Transform Still Images into Engaging AI Videos Using Runway Gen-3 Alpha

Despite recent controversies surrounding data scraping and training practices affecting many AI developers, the New York City-based startup Runway is advancing with impressive new features for its generative AI video platform.

Today, Runway announced via its X account and Discord server that its newly introduced Gen-3 Alpha AI video model, initially revealed in June 2024, now supports still images as prompts. This enhancement allows users to generate hyper-realistic videos within seconds from simple text or images.

To utilize this feature, users can visit Runway’s website, select the “Try Gen-3 Alpha” option, and upload still images or enter text prompts to create videos of either 5 or 10 seconds in length (user choice). Generating a 10-second video requires 100 credits, while a 5-second video costs 50 credits, according to Runway's pay-to-play and subscription tiers.

Gen-3 Alpha Performance

Preliminary tests of Runway’s new image-to-video functionality indicate not only rapid generation speeds but also high-quality results. During a brief test on my personal Runway account, I was impressed by the feature's efficiency—it took under a minute to generate a video from a still image, as shown in the examples embedded below.

Safety Measures Implemented

Runway’s model incorporates safety protocols to detect and block video generation from explicit still images or recognizable public figures, such as politicians.

This feature had a brief online release for select Runway subscribers over the weekend before being reintroduced today.

Initial Examples and User Engagement

Runway showcased 10 compelling videos generated from still images using the Gen-3 Alpha model on its X account. As stated in a recent tweet: “Image to Video is a major update that greatly improves the artistic control and consistency of your generations.”

Co-founder and CEO Cristóbal Valenzuela highlighted the new feature on his personal X account with a simple "it’s time," sharing his own image-to-video creation.

A Competitive Landscape in AI Video

Runway is competing with other companies like OpenAI (Sora model), Kuaishou Technology (Kling AI), Luma AI (Dream Machine), and Pika, all striving to deliver fast, high-quality generative video. However, OpenAI’s Sora remains private, whereas the other models are accessible to the public.

With Runway's latest image-to-video functionality, the landscape of filmmaking and video creation is shifting dramatically, enabling users to produce Hollywood-quality content on demand.

Yet, the future of innovation may hinge on ongoing legal challenges posed by creators claiming that AI firms engaging in data scraping without permission violate copyright laws. The outcomes of these class-action lawsuits could significantly shape the trajectory of AI video tools and creative software.