Black Forest Labs on FLUX 3: one backbone trained on images, video and audio
Latent Space · youtube · 2026-11-17
Latent Space interviews Black Forest Labs' Andreas Blattmann (co-creator of Stable Diffusion, co-inventor of latent diffusion) and Dustin Podell (lead author of SDXL, led FLUX Kontext) on how FLUX 3 trains a single backbone across images, video and audio.
Key topics:
- What each modality contributes, how compute is allocated across denoising stages, and why details that look great in stills flicker across frames;
- The case for autoregression vs. where diffusion still wins; world action models and robotics in 2027;
- Personal history: how Stable Diffusion changed their careers, the first FLUX training run in a French chalet, the photo experiment behind FLUX Kontext, and real use cases from restoring family photos to clips with synchronized sound.
Full chapter timestamps and links to BFL's official FLUX 3 image/video blog posts included.
More from Multimodal
- Seedance 2.5 4K Now Available on fal — petewoodbridge · 2026-11-17
- Higgsfield launches Short Drama Studio: turn a script into a full season — azed_ai · 2026-11-17
- Seedream 5.5 and Canva Craft reshape the text-to-image price frontier — ArtificialAnlys · 2026-11-17
- Probing activations restores diversity in distilled image and video generators — _akhaliq · 2026-11-17
- ReDiverse restores output diversity in few-step distilled diffusion models — anand_bhattad · 2026-11-17
- FLUX 3 trains on images, video and sound: BFL on visual intelligence — Latent Space · 2026-11-17