Although New York-based startup Runway has come under fire for its data scraping and training practices—like many AI model providers these days—it’s working hard on recent, standout features for its realistic generative AI video platform.
Today the corporate announced on its X-Account and be Discord server that Gen-3 Alpha, the brand new AI video model it unveiled in mid-June 2024 that’s able to generating incredibly realistic videos from easy text prompts in seconds, now also accepts still images as prompts.
All the user must do is navigate to Runway's website, click on the “Try Gen 3 Alpha” language, they usually can be taken to a screen where they will upload images and/or proceed to enter text inputs to drive recent AI video generations of 5 or 10 seconds in length (user chooses). A ten-second generated video requires 40 credits through Runway's pay-to-play and subscription tiers, while a 5-second generation costs 20 credits.
Tests show that Runway's Gen-3 Alpha image-to-video update is surprisingly fast while still maintaining prime quality
We tested the feature briefly on my personal Runway account, and I used to be blown away by each the speed (it took lower than a minute to generate the video from a still image I provided) and the standard. Check out my still image (generated using the separate AI service Midjourney) and the video created using Runway Gen-3 Alpha embedded below.
Safety measures in place
The model routinely tries to detect and block the creation of videos from explicit still images or well-known personalities similar to politicians.
For some Runway subscribers, the feature was also online early and briefly over the weekend before it was removed and re-uploaded today.
There are loads of initial examples
On its X account, Runway also highlighted ten interesting videos created from still images using the Gen 3 Alpha model.
In his tweet, Runway wrote: “Image to Video is a significant update that greatly improves artistic control and consistency across your generations.”
Cristóbal Valenzuela, co-founder and CEO of Runway, introduced the brand new feature on his personal account on X with the straightforward phrase “It’s time,” followed by a post of his own image-to-video creation.
Neck and neck within the race for video AI – one legal battle still pending
Runway is one in every of several corporations – including OpenAI with its Sora model, Kuaishou Technology with its Kling AI model, Luma AI with its Dream Machine, and Pika with its eponymous model – vying to supply users fast, high-quality generative videos.
However, OpenAI's Sora shouldn’t be yet public, while the opposite models can be found for public use.
But with Runway's recent still-to-video update promising users one other strategy to create Hollywood-quality videos in a snap, it's clear that the complete film and video creation space is about to be turned on its head like never before.
But will the courts suppress these innovations on the behest of the aggrieved artists and owners of copyrighted material?
Finally, Runway is one in every of several AI corporations currently facing class motion lawsuits from developers who claim that the common practice of AI corporations scraping and training publicly posted material – including copyrighted material – without explicit permission, authorization, compensation, or consent is a violation of copyright law. How the courts choose this issue will undoubtedly have major implications for the current and way forward for AI video and artistic tools.
