The Neuron: AI Explained AI Just Democratized Filmmaking (w/ LTX Co-Founder)
6 snips
Mar 12, 2026 Yaron Inger, CTO and co-founder of Lightricks and LTX, leads development of open-source video and audio generative models. He discusses LTX-2’s rise, why open weights matter for control and fine-tuning, and how local, consumer-GPU workflows enable new filmmaking tools. Topics include practical editing flows, integrating 3D/animation, distillation for efficient models, and possibilities for interactive, real-time educational video.
AI Snips
Chapters
Transcript
Episode notes
From Facetune To Foundational Video Models
- Yaron described Lightricks' origin: five PhD founders leaving academia to build mobile content apps like Facetune.
- Facetune bootstrapped them to ~30 employees and Apple awards before later pivoting to generative AI.
Open Models Enable Local, Private Video Workflows
- Open weights let creators run, fine-tune, and keep IP private by hosting models locally instead of relying on closed APIs.
- Yaron Inger highlights LTX-2 runs on consumer GPUs (4090/5090) and can be fine-tuned on private data for studio workflows.
Run LTX-2 Locally With ConfUI Or LTX Desktop
- Use ConfUI for graph-based experimentation or LTX Desktop for an opinionated nonlinear editor when running LTX-2 locally.
- LTX Desktop runs generations on your timeline locally and offers flows like first/last frame, audio-to-video, and retake editing.