Google just launched Gemini Omni 1.1 Flash, and the real headline is its upgraded long-form video creation. Instead of being stuck with one-shot short clips, creators can now extend footage like editing a timeline: the model can reference up to 10 seconds of earlier context, then generate ~10 seconds more each step—stacking up to a maximum of ~40 seconds. Compared to older versions that mostly relied on only the last second, this should help keep characters, scenes, and storytelling continuity much more stable. Even better: it introduces “first-and-last frame control.” You provide just the opening and ending frames (two anchor images), and the model fills in the middle automatically to produce a continuous sequence. That’s a big deal for teams that need smoother pacing, transitions, or repeatable playback loops. Google also added a 360p fast draft mode (up to 60% faster, ~1/3 the cost vs 720p) so you can iterate quickly, then upscale later to 1080p or even up to 4K via post-processing. Plus, you can use up to 3 seconds of video as motion reference. #AIvideo #GenerativeAI #VideoEditing #CreativeTools #GoogleAI #MLOps
Want to learn more? Visit Explore the world, stay updated on travel insights and international affairs, and discover authentic stories from real life
评论
发表评论