Tags: Testing
Foreward: I have decided to split the November post because it felt like half of the month was extremely productive, while the other half was less productive. Our goal was to go to Oslo, Norway, in order to test the prototype, visit the city and have proper discussion with the client. Everything went messy the moment Erasmus got involved, but I will talk about this more in depth. Also, this month we had our Midterm presentation, so I will talk about it, reflect on the peer review and do a general reflection.
Starting with the Pre-Trip "Journey". Since we had a goal to go to Oslo in order to do testing, we firstly asked our guiding teachers if it's possible to go there. That was when we got told about the Erasmus grant. We spent a few days, up to a week, in order to get information regarding Erasmus grant, create the application and send it properly. Unfortunately, everything got delayed due to the Autumn break and the fact that the teacher in charge of Erasmus went into a vacation. We were still extremely optimistic, but our Erasmus got declined because no foreign university was involved in the project, which wasn't even a requirement to begin with. After receiving the decline, we gave up on this idea and continued working on the project and the Midterm presentation and deliverables.
We told our guiding teachers about this turn of events but apparently our goal was not over yet. There was another possibility of going to Oslo. Saxion XR Lab could sponsor the tickets if the theatre was willing to pay for our accommodation. We had an "emergency" meeting with the people from the theatre and in the end they were willing to pay for our Airbnb. Because of this, we had to spend time again checking accommodation, legal stuff and so on. Fast forward the bureaucracy; we managed to go to Oslo for 4 days, 3 night. Out of these 4 days, 2 were travel days and 2 were working days.
The purpose of the Oslo Trip was to Test the prototype. The first day was a bit less successful since we encountered a few issues that weren't present in Enschede. However, on the second day, we actually managed to test both the ChatGPT integration and also the Metahuman with a professional actor.
The testing went this way.
We had a blockout map of the courtroom. In that blockout map we had a metahuman character. The first testing was the Metahuman testing. We had a professional actor whose face was tracked by an iPhone camera through an application called LiveLink. This tracking data was sent to Unreal Engine in order to give life to the MetaHuman character. The actor had a script in front of her and we wanted to test the functionality of the MetaHuman technology with an improv actor. Testing on our own is different, since we do not gesticulate the same as a professional actor. During this part of the testing phase, we encountered a few issues:
1. Since the actor was ready a script, the iPhone tracking also tracked the eyes moving from left to right. This motion looked extremely off and uncanny on the virtual character.
2. Since we iPhone is in front of the actor, some parts of the script weren't readable. The iPhone was blocking the way.
3. The actress was very expressing, which in real life is perfect, it is good and normal to show emotions. However, the tracking of those expressive gesture weren't captured properly by the iPhone. Raising the eyebrows posed an issue, the fast moving lips were another one. If the actors speaks very fast, the tracking does not capture the lip movement that well and sometimes even skips an animation frame.
The second part that we have tested was the ChatGPT integration which worked almost perfectly. I say almost because we had some technical issues, which were solved later on. We have tested ChatGPT inputs in both written form and with microphone input. We have used a Shotgun microphone to check the range possible of the microphone and if its feasible to have shotgun mics instead of multiple microphones for each person in the audience. The test was successful, the microphone picked up the sound from maximum range, the maximum range being the entire studio/play area. We tested in ChatGPT can recognize multiple languages too, since in Oslo people speak English but also Norwegian. We forced ChatGPT to only respond back in Norwegian or in both English and Norwegian; and it worked perfectly; sometimes so perfect it got creepy and dark, debating human life value with an AI which has 0 sympathy towards humans might not be a good idea..
At the end of the trip, we had a proper discussion with the people at the theatre about the future of the project, new deliverables and so on. We came to the conclusion that the bracelet is not needed anymore, since it doesn't provide any value to them. So, in the end we had to completely remove that.
Overall, the Oslo trip was worth it. It was extremely tiring, since we wanted to also visit the city. In the end we did manage to test the prototype, have a proper talk with the theater lead have fun too.
Midterm Review
These are the peer reviews I received from my teammates. Overall, I think I did decent enough, the overall grade is between 72% and 92% at Skills and between 76% and 88% at Responsibilities. I do agree that my 2 weaknesses are difficulties of motivating other people and being present on time, and I did try my best to work on them later on. I would like to at least maintain these scored and maybe increase them, if possible. For now, they do show my commitment in this project and I am happy that my teammates, in general, had a good time with me.
Comments