Weekly Journal

One   Two   Three   Four   Five   Six   Seven   Eight   Nine   Ten   Eleven   Twelve   Thirteen   Fourteen   Fifteen   Sixteen

Week One

BC IS SO BEAUTIFUL!!

I got introduced to my mentor, Melanie, and her postdoc, Colin, and her grad students. It’s a very chill environment here, and everyone is so nice. The lab also has a bunch of fun toys in it (don’t worry; I’m working more than playing!) and there is apparently more stuff on the way.

It took a few days to familiarize myself with the project. In order to fully get introduced to it, I acted as a pilot subject for it, which was pretty cool. It really helped me understand how other people are going to react to it when they don’t have all the background knowledge about it that I have. The coding was actually written in VB (which I’m not so familiar with), and the program churns out logs left, right, and centre! Regardless, the software is pretty awesome, and, I can’t lie, lots of fun to play with! I also got to pilot test the prototype for the Eye tracker, which is from Colin’s company. That was really cool, and it was really interesting to use the calibrating software to manually calibrate the video it recorded while I was wearing it and then see where I was looking.

Melanie found a really interesting conference for me to apply to. It’s basically a weekend conference at the end of the summer where students present their research projects from the past year/summer. I need to work on my application a lot if I want it to be accepted.

Week Two

This week was a week of small projects.

We realised that the colour matching task might never end if the participant is colourblind, so I put together a slideshow test (based off of the Ishihara colour test) that we will give to each participant at the beginning to make sure that he/she is not red/green colourblind. I took a lot of footage of the apparatus being used and screen shots of the program and edited it into a two minute Video Showcase (with audio!) that will accompany the “Call for Papers” application for the 2009 CHI conference. I also spent some time looking through videos and papers from the 2008 conference, and there were some really interesting things (some more practical than others)!

I finally got my website up and running, which was great because it helped me remember how much fun CSS and HTML are. If I have any spare time this summer, I’m going to teach myself PHP and SQL because I feel like those are pretty useful to know. I also made an online recruitment flyer for the study and put it up on my UVic webspace so that it will be easy to start recruiting participants when we are ready: http://webhome.cs.uvic.ca/~rdreezer/

I also got introduced to SPSS and got to break out some STATS knowledge from last semester. Some of the data from the Pilot tests were put into Boxplots so that we could get an overall view of how the experiment works, and get some sense of the results. But this isn’t a spoiler, so I won’t post the findings here.

Week Three

We’re waiting for the delivery of the new lens for the Eytracker (so that it doesn’t dry out the user’s eye). In the meantime, there’s not a lot left to do on the project. When the Eytracker is ready to go, we will need to do some more pilot tests – especially since we haven’t really tried testing the experiment while wearing the Eytracker. These pilots will also help me get comfortable with actually performing the test on people.

Because this project is stalled a little bit, Melanie has me starting a new side project. A few years ago she did an experiment about how people were able to locate areas of high concentrations of points on various visualization maps – whether it was easier for them to see it in 3D or flat or a specific colour scheme… Basically, multivariable data is read from a table and put into a 2D or 3D landscape map and a random dimension is used to control the z-coordinate. I am adapting this code (yay Java!) so that the z-dimension of the landscape (which is basically the height) corresponds to how dense the area is with points. This is also reminding me of how difficult it is to fully understand and use other people’s code – I don’t think I’ve ever loved Javadoc so much!

I’m also getting a little bored with my website. I think I might take some time this weekend or early next week to fix it up a little.

Week Four

I spent some time working on the java program for the landscape software. I managed to break the landscape down into a lot of intervals and calculate how many points are in each one. However, the next step would be to map those height values to the landscape's z-coordinates - which I'm having a lot of trouble with. Lucky for me:

The eye tracker is in. and fixed. We had to spend this week making touch-ups to all the equiptment in preparation for testing next week... the camera lens is a lot wider now, so it can display the whole touchscreen that the person is looking at on it. We ran a few pilot tests this week, and they all seemed to go pretty smoothly. The battery for the device that records what the two cameras are looking at is a little temperamental, so instead of just charging it, we got creative. We hooked up a giant power supply to it instead... and now it works flawlessly! Ah, to be an engineer and have so many fun toys to play with...

