[16:38 Mon,2.September 2024 by blip] |
Many of the most recently presented video AIs are Chinese developments, such as
Where OpenAI&s Sora had taken a first step forward, MiniMax seems to have taken a leap - in demo clips posted online, the people depicted walking, running and dancing actually produce quite stringent figures. Although you can still see that they come from a computer, and the kung fu still resembles a marionette fight, these examples are once again on a previously unseen level. At least as far as the main characters in the foreground are concerned - crowds of people in the background wobble and morph here too, as do cars flowing into each other and the like. Not every generation is equally convincing either; the movements don&t always work. Despite these persistent errors, MiniMax also displays an astonishing temporal coherence. MiniMax currently only supports text-to-video generation (more options are to follow), and it appears that prompts in English are also supported. In principle, the generator can also be used internationally after registering with a mobile number, and for the time being even free of charge. The maximum clip length is 6 seconds, the resolution 1280x720. As one of the major Chinese AI start-ups, MiniMax has already been offering text and image generation for some time, but the focus is on “social” applications, such as role-playing apps with AI characters.
Bilder zur Newsmeldung:
 more infos at bei hailuoai.comdeutsche Version dieser Seite: Neue kostenlose Video-KI MiniMax meistert menschliche Bewegungen |
;)
;)
 |
|
|