iCreatAI News
Skip to content
Trending storyWatching

Black Forest Labs on FLUX 3's multimodal training approach

1 reports1 sources1 days agoUpdated

Understand this story

AI overview

Andreas Blattmann and Dustin Podell of Black Forest Labs said on the Latent Space podcast that FLUX 3 is trained simultaneously on three signal types: still images, video, and audio. The two said this approach is intended to let the model learn visual intelligence from multiple modalities rather than relying on a single type of image data. What has been disclosed so far is the training approach; the podcast did not give FLUX 3's release date, performance figures, or availability.

AI overview based on source reports · Updated 1 days ago

Coverage timeline

Follow the coverage to explore different perspectives.

Nov 17
  1. Latent Space
    Building Visual Intelligence — Andreas Blattmann & Dustin Podell, Black Forest Labs

    In this podcast episode, Black Forest Labs' Andreas Blattmann (co-inventor of latent diffusion) and Dustin Podell (lead author of SDXL) discuss building visual intelligence, arguing that video and audio teach models things a still image cannot, and that FLUX 3 is trained on the three signals together. They cover flow matching, autoregression versus diffusion, the Kontext editing experiments, and how BFL spun out of Stability AI. Dustin also notes an unpredicted use case: people restoring old family photos with FLUX.

Activity for this story

Current comparable activity
5
Peak comparable activity
10Nov 17 06:00
Change in the past 24 hours
-50%

Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.