Issue link: https://digital.copcomm.com/i/1524965
ARTIFICIAL INTELLIGENCE www.postmagazine.com 26 POST JULY/AUG 2024 CREATING BACKGROUND ASSETS WITH GENERATIVE A.I. Generative AI's ability to act as a power multiplier is amazing, making light work of previous heavy lifting. But as sci-fi has taught us, it matters quite a lot who is driving that mech suit. The more skilled the artist at the helm, the greater the effect is. We've developed ways to give creative control back to our artists while using our in-house GenAI and Adobe's Firefly, moving away from the slot- machine-like randomness that makes it so hard to use in a production environment. One of the ways we've been using AI is to create deep background assets for commercials, such as for Continental Tire and Genesis. With our artist's ability to quickly create layers of city that would match the client's look and iterate skies, we were able to generate an expansive and vibrant city- scape with limited reference from the client and offer them more options than we would have pre- viously in the same amount of time. We can use the same techniques during postvis to quickly iterate set extensions or paint fixes. We've also been able to bring on illustrators in ways we have never been able to before, as they can work loosely to capture the image, and the AI can assist in getting it to a far more finished result in less time, while the illustrator can correct and refine on top. This has allowed us to offer an entirely new set of visual styles, leading to several 2.5D pitch projects, with more to come. We do our own training and fine tuning of AI models in-house, allowing us to add our client's characters or environments to the base models or to generate a constant look. Because the AI space is so new, it's up to us to put our own guide rails in place. We never use outside artist's names in our prompts, or train on art that isn't ours or our clients'. Ultimately, GenAI is a fantastic and evolving tool, and we are committed to keeping artists at the helm as we evolve with it. MTI FILM USING A.I. FOR FRAME-RATE CONVERSION, RESTORATION HOLLYWOOD — MTI Film (mtifilm.com), known for its work in television post production and film restoration, is making significant strides in the realm of artificial intelligence (AI) with the development and deployment of MTai. The tool is derived from an open-source platform and has been refined by a ded- icated team of MTI Film developers who understand the process and math of image processing. MTai addresses the challenging issues of using AI in post production, particularly in frame-rate conversion and the repair of missing or highly-damaged frames. The company recognized the limitations of traditional frame rate conver- sion methods, which often resulted in subpar visual quality. To tackle this, they embarked on the creation of MTai, a generative AI tool designed to synthesize new frames and deliver smooth, high-quality results. MTai Frame Convert was developed to overcome the complexities involved in converting footage to dif- ferent frame rates while preserving the original aesthetic integrity and motion inside the same time base. MTai was employed for the remastering of Walker, Texas Ranger, star- ring Chuck Norris as a member of the Texas Ranger Division. The episodes, which were originally captured on film and transferred to video at 29.97fps, needed conversion to UHD at 23.976fps. Traditional methods like inverse telecine or drop-frame processes were inade- quate for this task. MTai Frame Convert, integrated inside of the company's Cortex post production system, synthesized new frames, ensuring a seamless and smooth visual transition. This approach was applied to multiple episodes and extend- ed to other shows, including CBS's Ghosts, which was acquired in the UK at 25fps and converted to 23.98fps using MTai. In fact, during COVID, when no production was taking place, MTai Frame Convert was used to repurpose the content for the US market. MTI Film introduced MTai FrameGen at NAB last April. The tool is designed for repairing and recreating missing or damaged frames in restoration content and harnesses the power of generative AI to reconstruct frames from jump cuts and frame slugs, offering restoration capabilities that surpass traditional image-based clean-up tools. MTai FrameGen was recently used in the digital restoration of the film Topkapi (pictured below). Jump cuts and frame slugs were seamlessly replaced, with the newly-generated content matching the motion, grain and color of the original source. "MTai FrameGen is now integrated into DRS Nova V.6.0, MTI Film's gold stan- dard in film-restoration toolsets," explains MTI Film CEO, Larry Chernoff. "This innovative addition, combined with existing tools for deflicker, paint, dewarp, and DRS Nova manual tools, empowers artists to tackle restoration projects with a new level of confidence and efficiency with stunning results." BY LINDSEY ROBBINS LEAD VAD ARTIST HALON ENTERTAINMENT HTTPS://HALON.COM