When you buy through affiliate links in our content, we may earn a commission at no extra cost to you. Learn 
Adobe has confirmed that AI video generation features will be available to web and Premiere Pro beta users by the end of 2024.
The upcoming additions will run on the Firefly Video model. Premiere Pro will get a Generative Extend option that adds up to two seconds of extra footage to a given video. This can help if a clip doesn’t quite match the soundtrack, or if you need a smoother transition.
Firefly Text-to-Video and Image-to-Video will be available on the web. Both produce five-second videos (this may grow longer) based on either prompts or pictures. You can use menus to choose elements like the shot type and angle (such as a high-angle closeup) describe subjects and scenes in extensive detail.
With images, you can create wholly new clips based on a reference frame from an existing shot.
The Premiere Pro and web expansions came just as Adobe renewed its efforts to reassure creators that it wouldn’t let generative AI undermine their work. Adobe Stock now bars AI content that implies it covers “newsworthy” events. It also forbids the use of stock material to mislead people, and has “tripled” its moderation team size to catch abuses.
Firefly won’t produce videos involving alcohol, drugs, or nudity, and doesn’t include training data for public figures.
The company is also promising generative AI features that will help tweak Stock content for sale. You could change it to match a client’s branding or your personal tastes, for example. In theory, this leads to customized content that’s more likely to sell.
The emphasis on short, carefully controlled videos contrasts sharply with xAI’s Grok 2. It not only allows longer clips, but also doesn’t have many restrictions on public personas or subject matter. There’s a concern this could easily be misused for misinformation or obviously copyright-infringing work.
