runway generative AI tool ‘gen-2’ makes realistic movies with just words

Text-to-video AI gen-2 by runway

Runway’s recently unveiled AI tool Gen-2 can produce movies out of words. Midjourney churns out hyperrealistic AI images that can easily invite viewers into believing what they see is real. The text-to-image program cuts down the invested time in producing images by relying on the user’s written words. Yet the artificial intelligence industry may want to tap more into the system, to tweak and test it to awaken its full potential.

The entrance of the publicly available AI program Gen-2 by Runway then poses to be a rational follow-up on the progress of artificial intelligence. The program makes these once dormant, surreal images come to life. ‘If you can say it, now you can see it,’ says Runway. The AI program does, in fact, fulfill its claim. It analyzes the written words and in minutes – even seconds – the program produces a video based on the prompt the user feeds into its system.

No lights. No camera. No heavy equipment. Gen-2 doubles as a video-ready production tool that realistically and consistently synthesizes new videos. ‘It’s like filming something new, without filming anything at all,’ says Runway. In one of its video models, Runway types ‘the late afternoon sun peeking through the window of a New York City loft’ and puts together a scenic view of the city from the window inside a plant-covered apartment. Albeit the video is short, the sun sets, the wind blows, and the plants sway, all visible in plain sight.

video clips courtesy of Runway

Content creation without complex settings

Gen-2 by Runway creates more realistic and surreal videos compared to its predecessor. The sharper pixel quality gives it away, but the videos’ GIF-like flow may sneak up and cue the viewers that the videos have been created through an AI program alongside the unusual characters spotlighted in the movies (thanks to the users’ input).

Based on the initial comments of those who have tried Gen-2 as soon as it went live, there seems to be a 3-second limit to the videos that Gen-2 can produce. Runway is yet to comment on this, but what the team has affirmed so far is that its AI tool is available to the public. The emergence of text-to-video AI programs like Gen-2 allows general users to create video content without relying heavily on the complex settings of movie making.

These AI tools open up more accessible options to those who may find conventional video editing software and tools a challenge to navigate. Educators, marketers, and even students who need to spice up their presentations can also turn to text-to-video AI to help them creatively simplify or tug the attention of their audience toward their reports and lectures.

users can just jot down what they want to see in a video

Potential risk of deepfake?

The release of Gen-2 coincides with the unraveling of other text-to-video, AI-driven programs. Philipp Tsipman, the founder of generative AI video platform Camcorder AI, posted that in a span of seven days, five AI video generators showed up, including Runway’s Gen-2. The news signals how fast the teams behind the artificial intelligence industry want AI’s abilities to drastically improve.

A few users on Twitter have expressed their excitement about Runway’s Gen-2 and toys with the idea of how text-to-video programs might change the video-making game. While the creative frenzy buzzes around, potential risks still sit around the corner as AI tools like this can also be used to drive deepfake content forward, especially if governing policies on the boundaries of its usage is not enforced.

Elon Musk and Jeff Bezos have joined the Star Trek realm before while a doctored video of Mark Zuckerberg demonstrates the dangerous power of deepfake content. With the arrival of an easier interface and better image quality of the AI programs, text-to-video creating might all be fun until a user instill a non-verified idea to the public, pushes it as the truth, and supports their claim with a lifelike clip.

Gen-2 might still be limited in its ability to understand the user’s request in some cases