ROUND TABLE
QJ: AI will replace the repetitive and boring tasks like conforming, derushing, but we’ll always want a human being in control. What we’re looking for is not just a button pusher, but an artistic input. Def: How is AI impacting virtual production techniques – and what are the practical and creative implications? DL: I’ve been impressed recently by the merging of analogue and digital processes in VP. VFX supervisors are commissioning model makers to construct miniature worlds, which are subsequently 3D scanned, and this data becomes the basis of the virtual production backdrop in a sound stage. Supervisors get real control over the VP imaging, and they retain some of the qualities of analogue production. QJ: In virtual production, I wouldn’t be surprised if LED walls ended up being
replaced by AI chroma keying and even AI live reflection of an IBL. Right now, I see the use of AI in virtual production mostly with scanning techniques for asset creation, a bit of DLSS and to help treat tracking data live. CN: Industrial Light & Magic pioneered the use of high-def LED video walls, now those are becoming commonplace. These days, AI programs like Cuebric work with Stable Diffusion and Unreal Engine to generate plates quickly and more cheaply. As Midjourney improves to more and more hyperrealism and 2.5 programs like Pika and Runway provide depth and motion, more AI studios will be born. In fact, we are building a hybrid AI studio, a model which uses real actors and crew but also uses the volume, AI- generated plates and Unreal Engine to create movies and documentaries. Def: Are there any specific AI tools you would recommend, and why? DL: So many! We should all be playing with Runway Gen-2. The important thing is to have creative fun with the technology, and then think hard about how it should or shouldn’t be used. CN: For pre-production, NolanAI looks good for scriptwriting (it’s a co-pilot program), while for previs and in some cases background plates, check out Midjourney and Pika. For post: the entire Adobe Creative Suite, Resolve and Topaz. MB: As well as Runway, I recommend Luma AI, which offers a convenient way
to turn videos of environments to 3D models and text to 3D generation. These tools are recommended for their ease of use and the value they add to the creative process. QJ: The first one I’d advise everyone to learn how to use is ChatGPT. It gives you a glimpse of the future, and from where we are standing now, we’re able to see the ‘guts’ of the machine and understand how it works. In five years from now, I’m not sure we’ll have the same crude tools we are using now: that’s when it may be too late for those who didn’t get to grips with them earlier. Def: As AI becomes more integrated into the industry, what do you think are the most important skills/knowledge areas for emerging filmmakers and professionals to gain? CN: I see the most important skills as creativity, imagination and storytelling. It goes back to the fine artists during the days of photography. They used their artisan skills to reinvent traditional art to become modern art. Impressionism, expressionism and abstract art. That’s how you keep ahead of AI, keep pushing the artistic vision of the industry farther into realms we’ve yet imagined – and use AI as a tool to get you there. You are still the visionary. AI follows your lead. DL: The accessibility of the new AI tools makes us think we should know how to use them all. But this is a danger for an emerging film professional: you still need to show that you have specific expertise in a small number of creative areas,
49
definitionmags
Powered by FlippingBook