Why text-to-video may be the next ‘big’ AI thing – Times of India

In the case of generative AI, there’s just one factor dominating the headlines — ChatGPT. Nevertheless, there’s much more on the planet of generative past ChatGPT-like language fashions. Textual content-to-image is already turning into part of mainstream conversations however brewing within the background is generative AI able to changing textual content to movies.
What’s text-to-videos AI?
Merely put, you’ll be able to generate AI-powered movies on the idea of nothing however your phrases. Sure, it’s precisely the way it appears like: key within the textual content and the AI mannequin will generate a video based mostly on it. US-based startup Runway showcased its Gen-2 mannequin, which is ready to do this with a caveat or two.
Is that this a ‘new’ factor?
Not likely as it is extremely very like Dall-E — developed by creators of ChatGPT — and works utilizing generative AI language fashions. The outcomes are fascinating sufficient and it may definitely catch the flowery of many the world over.
Is ‘Large Tech’ not concerned in text-to-video?
They very a lot are. Again in September 2022, Meta showcased a moderately clearly named software Make-A-Video. With only a few phrases or traces of textual content, Make-A-Video creates movies utilizing generative AI however these movies didn’t have any sound. Right here’s what Meta CEO Mark Zuckerberg had mentioned about it: “It’s a lot tougher to generate video than pictures as a result of past appropriately producing every pixel, the system additionally has to foretell how they will change over time.
Only a week later and on cue, Google introduced an identical mannequin. Google’s generative AI mannequin is named Imagen Video. “Given a textual content immediate, Imagen Video generates excessive definition movies utilizing a base video technology mannequin and a sequence of interleaved spatial and temporal video super-resolution fashions,” is how Google had described it.
Google additionally showcased one other mannequin referred to as Phenaki, which is aimed toward creating long-form movies on the idea of textual content inputs.
What are the challenges with text-to-video AI?
Multifold. From operational to moral, the challenges are far too many. Maybe that’s one of many the reason why solely demos of generative AI fashions engaged on text-to-videos have emerged. For starters, producing a video with textual content may sound ridiculously simple and equally fascinating however think about making a video with simply phrases. One must be extremely exact with the instructions or it may generate the video equal of gibberish.
Then comes the moral challenges. AI-generated movies could possibly be the following weapon within the misinformation arsenal. Deepfakes may change into a good greater drawback that’s at the moment encountered.
Contemplating the fast-paced developments within the subject of AI, it could possibly be a matter of time earlier than text-to-video will get out of exploration mode and change into moderately mainstream.

Image / Information Source

Leave a Reply

Your email address will not be published. Required fields are marked *