SBS Optimizes Their Special Effects Archive with TwelveLabs

Customer Profile

SBS is one of the leading South Korean television and radio broadcasters. From their popular dramas to sports, news, reality TV, and more, SBS has been a pioneer in the wave of Korean entertainment that has gained worldwide popularity in recent years. SBS aims to continue to connect with audiences across the globe, while maximizing and monetizing their vast library of video content, including special effects from both their corporate archive, as well as footage uploaded and stored by individual team members.

"TwelveLabs' multimodal AI technology can be an excellent choice, as it has already proven to be among the best in the world."

— Senior Director, SBS Technology R&D Center

Executive Summary

SBS's vast content archive is a treasure trove of scenes and special effects that hold tremendous value, yet were hard to access and time-consuming to manage, until TwelveLabs AI. This case study examines how implementing TwelveLabs multimodal AI models helped SBS enable reuse of their media assets, and enable scene-level search across internal and individual archives.

Challenges

SBS faced several challenges in their content operations.

1. Content Retrieval — Locating previously archived visual effects was difficult and time-consuming.

2. Access to Individual Team Uploads — Individual team members uploading their own work makes for another layer of metadata that goes unaccounted for in the corporate archive.

3. Lack of Scene-Level Search — Trading and sharing footage based on a particular scene is a time-consuming process requiring deep-dives into mass amounts of video data.

Solution

Implementing TwelveLabs' multimodal AI technology with SBS's proprietary scene search technologies would address these challenges.

Phase 1 — VFX Reference Search & Scene-Based Retrieval

Implementation

• Development of scene search service using Marengo 2.7 for dramatic special effects production

• Create comprehensive video embeddings capturing visual elements, spoken content, and text overlays

• Index footage by semantic content rather than just keywords

• Generate timestamps for key moments, relevant shots, and VFX types

Benefits

• Immediate access to searchable video content

• TwelveLabs will add much more searchable detail

• Automated identification of key VFX elements

• Reduced time spent on initial content search

• Supports quick access to high-demand scenes for domestic and international trading and distribution

Phase 2 — Intelligent Statistical Analysis

Implementation

• Development of a statistical analysis service to identify trends in digital content production and consumption across various topics

• Integrate Pegasus 1.2 to automatically suggest clips based on context/prompts

Benefits

• Detects overlooked issues as it pertains to digital content trends and consumption

• Identifies areas of expansion beyond news (or other verticals) to diverse genres

• Reduces manual search time

Phase 3 — Short-Form Summarization

Implementation

• Development of AI-powered short-form summarization service, using Pegasus 1.2 to enhance the efficiency of digital content production

Benefits

• Improved productivity and usability of original content by enabling quick text summaries for content re-purposing

Conclusion

The partnership between TwelveLabs and SBS demonstrates the transformative potential of advanced multimodal AI in the TV broadcast landscape. By implementing TwelveLabs' multimodal AI technology, SBS can streamline the VFX creation process by re-using or repurposing archived content, as well as finding specific scenes for domestic and international demand. This partnership facilitates broader workflow optimization by aggregating internal needs and integrating those needs with the power of AI. As both technologies and strategies continue to evolve, this collaboration sets a new standard for how video content can be created, distributed, re-purposed and experienced in the digital age.