This Friday, we hosted our soft opening as preparation for the ETC Festival. Since we’re also pushing toward our documentation deliverables for the final two weeks, we decided to finalize everything this week regarding what will be showcased at Festival.
This week brought major progress across tech, art, and design.
Tech
We completed the entire hardware setup, including the TV, adaptive controller, speakers, microphone, and new sensor system. The sensor now detects when someone enters the play range, creating a more intuitive onboarding moment for starting the experience.
Art
Most clipping issues have been fixed, and new visual props were added to Athena to strengthen her character impression. Both Athena and Poseidon now have additional animation sets, including idle, thinking, and body movement animations, significantly improving their screen presence.
Design
Outside of finalizing the overall experience flow, we are also considering adding a buffer interaction so users receive feedback even while waiting. It’s important because latency sometimes interrupts the pacing.
Overall, the sensor performed perfectly with the full hardware setup, and the experience flow is now fully complete for Festival.
Latency Issue
The main challenge we encountered was latency. Some turns took around 15 seconds, which impacted the user experience. Our initial diagnosis suggests this is caused by network speed or temporary outages from ChatGPT or ElevenLabs, not a system flaw on our side. We’ll run further tests on Monday at ETC.
Documentation Work Begins Monday
Next Monday, the whole team will begin focusing entirely on documentation.
Additional Notes
During soft opening, we received a lot of thoughtful feedback about making the experience more interactive and immersive. Several distracting elements were identified, and we aim to refine those before Festival.
We also discussed our documentation plan with several faculty members. A recurring suggestion was to document not only what we built, but also the challenges we solved and the challenges the next team should expect, aligned with what we talked about in last week’s meeting.
In a conversation with Mike, we learned that the project may eventually be hosted for a museum client, potentially involving the use of local LLM models. We will include this context in our documentation for future teams.