I created the video above using a variety of AI tools, particularly ChatGPT, Gemini (Nano Banana model), Kling AI (video 2.1), and Suno, an AI music application. To bring it all together, I used Adobe After Effects. I describe the process below in more detail. I hope you enjoy and get to try this yourself.
Creative AI tools are continually improving. The things people create with them range from highly conceptual, fantasy-driven, and otherworldly artistic short-form images and movies to more practical solutions. Recently, I have been experimenting with AI to see how practical I can make my results. In a previous article, I utilized JSON to maintain consistency for a 3D icon illustration set I created using ChatGPT. However, I have noticed many possibilities and creative outlets for product placement, as well as tools that can help build a small AI commercial. I decided to create a concept ad for Nike Air Zoom sneakers. Of course, this is unaffiliated and is a passion project. As someone who runs, I like their sneakers, and as a designer, I admire their brand. That is why I chose Nike.
To get started on this concept, I envisioned how I wanted to present the product. I sketched out a simple storyboard and began creating the assets. Before diving into the AI part, I grabbed an image of the sneakers I planned to use. I then used that image along with a prompt in Gemini, using the Nano Banana model, to create the frames for the animation. Each static image or frame is a key component in the later animation process. I used unique prompts as well as the sneaker image for each AI-generated frame. Lastly, I did the same for the human model. Using a highly detailed prompt and the sneaker image, I successfully achieved the desired results. Since ChatGPT did a better job at creating a model, I decided to use that image as the frame for the model.
When writing prompts, it's important to put yourself in the shoes of an art director telling a vendor what you want them to create. Include cues such as camera angle, fps (frames per second), shutter speed, aperture, lighting conditions (natural vs studio), and the type of angle on the product—for example, I instructed Gemini (using Nano Banana) to use a 3/4 angle view of the product. Add other keywords such as 'levitating' and include the hex colors you want to use (if adhearing to brand guidelines), or just describe the colors and lighting you prefer. Compositionally, it's helpful to sketch out your ideas so you can better describe them to the AI tool. I found this useful when I needed to detail certain parts of the prompt. For instance, for the first frame, I mentioned in the prompt, "to make the background of the sneaker exploding crystals in purple and orange." Also, it's important to assign a role to the AI model you're using. For this, I used, "You are an expert 3D animator and designer, cinematographer, and product photographer with expert skills in visual effects." Then, set a goal, which I defined as a "stunning advertisement." This helps the AI tool you are using produce a more desirable outcome faster. Since it cannot read our minds, it's crucial to be as detailed as possible.
This is what the final prompt looked like for the first frame.
"You are an expert 3D animator and designer, cinematographer, and product photographer with expert skills in visual effects. Turn this simple photo into a stunning advertisement. The shoe is 3D in a 3/4 dramatic shot, 24 mm lens of the shoe facing the camera, floating, levitating in mid air with dramatic streams of wind and air around it flowing from bottom to top and circling the shoe in an effervesce. The background is black, with dramatic studio lighting highlighting the shoe's features. Behind the shoe, there is a purple and orange explosion of glass crystals exploding from behind the shoe, and flying toward the camera, all captured in a 1/3000th of a second exposure. The final is highly detailed in 4K, HDR, and prime-time commercial quality. "
I then modified this promt for the other two images based on my vision for other sequences, which indlucded water and earth type movements and bursts for dramatic effect.
For the final image of the model I used ChatGPT and a similar principles of including a product image as well as being highly descriptive as an art director to create that image frame. As a reuslt this is what I got as my frames to animate.
Frame One
Frame Two
Frame Three
Frame Four (jogger)
The animation process...
I envisioned this product rotating and morphing from one composition to another. To do this successfully, and make each frame morph into the next. I used Kling AI to set 'start" and "end" frames, which resulted in two videos: the frist being where frame "one" morphed into frame "two", and another where frame "two" morphed into frame "three". I used the first image as the initial frame and the second as the final. I described the type of animation I wanted Kling AI (AI, video 2.1) to make. It's essential to describe in the desired results in detail; I suggest to write the prompt and check if you're missing any key details for the animation.
Again, I described in detail what I wanted and adjusted the prompt as needed. I used this prompt for the actual 3D animation and got the results below:
"You are an expert 3D product animator and videographer. Create a dramatic, highly creative, and impactful animation of this product, with a dramatic zoom and smooth morphing with no cuts, where the shoe spins to change from one scene to the next."
I modified the prompt for other instances to accommodate the changing effects.
For the last sequence, I wanted to place the product on a person, so I used the frame I generated in ChatGPT of the model (Frame Four) wearing the sneakers and set it to only the "start frame" in Kling AI. I described the scene in detail.
This is the prompt I used for the last animation.
"You are an expert videographer filming a prime-time sports running commercial for a popular sneaker company. The goal is to create a highly creative, dynamic short clip of a runner tying her shoe. At the same time, the camera quickly zooms into the sneaker for a close-up of the sneaker, and then zooms out quickly to reveal more of the park environment as the runner gets up and starts running with her back to the camera. Use smooth animation, with no cuts, incorporating smooth zooming with ease-in and ease-out animation."
Lastly, I wanted to add a soundtrack to the final ad. So I used SUNO, a music creation AI tool, and asked it to create "an energetic workout track for running in the park."
As a result, it created a whole song (see below), which I only used the intro of because it fit perfectly with the length of the video.
Once I was satisfied with all the results produced by the AI tools, I assembled the entire video in After Effects. Setting the final animation, pacing, and length, as well as adjusting the musical cues, I added some animated text. I used Adobe Fonts to find a suitable font that works with the Nike brand and applied suitable animations to the text. I like isolating each element in After Effects using compositions, which makes the text animation more manageable and easier to work with. After assembling everything, I run the video repeatedly and make minor tweaks in the timeline to ensure the animation is smooth. I worked through three variations (which are not included for brevity).
In conclusion, these rapidly evolving AI tools provide an effective way to produce short-form content for various channels. They also serve as valuable tools for prototyping complex ideas and projects in a hurry, helping to plan and set expectations for larger-budget commercials and TV spots. It's truly exciting to learn these new tools and witness their power in generating excellent results in short amounts of time. As technology advances—and I'm sure new methods and tools will emerge in the coming weeks—I'm eager to see what the future holds for AI-driven design.