When it comes to generative AI, only ChatGPT is dominating the headlines. However, the generative world has much more than language models like ChatGPT.Text-to-image conversion is already becoming part of the mainstream conversation, but brewing in the background is generative AI You can convert text to video.
What is Text-to-Video AI?
Simply put, you can generate an AI-powered video based solely on your words. Yes, exactly. Enter text and the AI model will generate a video based on it. US-based startup Runway showcased a Gen-2 model.
Is this “new”?
Not really, as it is very similar to Dall-E, developed by the creators of ChatGPT, and works using a generative AI language model. The result is attractive enough, and there is no doubt that it will fascinate many people around the world.
Is “Big Tech” Involved? text to video?
they are very much. Back in September 2022, Meta introduced a tool with a rather self-explanatory name, Make-A-Video. Make-A-Video uses generative AI to create videos with just a few words or lines of text, but those videos had no audio. Meta CEO Mark Zuckerberg said:
Just a week later, Google announced a similar model. Google’s generative AI model is called Imagen Video. “Given a text prompt, Imagen Video uses a base video generation model and a sequence of interleaved spatial and temporal video super-resolution models to generate high-definition video,” Google said. was
Google also introduced another model, called Phenaki, aimed at creating long-form videos based on text input.
What are the challenges of text-to-video AI?
multifold. From operations to ethics, the challenges are too many. Perhaps that’s one reason why there’s only a demo of a generative AI model that converts text to video. For starters, making a video with text is ridiculously easy and sounds just as appealing, but imagine making a video with just words. You need to be incredibly precise with your commands. Otherwise, the equivalent of gibberish video may be generated.
Then there is the ethical issue. AI-generated video could be the next weapon in the misinformation arsenal. Deepfakes could be an even bigger problem we’re encountering right now.
Given the fast-paced development in the AI space, it may only be a matter of time before text-to-video conversion gets out of exploration mode and becomes mainstream.
What is Text-to-Video AI?
Simply put, you can generate an AI-powered video based solely on your words. Yes, exactly. Enter text and the AI model will generate a video based on it. US-based startup Runway showcased a Gen-2 model.
Is this “new”?
Not really, as it is very similar to Dall-E, developed by the creators of ChatGPT, and works using a generative AI language model. The result is attractive enough, and there is no doubt that it will fascinate many people around the world.
Is “Big Tech” Involved? text to video?
they are very much. Back in September 2022, Meta introduced a tool with a rather self-explanatory name, Make-A-Video. Make-A-Video uses generative AI to create videos with just a few words or lines of text, but those videos had no audio. Meta CEO Mark Zuckerberg said:
Just a week later, Google announced a similar model. Google’s generative AI model is called Imagen Video. “Given a text prompt, Imagen Video uses a base video generation model and a sequence of interleaved spatial and temporal video super-resolution models to generate high-definition video,” Google said. was
Google also introduced another model, called Phenaki, aimed at creating long-form videos based on text input.
What are the challenges of text-to-video AI?
multifold. From operations to ethics, the challenges are too many. Perhaps that’s one reason why there’s only a demo of a generative AI model that converts text to video. For starters, making a video with text is ridiculously easy and sounds just as appealing, but imagine making a video with just words. You need to be incredibly precise with your commands. Otherwise, the equivalent of gibberish video may be generated.
Then there is the ethical issue. AI-generated video could be the next weapon in the misinformation arsenal. Deepfakes could be an even bigger problem we’re encountering right now.
Given the fast-paced development in the AI space, it may only be a matter of time before text-to-video conversion gets out of exploration mode and becomes mainstream.