I just submitted my application to be sponsored to attend the Rising Stars of Research conference at the end of the summer. I really hope I get it. The idea is that there's like a giant poster competition of people's research projects from the summer (and the past year) and there are sessions on like applying for scholarships for grad school and things like that. It seems like it could be pretty cool - especially since its not just a computer science thing, but an 'all areas of science' thing...

Week Five

This is the testing week! 12 tests in 5 days - whoa baby... So, it's only Thursday, but so far this week (well, since Tuesday) I've managed to convince 9 of my friends to take an hour to come into my lab and be participants in my experiment. It's really interesting to adjust the cameras so that they show the pupil of the eye in the center of the screen. Everybody's eyes are shaped differently, and located at different depths on their face - some are more sunk in etc. For the most part, feedback from the participants has been good. A few of them felt like sometimes the matched colours didn't really match, and other times, they matched the colours very closely but it did not say matched. However, the whole point of using the Lab colourspace is that it is linear. I'm pretty sure that the inconsistensies are just in how we perceive colour... or maybe I'm just very protective of the program - even though I didn't write it.

The people in my lab are plotting hardcore. Today, I have my first group meeting with my lab. One of the graduate students is leading a discussion on some software that runs on a tabletop display which architects will use to plan interiors of buildings on instead of using paper... it seems pretty cool. Anyway, they've arranged it so that while he is presenting, an engineering student in a hotdog costume is going to come in and sing a song about my boss and give her a sausage. never a dull moment!

Week Six

Now that all the tests have been run, there is nothing to do except analyze, analyze, analyze! But that's not happening yet. Colin's business partner is tweaking the algorithm for the eye tracking software, so analyzing is stalled for a few weeks. In the meantime, I am working on the java landscaping program, and making it into a second research project. This will test memory - whether people can remember if they've seen a 3D graph before vs. if they are better at remembering when they've seen 2D graphs, or even just a set of points. It should be really interesting. I am still working on the code which will calculate the height.

Melanie found me an amazing scholarship to apply for - I really hope I get it. I just have so much to do in order to prepare for it. It requires an official transcript, a resume, and answering a few questions about my direction and about how i feel about technology.

Week Seven

My code will finally calculate the height and graph a 3D landscape given a multivariate set of points. I am spending most of this week reworking the code so that it works as smoothly for all data sets that are used as input files instead of just the dataset it was tested on. There are a lot of factors to consider as well. How many points should be used in the graph? What colour scheme should be used to highlight the points? How many contour lines should the graph have? I have started taking screenshots of completed graphs to be used in the experiment. Also, I started exploring DirectRT for windows vista (it's a beta version) to see if it is usable. It was totally running smoothly (minus an audio error which doesn't matter since I'm not using audio) but then it just started getting this weird error. Of course, who knows if the error is in my input data or in the program itself...

Week Eight

The error was with my input file; consider it fixed. It's hard to tell when a beta version goes wrong whether the problem is you or whether it is the software. I guess its probably a better idea to assume that its with you, and do everythign you can to fix it. This was my week of screenshots. I literally spent like all week setting up separate eclipse projects for each dataset and taking screenshots. That is: 13 datasets x 3 datapointValues/dataset x 3 elevations/pointValues x 2 angles/elevation... that is 234 screenshots in all! I sincerely hope they all look distinct enough, because if Melanie asks me to generate another set of landscapes using a dataset, I might cry. Not that this was hard or anything. I found it to be sort of amusing. but slightly boring work. It was definitely doable, but it was quite tedious... it just took forever and ever! Ah, whatever. I'm done done done. Now all I have to do is start compiling the screenshots into a working program using DirectRT, ! but that seems like a job for next week. Let's be honest, Canada day long weekend is so close that I can almost touch it... and I can't wait much longer.

