Weekly Blogposts

Week 7: Survey to Evaluate Kids' Perception of Agent


This week was generally less hectic than previous weeks. On Monday, I worked more on drafting questions that Dave and I could use to evaluate how kids perceive our agents - whether they see our agents as having the right gender, age, personality, etc. I also continued to work on the documentation for the models that I'd started last week. As mentioned before, I want to make sure that the documentation is detailed enough to provide sufficient context for the models - the tricky part is knowing which details are superfluous. At the last lab meeting, Justine had also stressed the importance of backing up files on the server; so in addition, I made sure to upload all art files onto the server. Of course, a great deal of them (at least 100 or so files) are ZBrush files and cannot be opened in any software program besides ZBrush. However, they are mostly old models that I saved in the process of working up to where I am now. For the final models, I will make sure that they are in OBJ format so that they can be opened in Maya or other modeling programs.


On Tuesday, I continued working on the documentation. I also presented my list of survey questions to Justine. Justine, Dave, and I sat down to discuss how to best structure the survey and how to best word the questions. Besides that, Justine also mentioned that it would be best to have a racially ambiguous model. Our current model is rather obviously Caucausian; thus, Dave suggested I mock up a few models that I felt looked racially ambiguous. We could then present these models during Thursday's lab meeting for general feedback. Dave also mentioned the possibility of setting up an online survey, which I will have to look more into.

As a side note, Justine's gorgeous dog Inca recently underwent surgery. I'm a dog person and I have a soft spot for border collies (since our first dog was a border collie mix), so I felt very bad for Inca, who was lying immobile on Justine's floor during our conversation. I hope she recovers speedily!


I mostly worked on producing racially ambiguous versions of the model. Ultimately, the models I produced differed from the original only in minor details, like skin color (obviously) and slight adjustments to features (thinning of the eyebrows, for instance). However, in the development process, I did try a variety of different things to get my models looking more racially ambiguous. I attempted, for instance, to change the boy character's hair to a more "moppy" variety; however, this only succeeded in making him look nerdy. I've found that it's quite challenging to make racially ambiguous models using our current style. What are racially ambiguous features, anyway? Most of the racially ambiguous people I know are celebrities and most of those celebrities look very Caucausian (Jennifer Lopez, Mariah Carey, Jessica Alba, Rashida Jones...) But that's probably the result of Hollywood trends, in which people who look more Caucausian are chosen over people who look otherwise. In any case, I'll need to look at more reference photos of racially ambiguous children.

Besides working on the model, I also formatted the survey questions and emailed them to people in our lab for feedback. Almost everyone is gone this week (Samantha, Evelyn, Amy), but I hope we're able to gather feedback nonetheless.

Dave and I also had a long discussion on Wednesday regarding my future plans. We talked about the pros and cons of graduate school, the kind of people who should (and should not) go to graduate school, and many other topics besides. I really enjoyed hearing Dave talk about his experiences in and out of graduate school; I definitely think our discussion will help me figure out what exactly I want to do in future years. Currently, I am about 70% certain about graduate school - but far less certain (maybe 50% certain) about a PhD. A great deal of self reflection and contemplation will be necessary before I make any final decisions.


On Wednesday, I mostly worked to refine my models - with the hopes that they would be the final iteration before I started on rigging and animating. These final touches included fixing the girl model's shoes (they were a bit lopsided) and the hands of both models (still somewhat mangled looking). At last, I reached a point at which any changes I made to the models actually made them look worse than better. It was at this point that I decided to wait for the lab meeting (moved from Thursday, the 4th of July) to see how I should proceed.

In the meanwhile, I began writing a documentation of the modeling process. Although ZBrush has been gaining popularity in recent years, especially in the gaming industry, I'm not sure if CMU-hired artists would be familiar with it. It seems to me that most artists hired by the ArticuLab are probably people with both technical and artistic skills, who are maybe CMU-affiliated or trying to apply to CMU's Entertainment Art industry. That is, they probably have to stretch their skills over a wider range of topics (both the technical and the artistic) and thus can only dedicate themselves to the most universally used modeling programs like Maya. In that case, anybody who adopts the Peer Tutoring project after me will need to (or at least they should) know how I developed the models, in case they need to go back into ZBrush to make changes.

In my documentation, then, I've not only described the general process by which I created the models, but have also given tips on how to change aspects of the model (the texture, for instance, or materials). I'm not done with the documentation, but I will try my best to give a thorough description. Having worked with Alex, I know how frustrating it can be working with someone else's files, not knowing exactly what they did before you came. Of course, in the best case scenario, I would finish all the modeling work and start on some of the basic animations - anyone coming in after me should only need to worry about animating a ready model in Maya. But it's always good to prepare for the worse-case scenario in which I am not able to finish the models by the end of the 10 weeks and some one else may need to pick up the modeling work after me. I think ZBrush is pretty intuitive compared to Maya, so using it for touchup work would not be exceedingly difficult.

Around 2 PM, we held our usual weekly lab meeting. During this meeting, Justine told me that the model looked "compelling" and was finally in the right style; now, we had to test it. As such, my next task is to write up a series of questions that we can ask middle school/early high-school students. These questions will be used to gauge how effective our model will be as a peer tutor. Dave and I will present these questions to Justine next Tuesday at 2:45 PM. If they are approved, we can schedule them into Dave's user studies.


Thursday was July 4th, which was supposed to be a day off. However, Id be going to Philadelphia on Friday (a regular work day), so I made up for my Friday hours by working on Thursday. I worked mostly on refining the survey questions regarding the models. In addition, I organized all the art files so that they could be transferred onto the ArticuLab server.


I went to visit Tiffany in Philly on Friday - no blogpost!