Adobe Firefly: New Tools for Video and Image Editing, and Support for Custom Models

Adobe continues to develop its Firefly AI platform, and the March update was quite packed. Several areas received significant improvements: video editing, image processing, and the ability to train custom models for a specific visual style. Let's dive into what exactly has changed and why it's useful.

Your Own Model – Your Own Style

One of the key innovations is support for custom models. In short, you can now 'teach' Firefly to work in a specific visual style, such as a brand's corporate aesthetic or an artist's signature style.

Previously, AI generation produced results in a generalized style that then required extensive manual tweaking. Now, the model can be configured in advance, and it will produce results close to the desired sample from the very beginning. This is especially useful for designers working with brands that have clear visual standards or for creative teams wanting to maintain a consistent style from project to project.

Simply put, you no longer need to explain to the tool “what it should look like” every time – you just need to configure the model for yourself once.

Video Editing Made Easier

Firefly has gained enhanced capabilities for working with video. Here, Adobe is aiming to shorten the path from raw footage to the final result – without extra iterations and manual adjustments for every frame.

The specific tools are aimed at giving more control over how the video sequence looks and unfolds. This is important because video is one of the most labor-intensive formats in visual production: even a small edit can require reassembling an entire scene. AI tools here take on the routine parts, leaving the creative decisions to the user.

Images: More Precision, Fewer Iterations

The image editing tools have also been updated. The focus is on greater precision when making edits and on more predictable results. This means that targeted edits can be made without the fear that the AI will “reimagine” the entire image.

This logic is familiar to anyone who has worked with generative tools: sometimes you ask to change the background, and the model also changes the lighting, pose, and mood of the entire scene. Firefly's new features are specifically aimed at keeping control in the user's hands.

The Core Idea: From Concept to Final Product – Faster

Looking at the update as a whole, Adobe is consistently moving toward a single idea: fewer technical barriers between concept and result. The user should be thinking about what they want to create, not about how to force the tool to do it.

Custom models eliminate the need to define the style from scratch every time. Improved editing reduces the number of iterations. The video tools speed up work with this complex format.

This isn't a one-time leap, but a systematic progression, and the March update fits organically into this logic.

Who Will Benefit Most?

The Firefly updates are primarily aimed at those who work with visual content regularly and in large volumes: designers, video producers, and brand content teams. But for those who use AI tools from time to time, the lower barrier to entry is also a tangible benefit.

Custom models will likely be most interesting to professionals and teams who value stylistic consistency. For individual users, it's also a viable option, especially if they have an established personal style they want to reproduce consistently.

The video and image tools are useful to a wide audience, from marketers to independent creators who want to achieve high-quality results without a deep dive into the technical details.

What Remains to Be Seen

It's still difficult to assess how flexible custom models will be in practice: how easy they are to set up, how many examples are needed, and how consistently they reproduce the desired style across different scenarios. Real-world use will tell.

AI-powered video editing is still an area where tools are evolving rapidly, but user expectations are growing just as fast. We'll see how well Firefly's new features meet current demands.

Overall, the March update looks like a meaningful step forward – not a flashy announcement, but a practical expansion of what already works.