Outlook 2020: How AI can automate video conform & comparison
Anthony Matt
Issue: November/December 2019

Outlook 2020: How AI can automate video conform & comparison

In the late 1980s, TV studio executives had a clever, but short-sighted idea: What if they did not have to pay to cut the negative on their film-originated programs? They could save the expense of future proofing their studio’s content and take a shortcut to deliver only the Standard Definition (SD) video masters needed for broadcast. Certainly, the state of the art 1-inch Type C and D2 tapes would serve as perfect masters for the foreseeable future — or at least until the studio executives retired. 

Fast forward to today, and those analog and early digital masters are no longer serviceable. Thousands of popular episodic television programs were mastered to these obsolete formats, while the high-resolution film negatives remain in vaults with no editorial roadmap for retransfer to 4K HDR. Up-conversion has served to sell some of these shows to recent audiences, but they do not match the quality of 4K HDR remastering from film. As a consequence, audiences for these once popular programs continue to shrink, grow unfamiliar and fade away. 

How to make remastering faster and more cost-effective

Eye matching a TV episode’s dailies footage to the final SD master can be done manually, but this is a very time-consuming process, taking 60 to 100 hours of labor per episode. Artificial Intelligence (AI) can help address this challenge by automating the re-conform process. AI solution providers like Prime Focus Technologies have put in massive development efforts to build a solution that automates the processes of conform and comparison. It automatically matches the dailies proxies to shots used in the final edit and exports a new EDL. Then the selects of film can be re-scanned at high resolution. This significantly reduces the time and effort spent on eye-matching, making the process of remastering uncut film content up to five times faster. 

AI-led conform technology can detect a majority of the matching segments from original dailies footage — even if the content has VFX, zoom-ins etc. The machine learning process uses several image identification and processing techniques to compute the structural similarity between two scenes and validate the best match for a given scene. The solution can also compare different versions (like the network cut, international cut, etc.) instantly without having to manually compare the content shot by shot, resulting in almost complete automation of the version editing process.

Such tools enable content owners to greenlight projects that were once deemed too expensive to re-master. They can also help free up editorial teams, enabling them to focus on far more creative pursuits. To further reduce the cost of remastering, certain service providers also offer a package for film scanning, picture/sound restoration, color, VFX and deliverables creation, which can help broadcasters and studios re-release their legacy content with greater ease and showcase it across multiple platforms for global audiences.  

Anthony Matt is the Executive Director of Cloud Media Services for Prime Focus Technologies (www.primefocustechnologies.com), which has US offices in New York and Los Angeles.