Artwork
iconShare
 
Manage episode 521030855 series 74017
Content provided by Monstrous Moonshine. All podcast content including episodes, graphics, and podcast descriptions are uploaded and provided directly by Monstrous Moonshine or their podcast platform partner. If you believe someone is using your copyrighted work without your permission, you can follow the process outlined here https://staging.podcastplayer.com/legal.

AI is revolutionizing cinematic lighting control with Beeble AI's Switchlight. Founder Hoon Kim explains how his tool, originally a general AI concept, became a powerful VFX asset by tackling the difficult process of relighting. Switchlight "unlights" any video footage to figure out the fundamental physical properties, like the shape (normals) and texture (metalness/roughness) of objects, and then uses this data to apply new, photo-realistic lighting instantly and securely. The desktop application is quickly becoming indispensable for both small production teams and major studios who need precise creative control over their shots.

The conversation reaches a pivotal point when host Chris, an expert in real-time rendering, mentions his work with the real-time ray tracer Vantage, leading to mutual excitement about integrating their technologies. Switchlight provides the control that other generative AI tools lack, and Hoon sees its PBR data as a perfect control signal for future generative video models. They agree the tool's true value is creating "relightable footage" that can be manipulated layer by layer, just like in professional compositing software. This technical precision promises to blend the creative freedom of AI with the consistency and detail demanded by professional filmmaking.

Beeble AI Official Website >

Hoon Kim's LinkedIn Profile >

Google Research Total Relighting paper >

Paul Debevec's Parthenon project (2004) >

Chaos Vantage (real-time ray tracing) >

This episode is sponsored by:

Center Grid Virtual Studio

Kitbash 3D (Use promocode "cggarage" for 10% off)

  continue reading

357 episodes