Austin Unveiled: A Tour of the Music Capital Crafted by AI

One of the biggest (and most controversial) topics of 2023 was the rise of AI, specifically AI tools that are now accessible enough for the common person to use, and that are powerful enough to complete tasks in seconds that would otherwise take a “human” hours, days, or even weeks.

And while the conversation surrounding AI tools (specifcally in art, design, journalism, and more) is constantly evolving - especially in regards to ethics, use cases, and the overall effect it will have on society - we wanted to see just how “creative” it was in terms of putting together a simple “Visit Austin” type of video.

Everything you see in the video above was created by an AI tool. From the images and animation, to the music and voicover.

We started off by directing ChatGPT to create a simple script for a 30 second video about visiting Austin, Texas. It did this in about 5 seconds, and gave us a complete rundown of what shots to create, and the accompanying voiceover verbiage.

We then went to the AI image tool “Midjourney” and used simple text prompts to get the shots that ChatGPT called for. For example “Austin skyline at sunrise” or “people dancing on Sixth Street.”

While Midjourney did its best to create photorealistic images, one downfall is that it certainly isn’t perfect yet. The Austin skyline isn’t quite right, the “I Love You So Much” mural is incorrect, among several other inconsistencies.

Regardless, they were still pretty good. So once we had all our shots, we took them over to the next AI tool, RunwayML. This impressive tool can generate 4 second video clips from nothing but a text prompt, however it generally does better if you feed it an image to work from first. Using each of our shots, it then animated them one by one. Again, this isn’t close to perfect yet… as you can see from the video, there are strange visual artifacts floating around, odd glitch type errors, and several other identifiers that the video is AI generated.

Once we had our animated clips, we simply input them into a simple video editor to put them in order. After that, we grabbed the voiceover script from ChatGPT and fed it into the AI voice generator over at ElevenLabs. This gave us the calm, smooth voiceover we were looking for… but again… it still wasn’t the exact tone and phrasing we wanted.

Finally, we had the AI music generation tool SoundRaw pump out a simple tune to match the vibe we were looking for… and once we added everything together… boom, an Austin tourism video generated completely by AI assets.

Is this the future of video production? Will travelers have to be more and more aware of AI images and videos when planning a trip? These are just a few of the thousands upon thousands of questions that all of us will have to keep in mind in the near future, and going forward.

Maybe we’ll create a “Visit Mars” tourism video next.