Adobe teaches Premiere to erase objects from video, while Firefly learns to build workflow graphs

Anyone who has edited video knows this pain: a boom microphone dips into the frame, someone else's sign sticks out in the background, and a reshoot is no longer an option. In the past, that meant manual rotoscoping and many hours in After Effects. At the Max conference, Adobe showed that it wants to turn this kind of work into a single click, and across three products at once.

Removing the unwanted: from still frame to video

The headline addition is Generative Remove in Premiere, currently in beta. It works like this: you mask an object, for example a microphone above the subject's head, Premiere tracks it through the clip, you hit Remove, and a generative model "fills the area with realistic detail." According to Adobe, the rest of the frame stays unchanged. The company showed a similar feature called Project Frame Forward at its Sneaks showcase of experimental projects last year. Now it has reached the timeline.

In Photoshop, the Remove tool has been updated. It has become better at detecting and reconstructing things associated with the object being removed: reflections, shadows, footprints in sand. This matters because it is usually the shadow that gives away that an object was erased. Photoshop also gets Quick Compare, a before/after split screen with no need to toggle layers, and Relight, where light sources are placed directly on the canvas with adjustable tint and brightness.

In Lightroom, one-tap skin retouching moves from the mobile version to desktop. In addition, Open Eyes arrives in public beta: it replaces closed eyes with open ones, using another photo of the same person from your library as a reference. For Premiere, Adobe also added audio separation into dialogue, music, ambience and effects, while Generative Media and Color Mode are now available to everyone.

Firefly tries to become a working environment

The second line of announcements matters more to product managers and founders. Quick Actions, ready-made workflows such as video thumbnails and product ad banners, have left beta. Firefly Design has launched in beta: a single space for illustration, image editing, animation, 3D and print. Alongside it comes Firefly Graph, where you can build and repeatedly run your own chains of operations. The Verge notes that it resembles Weave, Figma's platform from last year.

Firefly Builder, a tool for prototyping websites and apps, is in closed beta. In the publication's assessment, it is similar to Figma Make. It is clear that Adobe is ceasing to be a set of separate editors and is moving toward nodes and workflows in which AI steps connect like a conveyor belt.

What to do with this

If you make content, think about which routine operations are worth moving to the new tools. Removing a microphone or clutter from the background is a typical candidate. But Generative Remove is still in beta, and Adobe has not published quality metrics. Check the results on motion, on complex backgrounds and in close-ups, where artifacts are most noticeable.

For Open Eyes and skin retouching, decide in advance what counts as an acceptable edit. Swapping eyes based on a reference shot is no longer cleanup but a new photograph. Teams that work with clients or in news need rules for disclosing such edits.

For product managers, the takeaway is different. Adobe is copying patterns that Figma launched a year ago: a workflow graph editor and prototype generation. This signals that "repeatable AI pipelines inside a design tool" are becoming the genre standard. If you are building a niche AI editor, your competitor now is not just a startup but also a platform that is already installed on your customer's desktop.

Finally, a caveat: this is only part of the Max announcements. Adobe has posted the full list on its blog, and many features are still in beta or closed access, so it is too early to plan production around them.