Let’s start with an overview of new tools and feature updates since Adobe MAX. As of late October 2025, Adobe has released new AI-powered capabilities across the company. creative cloudincluding generative fill and generative upscaling photoshopAI object masks and fast vector masks Premier Pro (Beta)and AI-assisted screening Lightroom (beta). New beta apps and features include: Premiere on iPhone, adobe express Featuring new AI assistants and prompts, Fireflies Image Model 5, Soundtrack Generation, and Speech Generation (Public Beta).
Released features
- photoshop: Generative composition with Harmonize, next-level generative fill, and AI-powered object selection.
- premier pro: Fast track masking and AI object masking for video editing.
- adobe express: Includes new AI assistants, prompts, and editing features.
- light room: New AI-assisted culling helps you choose the best photos faster.
- fireflies: Firefly Boards for mood boards and integration with partner models from Google, OpenAI, and Luma AI.
Beta features and apps
- premier pro: AI object masks, rectangle, ellipse, pen masking, and redesigned fast vector masks.
- adobe express: New AI assistant.
- after effects: Enhanced 3D and vector workflows, plus new audio effects like gates, compressors, and distortions.
- light room: AI-assisted culling function.
- fireflies: Image Model 5, Generate Soundtrack, and Generate Speech are in public beta.
- Illustrator on the web (beta): A web-based version of Illustrator.
Other updates
- adobe’s new creative cloud is the fastest product ever with a phone-to-desktop editing flow and enhanced Media Intelligence search capabilities.
- A new subscription tier is now available, allowing subscribers to generate an unlimited number of images.
Adobe Firefly Tools Online
Adobe has compiled all of our AI and Express tools into an easy-to-navigate online hub where you can find the tools you need at https://firefly.adobe.com/.
@jasongandy on YouTube gives you an overview of where new tools for Adobe Firefly can be found online and how to access them.
I’ll explore these tools more in future articles, but I’ve mentioned a few that I’ve played with in a few projects to see how they react. As we progress, we’ll let you know if it helps or if it’s not ready for prime time. As you know, there are other AI tools out there, but none of them do everything, and they don’t do the same thing well every time.
A big consideration is cost. Almost all AI tools have a limit on the number of renders you can get with a subscription and an offer to sell you more credits, but beware of subscription traps. All of this adds up quickly. As you might imagine, Adobe credits aren’t cheap.
Unfortunately, by the time this article is published, Adobe’s free rendering credit offer will expire on December 1, 2025.
Of course, I can’t cover everything that just dropped this week, but I’ll touch on a few that I’ve played and, as always, share the good and the bad. However, it will take some time for this to roll out, so we plan to test various tools/features both in the online (web/phone) Firefly tools and in the Adobe desktop app update to test how the tools can help video producers and content creators.
hands-on experience
As we’ve done in all of our articles, we’ll break down workflows that typically require several different tools to accomplish.
Start with an AI-generated host avatar created in Midjourney. Photoshop 2025 (beta). Starting with this base image with dramatic studio lighting, I needed to even out the model’s lighting without changing anything about her appearance.
this is the new place be in harmony This feature really sped up my AI production workflow and gave me higher quality results. First, extract the model from the original background. Please select a subject after that selection and mask Refine hair and edges. Having the model on its own layer gives you much more flexibility to make changes and recreate this effect on any type of background you choose. In this case, I chose a bright house interior (also AI-generated with Adobe Firefly) to brighten her up to match the lighting.
In the first composite image, you can see that she is no longer integrated into the background scene with her original lighting and color. She is isolated on a new layer and simply floats above the green screen background before the background is revealed.
Before selecting the Harmonize button, hide the green screen layer and expose an internal shot of the background.
As you can see, Adobe Firefly allows you to choose between three results, just like you normally do in Photoshop.
The finished result is very clean and ready to animate.
Using the original Isolation layer, select the transparency of the layer, clean up the edges of the new Harmonized layer’s mask, and float it above the green screen layer. This allows you to animate your models and redesign them into different looks using other clothing.
