Black Forest Labs' FLUX 3 Video release offers generation from text and images, plus continuation from up to four seconds of existing video and audio. The company says the continuation uses both picture and sound to carry movement, dialogue and other audio into the next segment. This gives creators an existing audiovisual starting point for extending a scene. Release description.
The announced clips run up to 20 seconds with generated audio. Output starts at 720p, with 1080p delivered through upscaling. That resolution distinction matters when assessing what the model generates directly. Output specifications.
Terms in this piece · Glossary
multimodal — A model that works with more than text — reading images, audio, or video, and sometimes generating them too.
