. I love the idea of digging deep into the math of a problem, and then find something that makes sense to look at. After you have the infrastructure, visualizing the data should be simple. Here’s a great example in CFD: when two fluids mix together, how do you tell that something is a bubble, and how can you look at a coherent piece of a bubble as a simulation progresses? Even though we can usually tell right away, it’s much harder to describe one to a computer.

The answer to that question comes from Laney et al’s paper a couple of years back. It requires some math, but the results are surprisingly “simple”: the bubbles segmented with topological persistence (coming from Morse theory) look just like they should. Even better, they can then do quantitative analysis on the number of bubbles, etc. By the “I would love to have been on a paper” measure of paper quality, this one is right up there for me. Solid math, great results, and pretty pictures!

SIGGRAPH 2008, and image editing in the gradient domain

I just got back from my last visit to the west coast, this time for SIGGRAPH 2008. Unlike the other years I’ve been at SIGGRAPH, this time I was working, so I didn’t really have that much time to sit through technical talks. However, I did see a few talks and flipped through a few of the papers on the proceedings.

Let me start with the best talk I’ve seen in a long time. Real-time Gradient-domain Painting, J. McCann and N. Pollard. The presentation consisted of many, many live demos of the work and the techniques (using his own presentation software). This included intuitive explanations of the Poisson’s equation approach to image processing in the gradient domain, interactive diagrams of how multigrid solvers work, and other cool stuff. In particular, he wrote his own presentation software, and it just blows everything else I’ve seen out of the water (it reminded me of Alan Kay’s Squeak talks, only with decently rendered fonts). It’s hard to describe: if you can get your hands into the DVD video recordings, do so. It’s worth half an hour of your time.

The really refreshing part about the talk was that it had a great point of view: painting a small area at a time is a limitation of the physical world that’s just not there in a computer. We should do better than that, and they do it by letting the user paint desired edge changes in the image, and using the incredibly useful Poisson equation framework originally proposed by Perez et al. five years ago (yes, Adobe, I know about the healing brush). You should go check this paper out.

Back when I was an undergrad, I had the opportunity to monkey around with a really simple application of this method. At the time, Manuel Oliveira and my advisor were working on GPU relief mapping, and realized that the Doom 3 engine makes extensive use of normal maps, and that he needed depth maps instead for relief mapping. We quickly put together a solver that did a decent job of reconstructing the depth information. After a fair amount of hacking (of which I wasn’t a part, unfortunately), he and his collaborator now have a plugin for Doom 3 Quake 4 running with relief maps. Check it out (wmv, unfortunately), it’s pretty cool - they even get self-occlusion and shadowing right, in texture space.

The other cool SIGGRAPH papers I’ve seen or read about are Lipman et al.’s deformation paper and Snavely et al.’s (slashdotted) video enhancement paper. More on those soon.

Eventful day

This week I’m out of town again visiting JPL to work with Eric Gurrola, Giangi Sacco and Mark Simons on making VisTrails work with their interferometric synthetic aperture radar software (in short INSAR). Tomorrow we’re flying to Stanford for the last couple of days of this workshop where we’ll talk about the stuff we’ve been doing.

JPL is by itself awesome, and just so happens to be 10 miles away from the biggest quake in SoCal in 14 years. Things started shaking, and next thing I noticed I was hiding under a table. Eric, who was on the phone, (and has been living in California for a long time) didn’t even stop the conversation, just moved under his desk. About 20 seconds after, things stopped shaking as hard, with only a few jolts for the next minute or so. And then everyone went back to work.

The second event of the day was the (non)-arrival of my brother in Salt Lake City. He was supposed to arrive around 7:30PM, but he happened to be on the plane that blew a tire on JFK this morning. The plane made a safe emergency landing, and he’ll eventually get to Salt Lake City (Nando, if you’re reading this, send us email! We don’t know what flight you’re going to be on!)