Adobe launches its AI video model Firefly Video Model in public beta in Premiere Pro, including letting users extend footage and generate video from prompts
it supports text-to-video, image-to-video and designed for immaculate prompt coherence. Those sample generations look quite impressive — excited to go hands on! I'm at Adobe MAX this week so [video] Ed Ludlow / @edludlow : ICYMI + LONG POST: Adobe came out with its Generative AI video tools today (Firefly in Premier). The technology can create and modify video. For example, you can take an original video clip and extend it using generative AI *which is kinda cool* Right now there's no usage Bilawal Sidhu / @bilawalsidhu : Extremely entertaining Illustrator demo. Only at Adobe MAX will the whole room of 10,000 attendees lose their minds over the simplest quality of life updates. Gotta love it 🥰 [video] Bilawal Sidhu / @bilawalsidhu : 2) Auto masking directly inside premiere. Basically akin to Segment Anything Model 2 right inside premiere. This one alas is “coming soon” but one I'm most excited about because I won't need to move into After Effects for simple compositing tasks like this. [video] Bilawal Sidhu / @bilawalsidhu : Woot! Two REALLY cool features coming to Adobe Premiere. 1) Generative Extend available today in beta. Take any clip and extend it — with the video and audio automagically extended :) [video] Andrew Curran / @andrewcurran_ : Adobe's video model is launching today. The clip is from the beta. There are multiple other video models also approaching launch: - Grok's image/video model. Elon said Flux is a temporary solution to Grok's image generation until he finishes training his own model. The image [video] @adoberesearch : Create life in your scenes with a single click with Project Generative Physics! In this @AdobeResearch sneak, realistic physics are a snap with a simple text prompt that can generate subjects which interact with their environment. https://www.linkedin.com/... [video] @urinieto : Generative Extend is out! So proud to be part of this incredible effort with such big titans like @justin_salamon and @premseeth. Actual audio examples in the thread. Hit me up if you're a PhD student looking for internships for 2025! :) Bobby Miller / @bobbymiller : V cool. Adobe has gone from disdain for artists to straight up eliminating jobs for them. Alexandru Costin / @acostin : @Adobe #AdobeMAX premiere pro GenExtend is going to save so much time when editing video! #AdobeFirefly [video] William Lamkin / @williamlamkin : New generative AI video editor tools in @Adobe Premiere Pro Including: • Generative Extend • Object Addition/Removal • Text-based Editing • Audio Category Tagging • Speech to Text • Enhance Speech • Scene Edit Detection • Auto Color • Color Match • Morph Cut • Auto @nvidiastudio : 🎥🎨 Announced at #AdobeMAX: New RTX-accelerated Adobe AI app updates speed up content creation in Adobe Premiere Pro, After Effects, and Substance 3D. From Generative Extend in #PremierePro to the new @Substance3D Viewer app, your workflow just got faster. 💻 Learn more 👉 [image] Pete Pachal / @petepachal : Interesting: The new Adobe Premiere Pro will directly use the NPU in Microsoft's Copilot+ PCs for its generative features, (like Generative Extend), giving a performance boost #AdobeMAX [image] @zuryart : This feature to automatically track and object select in a video is HUGE!! What a time saver! Thank you @Adobe!! #ADOBEMAX [video] @joematoske : Increasingly, Adobe becoming an anti-creative company- Continually sounding more and more like their audience are blockchain enthusiasts. As a long time premiere editor, hate to see it. Justine Moore / @venturetwins : Got access to @Adobe's new generative expand for video ✨ You can add up to two seconds to the end of a clip (watch for the black bar). I tested it on a few styles, and it's good at maintaining the context and natural motion of a scene - watch to the end for a blooper 😂 [video] Max Einhorn / @maxeinhorn : Commercially safe! All other video models, take note. Licensing your training data properly is going to be an expectation going forward. The fact that Adobe is doing it as a market incumbent is going to sway how this “fair use” question shakes out. If they can do it, why won't you? Justin Salamon / @justin_salamon : Adobe just announced Generative Extend for Premiere Pro (beta) at #AdobeMAX! Use GenAI to extend your video clip *including the audio* @pseetharaman @urinieto and me in the Sound Design AI Group at @AdobeResearch worked on the audio part and we're so excited to see it go out! [video] LinkedIn: Adolfo Hernandez S. : 🎉 Now in Premiere Pro (beta), you can click and drag to seamlessly add a few more frames of video or audio, including room tone and sound effects, with Generative Extend! …
Context & Ripple Effects
Adobe has been moving Firefly from prompt-based enhancements toward video production: its earlier natural-language editing upgrades targeted effects and color work, while a September update set expectations for Firefly Video features in the Premiere Pro beta.
This release makes that planned capability usable inside Adobe’s flagship editing workflow, rather than leaving video generation as a separate demonstration or web experience.
First-order effects
- Premiere Pro beta users can generate video from text or images and use Generative Extend to lengthen clips with matching room tone and effects, adding synthetic material directly to an edit.
- Adobe gains a public testing channel for Firefly Video’s prompt coherence and editing fit; the beta also exposes its broader AI-assisted editing suite to working video workflows.
Second-order effects
- Editors can treat missing handles, short transitions, and certain insert shots as generation tasks within Premiere, potentially reducing the need to leave the timeline for narrowly scoped fixes.
- Premiere becomes a more consequential distribution point for video models, building on Adobe’s earlier exploration of letting users access third-party video models in Premiere; model providers will need to compete on how well their output fits production workflows, not only on standalone demos.
Third-order effects
- If these tools prove reliable in production, video editing software is likely to evolve from a tool for arranging captured footage into a system that both manages and creates media assets.
- The durable competitive advantage may shift toward workflow-native controls—asset handling, editability, and consistency—rather than text-to-video generation alone; public-beta use will test that proposition.
The trend: Generative video is moving from standalone creation interfaces into the professional editing timeline, where usefulness depends on controllable, workflow-compatible output.