We recently delivered our first AI-generated commercial.  Below are a few takeaways from that experience.  But first, here’s the spot we made for Primo’s Donuts:

Tools used: Google VEO 3, Adobe Firefly, Adobe Express, and Adobe Premiere Pro.

Generative AI Video Takeaways

Generative AI Video Budgets vs Traditional Video Production Budgets

Obviously, money is the biggest player when it comes to generative AI videos.  It only cost a few hundred dollars and just a few hours to create this spot for Primo’s Donuts.  It took some time to come up with the idea (human-generated), but once the idea was in place, it didn’t take long to write the prompts (AI-assisted), generate the clips, and put it all together.

If we had shot this with real actors on a sound stage with a green screen, it would have taken several days for pre-production, a full day to shoot and several days to edit and comp.  The budget for that would have been at least $15,000.

And if we went all out and shot this at the actual locations, it would have been in the six figures.

There are some budgetary limitations with Generative AI.  For this video, we generated about ten different clips and worked with just four.  These four clips were not perfect but we had to make it work since there was not enough budget to generate additional clips.

The bottom line is that generative-AI allows small businesses such as a donut shop to produce high-quality productions.

Generative-AI Videos are not 100% automatic

As mentioned above, to get four clips that we could work with, we had to generate ten.  And those four weren’t exactly what we intended.  For example, the resort singer in the last clip was supposed pause his singing, turn to the camera and say his line.  Instead, AI decided to have the man sing the line.  This was actually a pleasant surprise, but most of the clips had strange, and at times, eerie, results.  We had people chewing before taking a bite of the donut, staring blankly into the camera, and body parts inexplicably disappearing.  There were also sound effects that we had to fix.

Since we didn’t have budget to generate more than ten clips, we took the clips and manually adjusted to make them work.  That included creating new sound effects and manually adding them in.

The text graphics also were not generated properly so we had to create those on Premiere.  Simple enough, but again, not automatic.

Generative-AI requires human creativity

There were two layers of human-creativity required for this project.

AI’s creative abilities are taking leaps not steps, but as of now, human creativity is still a major requirement.

We of course tasked ChatGPT, Copilot and Gemini to come up with a concept, but none of their ideas felt right.  We wanted to convey the idea that donuts, in a way, represented the American Dream and was a part of the immigrant story.  None of the ideas felt right.

So our creative team came up with the “We came for the donuts” storyline.

That was the first layer of human creativity that this project required.  The second was what we discussed in the previous section.  Since AI didn’t generate exactly what we wanted, we had to take the clips we had and get a bit creative about how to use them.  This required some interesting editing and creating some additional effects.

Overall, we feel AI will be able to do a lot of creative stuff, but for this spot, it still needed human assistance.

The overall takeaway is that AI has democratized high-quality videos.  Mom and pop stores, cash-strapped startups and other small to mid-sized businesses can produce amazing-looking videos, and do so on a regular basis.

We’re very excited about this first spot and what we will accomplish in the near future.

To learn more about Picturelab’s Generative AI services, including pricing, check out our AI page or contact us here.