What happens when AI becomes part of the production pipeline? A new experiment in Los Angeles offers an interesting glimpse at where virtual production workflows may be heading.
A recent CNBC report takes viewers inside Innovative Dreams, a new production venture bringing together virtual production, performance capture, VFX and generative AI within what the company describes as a “Realtime Hybrid Filmmaking” workflow.
Its first major test is The Old Stories: Moses, a three-part production starring Sir Ben Kingsley. The production was prepared in a matter of weeks and shot on a Los Angeles soundstage in just one week, using virtual production alongside AI-generated environments and imagery. Director Jon Erwin told the Los Angeles Times that AI was used for elements including wide shots, large crowds and stunt-heavy sequences, while the Red Sea sequence could be explored through multiple generated variations before being presented to Kingsley on the LED stage. The speed is striking.
For the Virtual Production Studio Network, however, the more interesting story is what this tells us about the continued evolution of production workflows.
Virtual production is becoming part of a much wider technology ecosystem
Virtual production has always brought different disciplines closer together. Real-time engines, camera tracking, LED volumes, VFX, cinematography, production design and traditional filmmaking knowledge increasingly meet on the same stage. That convergence has already required crews to develop new ways of communicating and collaborating.
Generative AI adds another layer. Innovative Dreams describes its approach as combining performance capture, virtual production and visual effects, with generative AI supporting stages from early concept development and previs through production and post-production. That makes this particularly relevant to the VP community.
AI is beginning to interact with the infrastructure and workflows already developing around virtual production. Sets and environments can be explored differently. Visual ideas can be iterated rapidly. Backgrounds, crowds, extensions and other elements can potentially be generated or adapted during the filmmaking process. The boundaries between pre-production, production and post-production become increasingly fluid. And that changes the skills required around the stage.
Understanding the technology matters
Every major technological shift in filmmaking brings a period of uncertainty. Generative AI has accelerated that uncertainty considerably. There are legitimate questions around employment, copyright, consent, creative ownership, sustainability and the long-term structure of production. The CNBC report places the Innovative Dreams experiment within the wider contraction of production employment in Los Angeles and asks an important question: can these technologies help more productions become financially viable, or will efficiencies ultimately reduce employment? There is unlikely to be a simple answer.
What we can say with considerably more confidence is that people working across film, television, education and virtual production need opportunities to understand how these systems actually work. That means moving beyond discussions about AI in the abstract. We need opportunities to test workflows, understand their limitations, identify where human judgement remains critical and examine how individual technologies connect across a production pipeline.
This is particularly important for students and early-career practitioners who may enter an industry where these workflows are already part of the production landscape.
Access will shape who gets to experiment
There is another important dimension to the conversation: accessibility. One of the long-standing promises of virtual production has been the ability to create locations, environments and production possibilities that might otherwise sit beyond the resources of a project.
AI potentially pushes that further. A filmmaker may be able to explore more concepts before committing resources. Smaller productions may be able to attempt environments or visual effects previously outside their budgets. Educators may be able to introduce students to sophisticated production concepts without recreating every element of a traditional studio pipeline. Of course, accessibility involves more than having access to software.
There is a substantial difference between opening an AI application and understanding how to incorporate it responsibly and effectively into a professional workflow. Access therefore needs to include technology, infrastructure, knowledge and experimentation. This is where studios, universities, training organisations and collaborative networks have an important role to play.
New workflows require new conversations
Perhaps the most interesting aspect of Moses is how many technologies and disciplines are being brought together. Innovative Dreams operates a virtual production stage at the MBS Media Campus in Manhattan Beach and positions itself as both a production services company and an R&D environment. Its model brings filmmakers and creative technologists into the same production process, supported by AWS infrastructure and Luma’s generative technology.
That kind of convergence makes interdisciplinary knowledge increasingly valuable. A cinematographer does not necessarily need to become an AI engineer. A director does not need to become an Unreal developer. A production designer does not need to understand every component of a cloud infrastructure stack.
But everyone benefits from understanding what the other technologies and disciplines can contribute. That shared understanding improves communication across the production. It also gives creative teams a much stronger position from which to make decisions about when a technology is useful, when another approach is better, and where its limitations or ethical implications need to be considered.
The value of experimentation
For VPSN, this is ultimately why projects such as Innovative Dreams are worth watching. They give the industry something tangible to examine. Rather than speculating about what an AI-enabled production pipeline might eventually look like, we can begin studying productions that are actively combining these technologies at scale.
Some approaches will become established practice. Others may prove less useful than anticipated. New roles will emerge, existing roles will evolve, and the technology itself will continue changing. The ability to experiment, evaluate and share knowledge throughout that process will be critical.
Virtual production has already demonstrated how quickly filmmaking workflows can evolve when previously separate disciplines begin working together in real time.
Generative AI introduces another participant into that increasingly connected production environment.
Understanding how it fits — creatively, technically and responsibly — is becoming part of understanding the future of virtual production itself.
