Adobe Firefly Gets New AI Video Editing Tools, Partner Models, and Browser-Based Editor

AISoftware Update

by Nino LeitnerDecember 17^th, 2025

Subscribe to News - Software

Adobe has expanded its Firefly creative AI platform with new video editing capabilities, third-party model integrations, and a lightweight browser-based video editor designed to move beyond “prompt roulette” toward genuine creative control. The update introduces Runway’s Aleph model for precise clip editing, Topaz Astra for resolution upscaling, and Black Forest Labs’ FLUX.2 for enhanced image generation.

For filmmakers and content creators who have followed Adobe’s steady development of AI-powered tools, this update represents another step in the company’s strategy to position Firefly as a unified creative hub rather than just another text-to-video generator. The additions address two persistent criticisms of generative video tools: the lack of precise editing control and the need to jump between multiple platforms to achieve professional results. Adobe’s approach now lets users pick the right model for each task while keeping everything within a single interface.

We have covered Adobe’s Firefly video capabilities extensively here at CineD, from the initial announcement of the Firefly Video Model to its public beta launch earlier this year. The platform has evolved considerably since those early days, and this latest update continues that trajectory by focusing on workflow refinement rather than just raw generation capability.

Prompt to Edit with Runway’s Aleph model

The most significant addition for video professionals is the new Prompt to Edit feature, which integrates Runway’s Aleph model directly into Firefly. Rather than regenerating entire clips from scratch when adjustments are needed, creators can now upload existing footage and make targeted edits using natural language instructions. The workflow is straightforward: generate a video in Firefly, then refine it using specific edit commands powered by Aleph.

Runway introduced Aleph in July 2025 as an in-context video model capable of multi-task visual generation. The model can add, remove, and transform objects within clips, generate new camera angles, modify lighting conditions, apply style transfers, and handle a range of post-production tasks that would traditionally require manual compositing work. According to Runway’s documentation, Aleph analyzes uploaded footage to understand scene context before implementing changes, maintaining temporal consistency across frames.

For Adobe, this partnership represents a departure from relying solely on in-house models. By integrating Aleph, Firefly gains sophisticated video manipulation capabilities that complement its existing generation tools. The combination means creators can generate initial content with Adobe’s commercially safe Firefly Video Model, then apply more targeted edits through Aleph without leaving the platform.

New camera motion controls

Adobe has also introduced enhanced camera motion controls that give filmmakers more say in how the virtual camera moves through generated scenes. The workflow uses a reference-based approach: upload your starting frame image alongside a video clip demonstrating the camera movement you want to recreate. Firefly then applies that motion pattern to your new content.

This addresses one of the fundamental challenges with AI video generation, where camera behavior often feels random or disconnected from cinematic conventions. By using actual footage as a motion reference, creators can pull from their own shot libraries or existing production work to guide the AI’s output. It is not quite the same as manual keyframing, but it offers significantly more predictability than text prompts alone.

Topaz Astra integration for video upscaling

Resolution limitations have been a consistent pain point for AI-generated video. Most tools output at 1080p or lower, which restricts their usefulness in professional pipelines where 4K or higher delivery is standard. Adobe is addressing this through integration with Topaz Astra, the cloud-based video upscaler from Topaz Labs.

Astra uses the company’s Starlight AI models to upscale footage to 1080p or 4K while enhancing detail and clarity. Unlike simple resolution scaling, the tool offers two distinct modes: Precise mode maintains the original look while cleaning up artifacts, while Creative mode can “imagine” new detail into the footage. For AI-generated content specifically, Creative mode can help smooth out the telltale inconsistencies that often plague synthetic video.

The integration means Firefly users can generate clips at the platform’s native resolution, then push them to broadcast-ready quality without exporting to a separate application. For creators working with older or lower-quality source material, Astra can also help restore clarity and reduce noise, essentially serving double duty as both an upscaler and a restoration tool.

FLUX.2 from Black Forest Labs joins Firefly

On the image generation side, Adobe is adding FLUX.2 from Black Forest Labs to Firefly’s growing roster of partner models. FLUX.2 is the latest iteration of the FLUX model family and represents a significant leap in image quality, prompt following, and multi-reference consistency.

Within Firefly, FLUX.2 generates and edits images at up to 1 megapixel resolution with flexible aspect ratios. The model delivers what Black Forest Labs calls photorealistic detail with advanced text rendering capabilities, an area where many AI image generators still struggle. Notably, FLUX.2 can work with up to four reference images simultaneously, enabling character and style consistency across multiple generations.

The model is available in Firefly’s Text to Image module, Prompt to Edit feature, and Firefly Boards. Adobe is also bringing FLUX.2 to Photoshop desktop immediately, with Adobe Express integration planned for January. This cross-platform availability reflects Adobe’s broader strategy of making capable AI models accessible wherever creative work happens.

Firefly video editor enters public beta

Beyond new models and features, Adobe is launching a dedicated Firefly video editor in public beta. Described as a lightweight, browser-based assembly space, the editor lets creators combine AI-generated clips with personal footage and audio in a streamlined interface.

The tool is not intended to replace full-featured NLEs like Premiere Pro. Instead, it serves as a finishing environment for AI-generated content, where creators can assemble generated clips into coherent sequences without switching to heavier software. Think of it as the final step in an AI-first workflow: generate your footage, apply edits and upscaling, then cut it together in a simple timeline before export or handoff to more sophisticated editing tools.

Unlimited generations through mid-January

As a promotional push, Adobe is offering unlimited image and video generations to Firefly plan subscribers through January 15. This applies to all Firefly and partner models for image generation, plus unlimited video generations using the Firefly Video Model specifically. For creators who have been rationing their generation credits, this window offers an opportunity to experiment more freely with the platform’s expanded capabilities.

The bigger picture for professional workflows

Adobe frames this update around a key insight: creators do not want to be locked into a single AI model or forced to hop between disparate tools to complete a project. The combination of Adobe’s own commercially safe models with industry-leading partner tools like Runway, Topaz, and Black Forest Labs creates what the company calls a practical, end-to-end creative workflow.

This approach carries particular relevance for professional users who need both creative flexibility and legal clarity. Adobe has consistently emphasized that Firefly models are trained on licensed content, making generated assets safe for commercial use without the copyright ambiguity that surrounds many competing tools. By hosting partner models within the same environment, Adobe extends some of that workflow integration while letting creators access specialized capabilities when needed.

Whether this multi-model strategy will satisfy professional cinematographers remains to be seen. The tools are impressive on paper, but real-world production demands a level of precision and predictability that generative AI has not yet consistently delivered. Still, for previsualization, concept development, and certain types of content creation, Firefly’s expanding toolkit represents a genuinely useful resource.

What do you think about Adobe’s approach to integrating third-party AI models into Firefly? Are you planning to test the new Prompt to Edit feature with Runway’s Aleph model? Don’t hesitate to let us know in the comments below!