Adobe Launches Firefly AI Video Generator for Everyone: Here’s What to Expect

Adobe has made its AI generator for text and images to video available for anybody to test online. After a limited time of early access testing last year, Generate Video is now accessible in a public beta. Along with additional image generation and translation features, the beta tool is available through the updated Firefly web app, offering artists AI credit subscription tiers.

Although Adobe stated it would not announce prices for major users like studios until later this year, the company did unveil the first public version of an AI tool that can create video clips on Wednesday and disclosed its pricing.

By introducing a new stand-alone subscription service that allows consumers to access the company’s AI image, vector, and video-generating models, Adobe hopes to build on the early success of its Firefly AI models.

This is Adobe’s most audacious attempt to develop their Firefly AI models into a real product to date.

Additionally, the company is unveiling firefly.adobe.com, a revamped website where users can access Adobe’s AI models. The new Firefly AI video model is one of these, and it’s currently available in public beta on the Firefly website and the Premiere Pro Beta app.

Adobe’s Firefly Expands AI Video Tech

Adobe is referring to the service as the Firefly Video Model. It will compete with two other companies that presently provide video-generation services: startup Runway and Sora, a model created by OpenAI, the company that created ChatGPT. Meta Platforms, the company that owns Facebook, has also created an AI model for creating videos, but it has not said when it will be made public.

Read More: Adobe Launches a Global Initiative to Teach 30 Million People How to Use AI, Create Content, and Digital Marketing

Beginning with the Generative Extend tool in beta for Premiere Pro, which can be used to stretch the beginning or conclusion of a film, Adobe began releasing products in October that are powered by its generative AI Firefly Video Model. Today’s release of the Generate Video tool, which comes more than two months after OpenAI’s own text-to-video generator, Sora, was originally hinted at in September, includes a few minor enhancements.

Unlike its competitors, Adobe’s business strategy is focused on creating clips that work with Premiere Pro, its flagship video editing software, which is used by movie and television studios.

Therefore, a lot of the features that Adobe is highlighting are centered around feeding pre-existing photos into the video model and asking it to create films that correct or enhance shots that were taken on a real production set but didn’t turn out quite right.

Text-to-Video and Image-to-Video are the two features that make up Generate Video. As the titles suggest, Image-to-Video enables you to include a reference image with the prompt to provide a starting point for the video, while Text-to-Video enables users to create footage using word descriptions. Features like motion, shooting distances, camera angles, and style simulation are some of the features in Generate Video that help to improve or guide the results.

The original 720p quality of the video is now delivered at 1080p at 24 frames per second. Both Text-to-Video and Image-to-Video require at least 90 seconds to produce clips that are no more than five seconds long, which is less than the 20 seconds that Sora users can utilize. Additionally, Adobe says that a 4K model and a faster, lower-resolution “ideation model” are “coming soon.”

According to Adobe, the service will produce 1080p, five-second video. Adobe executives stated that most individual clips in most projects are only three seconds long, even though that is shorter than the 20-second snippets produced by OpenAI’s service.

According to Adobe, a user can create 20 clips for $9.99 and 70 clips for $29.99 each month. That contrasts with a $200 OpenAI subscription that can handle lengthier, higher resolution videos and 50 videos at a lower resolution for $20 a month with OpenAI’s plan.

Later this year, Adobe will make the pricing details of its “Premium” pricing plan available to studios and other heavy video customers. Adobe is working to create 4K video and will continue to prioritize quality over longer films, according to Alexandru Costin, vice president of generative AI at the company.

“We actually think that great motion, great structure, great definition scheme, making the actual clip look like it was film, is more important than making a longer clip that’s unusable,” Costin stated.

The Firefly web app, which houses a large number of Adobe’s generative AI tools, has also been updated. It now connects with Creative Cloud apps like Photoshop, Premiere Pro, and Express in addition to having a new user interface, which facilitates the transfer and editing of AI-generated content. Additionally, Firefly is safe for commercial use because it is trained on licensed and public domain content. To attract customers who wish to include AI-generated videos in movies without worrying about violating copyright protections, Adobe even markets its Generate Video tool as “production-ready.”

There is increasing competition for Adobe in the market for AI videos. Google is testing a second generation of their Veo AI video model in addition to Sora, which, based on early demo examples, appears to be more remarkable than OpenAI’s model. New generative AI tools with a video focus have also been announced by ByteDance and Pika Labs. Adobe’s main advantage is Firefly’s commercial viability, but it will still need to match the features and quality of its competitors.

Read More: Google’s Latest Generative AI Video Model is Now Available for Businesses

Starting today, the Firefly web app will also make two more tools available in public beta; however, they are not free to use. Using integrated 3D and sketching features, Scene to Image enables users to produce their own references for AI-generated photos. It appears to be based on Adobe’s “Project Scenic” project, which was revealed in October. The Translate Audio and Video tool, which lets users translate and dub audio into more than 20 languages while keeping the original speaker’s voice, is rather self-explanatory.

Adobe is introducing two new Firefly subscription plans that offer credits for using Adobe Firefly models. At $9.99 a month, Firefly Standard offers up to 20 five-second 1080p video generations and 2,000 video/audio credits. The more expensive Firefly Pro plan, which costs $29.99, offers up to 70 five-second 1080p video generations and 7,000 credits. Unlimited access to Firefly imaging and vector features is a noteworthy benefit of both programs.