A nostalgic trip with iconic startup figures reimagined in retro-style home videos: After watching the video, you might be wondering how we brought these moments to life—and which tools made it possible.
In this article, we’ll take you behind the scenes, step by step, revealing the tools that helped us capture that authentic, vintage feel. Spoiler alert: There's a lot more to it than just Midjourney and Runway.
From idea to video
When we met to define our strategy for the video, we knew that we were missing a full production crew. However, AI tools could help us create an impactful promo video.
No matter how well-executed an idea is—if it doesn’t communicate our message or isn’t compelling, it could be made by Pixar, and you still wouldn’t get the desired result.
For the video, we aimed to capture the essence of startups: the entrepreneurial drive, the grit, and the power to grow. Startups are about visionaries with ideas that could completely reshape our world.
Many ideas hit the table, but one stood out, both feasible and inspiring: What would the founders of the biggest startups have said as kids? What did they dream of back then that shaped lives today? How did their vision grow alongside them? Every dream starts small, and it needs space to thrive. Our message is clear: The Branx can grow with you.
The storyboard
After deciding which startup founders we wanted to show and after putting imagination into what they could have dreamed about as kids, we moved on to putting the theory into practice: Every video project needs a storyboard to bring it to life.
Contrary to popular belief, a storyboard doesn’t have to be hand-drawn. We used a mix of images, videos, and stock photos to set the scenes. Each piece was timed, and we carefully matched audio to visuals.
The storyboard is simple: an intro, six mini-stories (each with several scenes) featuring the founders as kids, and an outro. Needless to say, the big challenge lay in creating each mini-story: Where do we get the images from? How do we make these childhood sceneries feel as authentic as possible?
Creating iconic startup founders with AI
The founders' childhood
Our first challenge was figuring out how to portray these famous startup founders as young children (ages 1–6) without having any original photos of them.
We used Midjourney to generate images that represent each founder.
For the first founder—a music-loving individual from a Nordic country who coincidentally dressed in green—we aimed to create an image that could belong in their family photo album. Think 70s aesthetic, rich with nostalgia. To capture this, we crafted detailed prompts in Midjourney and included reference images to capture both the founder’s physical features and the era’s photo style. We also pulled in frames from vintage home videos to give the scenes a consistent, real look, enhancing the story's authenticity.
For the viewers' understanding, it’s essential that this final image subtly hints at the founder’s identity. Alongside a voiceover of a child talking about their love for music, we dressed the character in green and added headphones—a nod to the color associated with their startup.
We continued generating childhood images of other founders with Midjourney: one wears a black turtleneck and bites into an apple, while a girl mentions her love for meeting new friends, wearing a yellow shirt with a bee on it.
The voices
At this stage, we faced a new challenge: How would we create the voices for these children to say exactly what we needed? This turned out to be one of the toughest tasks, as AI-generated child voices are limited in variety. By combining some options and carefully editing them, we achieved a fairly realistic result—though these AI voice tools still have room to improve.
Here’s a list of the tools we used to generate the voices:
- SpeechGen: https://speechgen.io/en/child-tts/
- Artlist: https://artlist.io/voice-over
- Fineshare: https://finevoice.fineshare.com/
Bringing a Midjourney image to life with Runway
Once the images were created in Midjourney, the next step was to bring the characters to life, keeping the viewer immersed in that cozy, home-video vibe.
With Runway Gen-3 Alpha Act-One, we were blown away. By simply uploading the Midjourney-generated image and a reference video showing the gestures we wanted the character to make, it produced the result in seconds.
The driving performance video we used was a stock video from Freepik. While the video didn’t have the exact look or character we wanted, this AI can capture detailed facial movements. It can move eyes, and lips, and even generate teeth that weren’t visible in the original static image. It’s incredible!
Representing each founder's childhood with stock home videos
As you can see in the spot, we wanted to immerse viewers in the founders' childhoods, considering details like age, location, and family background. We used stock home videos to capture that era when these founders dreamed big and spoke their ambitions out loud. To create an authentic 8mm feel, we added video filters to mimic Super 8 film, combined with sounds of projectors and clicking keys—elements that give a nostalgic, realistic atmosphere.
We sourced videos from various stock libraries, both free and paid. A standout free source was Archive.org, which impressed us with its extensive collection of vintage 70s home videos and retro ads that matched the look we were after.
Editing in After Effects
The visuals
After organizing the AI-generated material, we moved on to assembly and editing in Adobe After Effects. The overall aesthetic of the video draws heavily from vintage home videos, reminiscent of analog cameras like the Super 8 or VHS recorders. By using various effects, filters, and layer blending modes, we created a unified look. This consistency was essential due to the variety of materials used (vintage footage, color grading, exposure, noise, filters, etc.).
.png)
We applied effects like Fast Box Blur and Noise to soften high-definition clips, giving them a vintage feel. For VHS-style footage, we used Wave Warp to create horizontal offset and Venetian Blinds to overlay horizontal lines, classic markers of that era.
We incorporated two types of scene transitions. The first uses camera lens flares, enhancing the cohesive, vintage aesthetic. The second transition, used between founders, is a classic cut, combined with broader contextual clips introducing the next featured entrepreneur.
Additional effects included color, saturation, and exposure adjustments, along with a touch of chromatic aberration. Lastly, we used the wiggle expression on some elements to simulate a handheld camera effect.
The sound
We applied grainy and noise filters at various frequencies to create a textured background for the overall composition. We had to pay special attention to the AI-generated voices. Initially, the raw audio sounded somewhat robotic and had a layer of noise. To clean this up, we used Adobe Podcast, which left the audio much clearer.
Next, we moved to Adobe Premiere and applied effects like FFT Filter and Distortion to these audio tracks. By adjusting various parameters and adding a touch of background grain noise, we achieved an "encapsulated" sound, similar to audio recorded on an old-school voice recorder.
The outcome: Childhood dreams, now real
With the help of AI, we crafted a realistic video that captures the spirit of the founders' childhoods, blending real-life scenes with images and voices so lifelike that we were genuinely blown away. The video connects with our target audience—founders, CEOs, and startup entrepreneurs—in a playful, engaging way, sparking the ambition within each of us. The message is clear:
Every big dream starts small. Let your vision grow.
Side note: The video creation process opened our eyes to the vast possibilities AI offers designers and content creators. As we explored these tools, we were both astonished and a bit shaken. Creating realistic scenes for a promo video is just one of many potential applications of generative AI, and human creativity is essential to fully unlock these possibilities. However, this experience also made us even more aware of potential misuse. We learned how easy it is to manipulate images, voices, and videos to create realistic simulations. Hopefully, this video serves as a positive example of how to use AI tools responsibly to entertain audiences.