Week Nine

This was a short work week. Monday I took off, and Tuesday was Canada day, so there wasn't much work going on at the start of the week. But, I've definitely made up for it the past few days. By the end of Thursday (and I stayed at work til almost 6:30 to get it done), I managed to complete the first copy of my landscape memory experiment. Melanie has been away sick the last few days, but hopefully she'll be better by Monday, and then I can show it to her and see all the changes that she wants made before we start piloting the experiment... and then piloting away! amazing! So, because I'm done that, I've gone back to my original assignment. I'm currently setting up the calibration files for all the videos from the Eye tracking device. Basically, I am calibrating the video with the eye on it by mapping the eye movement to the area of what the eye is looking at. Then, we will be able to tell where the person is looking the whole time, which is pretty cool. Ok, back to the video! analyzing!

Week Ten

The new experiment is completely complete. I've started running test subjects for it. I ran 6 this week. Feedback seems to be good. They like that it only takes like 25 minutes to do. Also, its interesting to note that almost everyone who does the experiment mentions that they try to form the landscapes into images in their mind to help them remember.

I'm back to organizing the data from the first experiment. I am generating calibration files, and processing videos, and generating fixation files, and annotating them. Basically, I am transforming the raw data videos into usable videos, and csv files that can be used to analyze the data and interpret the stats. It's a lot more time consuming than it seems... in fact, I should probably get back to work right now ;)

Week Eleven

The week eleven report is a little slow on going up, but only because i've been so busy. For each of the 12 hour-long videos for the eye tracking experiment, i had to go through the videos and (using a program) tag each eye fixation... thats a lot of hours of video and a lot of fixations! Good news is, all videos are completely tagged and ready to go. My boss transposed the excel logs of the fixations into the correct format of files so that SPSS will make them into boxplots and anovas, so we will get started analyzing the results in week twelve! I'm curious to see how the results look...

Week Twelve

Ah, the week of testing users. I finished testing 20 users for the landscape study this week. Feedback was pretty good. Here are some examples of the kinds of landscapes that they were shown. These are rough ones, though, that werent actually used. All the ones used were in green, and didnt have a scale, but they had a large point size, and a lot more points!

Most people said that they tried to associate the landscapes with images in their mind so that they would remember them. One of the participants said he would look at the image and then shut his eyes, and then open them and look at the image again.

After I ran the studies, I compiled all the logs into excel files. these files should hopefully be acceptable formats for SPSS so that they can be used to create anovas and boxplots in order to analyze the data. I'm pretty sure they'll work fine, but I guess I'll find out next week!

Week Thirteen

So the data looks good! Colin has already started writing the CHI paper (hopefully) for the slider study. As for the landscape study, the results were basically that there is no effect whether you use 3D or 2D or flat images in terms of memory (which is what we wanted to find out). People did slightly worse with remembering the 3D images, so the plan is to test 10 more people and see how that effects the results we've been getting. I was using a trial sample of a program called DirectRT to build the interface for testing everyone, and now the trial period has expired. I emailed the company and they sent me a new version of it to use instead, but when I install it, it says that the trial period has expired. I guess maybe there are some .dll files or hidden files that I did not fully manage to delete? I'll have a look at that now and see how it goes.

Well, I guess I'm done everything that they had planned for me to do this summer - which is good news! It means that I am efficient and that I didn't mess up too much! One of the masters students is using tabletop displays in his thesis, and i am helping the lab out by trying to figure out how to convert an AutoCad image into an image that vtk will recognize and love. This is a little bit harder than it seems since AutoCad produces files in .dwg and vtk won't accept them, so there has to be an intermediate conversion... But don't let me get ahead of myself! I spent the last few days learning how to generate 3D objects using a Java-based wrapper of vtk and doing fun things like making them rotate and zoom and move - all extremely basic stuff, but still kind of challenging!

