Inspiration
We were inspired to build something that showcases NYC, and what better way to learn than to revisit the past. Our immersive project highlights landmarks that has shaped what NYC has become today, paired with interactive details about subjects in the image.
What it does
Vantage is a dynamic website that enables users to choose a point to see significant historical events and a 360 image that you can look around in using VR or clicking and dragging via mouse input.
How we built it
We scraped data from a map then gathered information through Gemini to get the scene and key details, then used Nano Banana to create the panorama, then generate hotspot vision for labeling with Gemini vision + bbox detection, then generate summaries by scraping Wikipedia and verifying with Gemini and finally we render the info-panels for VR with text-to-speech
Challenges we ran into
Refining the image generation prompts, integrating the VR aspect with info-panels, setting the hotspot vision to generate details about image subjects
Accomplishments that we're proud of
Integrating many AI tools to create a pipeline, making the project dynamic so users can generate any NYC location, adding text-to-speech for accessibility, brand design
What we learned
Creating AI pipelines to generate creative experiences, using gemini hotspot/bbox to label subjects in the image
What's next for Vantage
Expanding the geographic scope and generating even more realistic depictions with more curated data.
Log in or sign up for Devpost to join the conversation.