<p>Another step towards generating video from text.</p>
<p>Google releases Phenaki - video generation based on descriptions. The main differences from previous attempts are two:</p>
<ol>
<li>The model can generate longer and more stable clips.</li>
<li>The description can contain a sequence of scenes (like a script) that will control the flow of the video.</li>
</ol>
<p>I think the time when we will watch personalized generated movies is not far off.</p>
<p>More details <a href="https://phenaki.video/">here</a></p>;
<p>Paper: <a href="https://openreview.net/pdf?id=vOEXS39nOF">here</a></p>;
<p>P.S. Interestingly, in the world of Generative Video and Generative AI, the equivalent of “Hello world” has become the generation of video and images of an astronaut sitting on a horse.</p>
<p>#ai #video #text2video #google</p>