Today, OpenAI - a company behind ChatGPT - released a new text-to-video model called "Sora", which can generate
extremely impressive video outputs.
I recommend clicking the link above and seeing them for yourselves because it's something you should see to believe.
I've often compared AI-assisted art generation to filmmaking from a director's perspective, in that you actualise your artistic vision by directing other people (or an AI) who have raw skills(i.e. acting, CGI, script writing, cinematography, lighting, etc.). With technology like Sora, the parallel can be more literal so that you'd be able to make your feature (*cough* porn *cough*) films at home.
There are still many obstacles to overcome until we get to that stage, though. First of all, unlike the case with image generation, we don't have fine control over how the final output would look. It'll take some time before we'll see an equivalent of control nets or some GUI tools resembling traditional 3D animation/video editing tools to address the issue.
Secondly, it may take quite some time before the open-source community can reach the level achieved by companies like OpenAI. Despite its name, OpenAI has nothing to do with open source, unlike AI companies like Stability or Mistral, which endorses and contribute to the FOSS community.
If you're not a supporter of the FOSS movement, you may think it wouldn't matter. But if anything else, most of the AI services provided by proprietary companies like OpenAI are heavily censored, so you aren't likely to generate much kinky content with them.
On the brighter side, projects like Sora show how exponentially fast things are moving in the AI field now, which gives some hope that the open-source community will keep evolving rapidly as well, even though they'd be several steps behind big corporations.
