AuraFlow v0.x series, to date the largest (6.8B) and highest fidelity (0.7+ on GenEval) open sourced text to image model.
Create short AI videos from text or first and last frames, with synchronized dialogue, effects, and ambience.