Week Fourteen

Another week bites the dust. It actually went by super fast. I am finally taking advantage of the whole 'flexible hours' concept. I came in to the office at 2 pm today, and I'll stay until 9:30 or 10 pm, which I don't mind doing. In fact, I'm pretty sure that I'm exponentially more productive when I am working in the evenings or at night than when I go to work in the early morning.

For example? All week I have been working in high level languages trying to learn vtk a little more in depth. I have settled on using Java, since it is much easier for me to compile and link using eclipse for some odd odd reason. I have been trying to create an applet that will open a 3D image file, and then apply transformations to it - the basic ones from vtk (pan, rotate, zoom). I have been strugging a lot, but today when I was looking something up, I stumbled upon a beta-ish version of a free program coded in Java that will open a .vrml, .wrl, .3ds, .obj (and many other image files) and treat it as though it were a vtk file. It was amazing! So, I downloaded the source code and I started using it and it is fabulous! The only problem is, it can only accept Autocad files which are .dxf (so i need to find a way to convert .dwg to .dxf) and it will only accept .dxf files from Autocad 2000 (and i need to figure out why), and it will only recognize certain entities of .dxf (lik! e line, and vertex, and other simple ones). Well, I have my work cut out for me, but at least its a start! Plus, I'm sure there are forums with threads in related topics that will help me out! It is extremely interesting to be working with 3D objects. I have realised that this is something I would be interested in doing once I graduate (and before I attempt a masters).

Things on the DirectRT front aren't going so well. I am still waiting to hear back from the company about another license key so that I can keep using the program. In the meantime, testing is completely stalled! Ah, well... I still have two weeks left to go. I installed DirectRT on another computer and got Colin to email the company from his ubc account, but they have not responded to him yet either.

Week Fifteen

I have DirectRT up and running again! It is currently being run on the Windows XP common computer, so I need to move that computer tower into my office so that the environment for the study will be consistent. I am still using the same Wacom Tablet monitor so that it will provide users with the same screen resolution and such. Also, the monitor that is currently being used on the common computer doesn't have a cable that fits into the back of my Windows Vista computer, so i guess i need to abandon it for a while - that or keep shlepping the monitor back and forthe between the two computer towers. Hmmm that could be a good workout! But back to the point, now that DirectRT is working again, Melanie and Colin would like me to find 10 more subjects to test so that their results can be stronger. Ah, to find ten more subjects... a quest indeed!

I got an email from the CDMP people today telling me that I need to finish the website and write a final report before I get the final installment of my award ($). So, I'm going to put the java programming on hold for a little bit and start writing my report. It shouldn't be too hard, except that I worked on two separate research projects (the second one still kind of in progress) so my report is basically going to be two reports in one!

I got an email from the Computing and Software Department at McMaster University outlining the undergraduate tutoring/marking positions that need to be filled. I applied to TA most of the first year computer science courses (and I will see in a few weeks if I get offered anything) even though I don't feel qualified. This summer has made me realize that I really am capable of doing anything - even things I am very unfamiliar with - I just have to try! So, if they give me the chance, I will do my best, and I'm sure it will be good enough!

Week Sixteen

My final week at work and the bosses are on vacation! This has been a crazy week... I've been running subjects like crazy, and trying to put together TWO reports - one for each of the research papers i did this summer. Time flies super fast, but this was an amazing experience. No regrets. I had a meeting with the guy I was doing the java programming for, and he was happy with the little progress that I made. I also need to clean out my computer. So much to do, so little time! And now it's all about submitting the website and the report, and packing to go home! What a summer!

Glossary

AutoCad - a CAD software application for 2D and 3D design and drafting.

CHI conference - The annual CHI conference is the leading international forum for the exchange of ideas and information about human-computer interaction (HCI). http://www.chi2009.org/

Eye tracker - glasses with two camera lenses on them: one camera records where you are looking, the other records the movement of your eye. The footage is calibrated to produce an accurate representation of where the eye is looking.

VTK - The Visualization Toolkit (VTK) is an open source graphics toolkit. It is a platform independent graphics engine with parallel rendering support. (courtesy of Wikipedia)