Adobe’s AI video model can generate videos and extend clips

Software giant, Adobe, has unveiled a new AI video model that can generate videos with both text prompts and still images. The Firefly video model takes on OpenAI’s Sora, Google’s Veo, and the recently announced Meta Movie Gen. Here’s everything about Adobe’s new AI video tool.

What is Adobe Firefly?

Adobe Firefly video model’s text-to-video capability can help you generate high-quality videos with a few prompts. However, it can only generate videos that are about four to five seconds long. But the good thing is that the model was trained on licensed content, unlike some other tools that use publicly available content.

For comparison, OpenAI’s Sora and Meta’s Movie Gen rely on publicly available and licensed content. Both models can generate videos that are up to a minute long.

Adobe also announced the image-to-video feature that can turn still images into short clips. Again, the generated videos are about four to five seconds. Still, this feature appears to be impressive, as it can let you bring your old photos to life.

However, as per some samples shared by Adobe, the image-to-video output turned out to be a bit jittery, as the AI model made some unnecessary parts move and wobble.

ALSO READ: Adobe Acrobat update adds handy new generative AI features

Other than this, Adobe also showcased its Generative Extend tool, which is launching in beta for Premiere Pro. The tool lets you extend the beginning or end of the footage, so your videos can reach just the length you want for being suitable for, say, YouTube.

Generative Extend, however, is limited as it can extend clips by only two seconds. It can generate videos in 720p or 1080p at 24 fps, depending on your preference. The feature can also extend sound effects in your footage by up to 10 seconds. But it cannot extend music or spoken dialogue in a video.

ALSO READ: With Adobe’s new generative AI tool, you will be able to produce music with text prompts

Adobe confirmed that these tools take about 90 minutes to generate videos, which is a long time. However, it plans to add a ‘turbo mode’ to shorten the processing time.

Adobe’s new AI video model looks great, and we can’t wait to use its features, especially, Generative Extend in Premier Pro.