I recently came across a fan trailer for Grand Theft Auto VII, Egypt on YouTube, created by German music video dierctor Yonatan Dor. His comapny, Dor Brothers, developed deep expertise in generative AI by using it for the production of more than 200 music videos. One of the things I’ve been tracking for Forbes is the rapid development of cinematic AI, so of course I was already familiar with The Dor Brothers’ incredible AI-enhanced music video for the supergroup SiM, Hide and Seek. With the release of their GTA trailer, I reached out on Instagram to learn his story, just as I had to AI filmmakers Caleb and Shelby Ward, Dave Clark, Nicholas Neubert and Nice Aunties. I wanted to learn more about how he used AI to make this.
Dor is a frequent speaker at AI events and is collaborating with teams at Microsoft and Pika Labs. “The Hide and Seek project for SiM kicked off when the UNFD record label approached us to craft a music video that captures SiM’s distinctive aesthetic. With only five days(!) for production, we quickly conceptualized a style blending traditional Japanese art with a modern explosive twist to match SiM’s artistic identity. We were provided with a very minimal brief by the label – “Devil/Demon, Bat, Resuscitation (Revival), Play Dead, Immortality” – this is where our imagination took over. We visualized scenes such as guitar-playing demons, massive speakers overwhelming a city, and pre-apocalyptic alarms signaling “The end is near, SiM is here,” reflecting the band’s intense live performances.”
In softer sections of the song, we leaned towards traditional Japanese aesthetics, reimagining “The Great Wave off Kanagawa” to match the video’s modern feel. The video concludes with a poignant image: a man in a destroyed city touching an eye that opens, symbolizing a return to pure, unadulterated art. To create this video, we generated a total of 2230~ images. This figure does not account for the additional unused clips produced using Pika Labs (+300-500).
For the GTA VII – Egypt project, Dor leveraged Midjourney’s “Style Reference” feature to reimagine iconic video games with a unique spin. “The idea of setting a GTA game in Egypt sparked our imagination, combining Rockstar’s trademark humor with a twist,” he said. To bring this vision to life, Dor used:
1. Midjourney for creating initial images using the style reference tool.
2. Pika Labs for animating these images.
3. Photoshop for image retouching and cleanup.
4. ElevenLabs for dramatic AI-powered narration.
5. Adobe Premiere for video editing.
6. Magnific AI for image upscaling and sharpening.
There is another Dor brother, who used to make music videos with Dor, but he’s doing other things now. “I hope he returns someday,” he said. “But I feel like the animators I work with are brothers, too.” Dor mentioned he got started with AI after discovering a YouTube video in 2022 that had only a couple hundred views, when a very early version of Generative AI was introduced. That was before Midjourney or even Stable Diffusion existed. It was called VQGan for those who might be interested. “I was Initially skeptical that it was real, I was astounded by its potential after trying it out, despite producing a horrendous first image. That experience convinced me of the tool’s potential, leading me to dedicate 8-12 hours daily to mastering it.”
“I initially found clients by offering AI Music Videos on Fiverr. It was actually the first time AI Visuals were ever offered on the platform.” Dor explained. “From there, I also created a few YouTube videos that garnered some attention and shared my images in public groups to build an early following. Now, we attract clients from various sources for a wide range of projects.”
Dor noted the dramatic changes we are witnessing in Generative AI tools, predicting that we will see a significantly different level of AI-generated content in the coming year. Yesterday, OpenAI revealed a new generative AI tool called Sora. Sora is revolutionizing the scene of AI video generation at a level which is absolutely unprecedented. So much has changed in the last few months, too much to summarize briefly. This technology is evolving at an incredible pace, with advancements nearly doubling every month. If I were to produce our older videos today, I’d have a much broader array of tools at my disposal. Today, there are numerous tools that allow for a fascinating blend of different techniques.” AI is threatening to disrupt media, social media, and our personal communications. What happens when we know what’s real anymore? We’ll soon find out.
Read the full article here