The promise is tantalizing: film your talent against any backdrop imaginable without leaving the studio. No location scouts, no travel budgets, no weather delays. Just pure creative freedom powered by artificial intelligence.
But as with most AI applications in production, the reality is more nuanced than the pitch. Our recent collaboration with Turkish production company PTOT Films on a banking commercial revealed both the potential and the complexity of AI-generated backgrounds in professional video production.
When Virtual Production Isn’t an Option
The project brief was straightforward: create a commercial featuring a celebrity traveling through multiple international locations with her banking card. Traditionally, this would call for virtual production with LED walls displaying high-resolution background plates.
“VP is good, but you need obviously a really big screen to pull off some of the shots they had,” explains Lucien De Vivo, our lead on the project. “They said that the screens they have in Turkey aren’t great, so they were left to blue screen.”
Rather than compromise on the vision, PTOT Films turned to AI background generation—a decision that would reveal both the creative possibilities and technical complexities of this emerging approach.
“The director wanted a tilted camera, high tilting up with the sun kind of coming from a certain direction…all of those things we could dial in with AI within reason.“
The Hybrid Advantage: Where Traditional Expertise Meets AI
What made this project work wasn’t just the AI tools—it was the integration of traditional filmmaking knowledge with generative technology. When the director specified camera types, lens choices, and lighting preferences, this information became crucial input for the AI generation process.
“All of that helped kind of reinforce and pump into the AI imagery because we could use those nuggets like 35mm lens or an 80mm lens,” De Vivo notes. “All of those things help change what the picture looks like, what the background looks like.”
This hybrid approach proved essential. The team’s understanding of film language, lighting, and camera work informed every stage of the AI generation process — from initial concepts through final compositing.
Creative Control vs. Creative Lottery
One of AI background generation’s key advantages over stock footage or traditional virtual production is flexibility. Where stock imagery locks you into existing perspectives and angles, AI allows directors to dial in specific moods, lighting conditions, and camera positions.
For the New York City sequence, the director wanted “a tilted camera, high tilting up with the sun kind of coming from a certain direction,” De Vivo explains. “All of those things we could dial in with AI within reason.”
But this flexibility comes with a caveat: the generation process can feel like “a bit of a lottery,” requiring multiple iterations and careful refinement to achieve the desired result.
The Complex Workflow Behind Simple Results
The technical process reveals why AI background generation isn’t the instant solution some imagine:
Concept Development
Using tools like Midjourney and Google’s Veo to generate initial reference images.
Upscaling
Running approved images through tools like Magnific AI or Topaz Gigapixel.
Video Generation
Converting static backgrounds to video using platforms like Runway or Kling.
Re-upscaling
Final video enhancement through tools like Starlight.
Traditional Compositing
Standard keying, tracking, and depth work.
“It’s a bit of a weird workflow to be honest,” De Vivo admits. The constant upscaling and downscaling between different tools creates multiple potential failure points and extends timelines significantly.
When AI Makes Sense (and When It Doesn’t)
The PTOT project took approximately six weeks for six shots — not dramatically faster than traditional methods. So when does AI background generation make sense?
De Vivo sees it as “another tool” rather than a replacement: “If a client came along and said, ‘Look, I don’t like how the Unreal scenes look. They look fake. I want it to look more realistic,’ then without having to go down the traditional matte paint route, it’s just another option.”
The technology shows particular promise for:
• Abstract or fantastical environments that would be expensive to create traditionally.
• Projects where specific creative control over mood and lighting trumps precision
• Scenarios where traditional virtual production infrastructure isn’t available
• Concepts that benefit from rapid iteration and experimentation
The Reality of Creative AI
Perhaps the most important insight from this project is that AI background generation and 3D modelling for AI-driven animation (for example), isn’t about replacing human creativity — it’s about augmenting it with new possibilities. The technology requires significant human oversight, traditional production knowledge, and patient iteration to produce professional results.
“It needs to be done with purpose,” De Vivo emphasizes. “You know, I want to have someone set on the moon with asteroids flying behind or something. And then that’s something that you could say, ‘Yeah, I could do that.'”
“It needs to be done with purpose“
Looking Forward: Evolution, Not Revolution
As AI video generation tools continue improving — particularly in resolution and consistency — we can expect more streamlined workflows. The current multi-step process of upscaling, generating, and re-upscaling will likely consolidate into more integrated solutions.
But the fundamental insight remains: the most successful AI applications in production will be those that leverage the technology’s strengths while acknowledging its limitations. AI background generation offers exciting creative possibilities, but it’s not magic — it’s a sophisticated tool that requires expertise, patience, and clear creative vision to use effectively.
For production teams considering AI backgrounds, the question isn’t whether the technology can create amazing results — it can. The question is whether your project, timeline, and creative requirements align with the current realities of the workflow. When they do, the possibilities are genuinely exciting.
Let’s Connect
As AI continues reshaping video production, the most successful approaches will be those that thoughtfully integrate new capabilities with proven production expertise — creating hybrid workflows that leverage the best of both worlds. Synima can help.
Head of Marketing at Synima