This is a great workflow for utilizing the same character in different scenarios and scenes, while still having some fun. adobe firefly Give her a whole new wardrobe. To start, simply send an image of your model to a green screen background and enter what you want her to wear or use. Gemini 2.5 Nana Bananathe results are the best among other apps I’ve seen so far.
It was easy to prompt and change several new looks, from business to character.
Check out my previous article AI Tools: HeyGen Avatar IV Gets Real to learn how to animate these different models and easily composite them in After Effects or Premiere. I will write more about this process in the near future. We also plan to introduce it at a free workshop we will be holding after the new year.
Character animation and sound
I needed to create a simple animation for a client’s website. Based on a character I created on my iPhone during a Midjourney meeting, I decided to animate it for an introduction and some marketing promotions.
Start with a character pose selected from in the middle of a journeyUploaded to. Generate Adobe Firefly video toolsAlong with the blank background extracted with photoshop Use the removal tool. You now have a starting frame (blank) and a final frame with your character posing in front of the camera.
I then wrote the required action into the prompt and it worked almost immediately on the first few tries. This was after an hour of failed attempts with Midjourney and other AI image conversion tools.
Good points: The detail maintained in this animation was better than any other tool I have used to date, even when using the VEO 3.1 model. The characters’ faces, eyes, and personalities are faithfully reproduced, and what’s even more impressive is the letters on their name tags.
Bad points: Currently, Adobe Firefly only produces up to 5 seconds of rendering and only 1080p. The video won’t be extended either, so you’ll have to take the last frame and rerun it to match the edit segment, or extend it. premier pro like I’ve done several times AI-generated extension tools.
Add AI sounds
The next step was to try out the new Adobe Firefly Generate sound effects tool. The great thing about this is that you can use your microphone to record the timing and impact of your sound effects. So you can record your animation as it plays and use your audio to trigger effects. It wasn’t as intuitive as you might think and I had to start over a lot. I actually think it would have been easier to record and change the mouth sounds directly in Audition, but it was fun to test this with the tool.
Finally, I generated all the components to mixdown everything together in Premiere. This project seriously could have been completed in an hour or so if I had been able to streamline my workflow and repeat the steps, but because the extensions added in Premiere didn’t take advantage of VEO 3.1 Nana Banana, the nametags were broken in the render, so I had to go back and forth between After Effects and Photoshop to motion-track a nice nametag on top of the rendered one to make it easier to read.
The final result of this process is:
So, while we still have a long way to go before AI-generated content becomes a reality, seeing all this integration and vastly improved results raises expectations for better tools and cleanly edited content generation.
AI Masking in Premiere Pro
My friends at Adobe know that I’ve been complaining about the Roto Brush tool in After Effects for over a decade. It’s fine for quick and dirty high-contrast selections, but in real production it was still very unreliable and took a lot of work to set up, reselect, etc.
So I ran a very quick test using a clip provided by Adobe to create a simple mask layer over text using Premiere’s new AI Object Mask tool. This took about 10 minutes to complete from start to finish.
I imported some test footage, duplicated the layers, and sandwiched a simple text layer between them. The new object selection tool allows you to draw around the area you want to mask and track.
Once you have selected your objects, track the scene so the tool can detect all edges throughout the shot.
In this shot, a tree branch enters the shot halfway through. I created a second mask on the same top layer and tracked it as well.
While I’m impressed with the speed and overall accuracy of quick masking this tool provides, it’s still not as sophisticated as just chalking mattes and feathering. In a real-world scenario, trying to get a clean composite would require you to bounce the mask and paint in areas it won’t hold, but for quick and dirty marketing videos or social titles, this will probably cover most people’s needs. It’s just not ready for the quality of broadcast or streaming capabilities yet.
Here are the results of a video test with both the original clip slowed down and sped up to normal speed.
We look forward to providing you with more details soon.
