跳到正文
postPerspective· Randi Altman·· 2 天前AI 评分20

Adobe MAX 2026 迈阿密大会综述:Firefly 与生成式 AI 工具成焦点

Adobe MAX 2026 Part 2: Conference Overview

AI 导读

Adobe MAX 2026 在迈阿密举办,超过 10,000 人到场,CEO 交接与 Adobe 达到 10 亿用户成为开场重点。

正文 · 原文

By Mike McCarthy

I thankfully dodged the airline ground stop last week and made it home from Adobe MAX, which returned to Miami this year after also being held there in 2024. The conference is most often held at the LA Convention Center, but Miami is more convenient for East Coast pros, as well as attendees from Europe, Central America and South America. The location might also attract more first-time attendees, which could be a plus for Adobe.

There were over 10,000 people at the conference, and most of them were in the room for the initial keynote presentation that kicked off the event. It started off with outgoing CEO Shantanu Narayen giving a talk and then introducing the audience to his replacement, Anil Chakravarthy. This was a previously announced transition, similar to Tim Cook’s transition at Apple, which I witnessed at WWDC earlier this year. They also highlighted Adobe recently reaching 1 billion users, which is over 10% of the planet. I assume that number is not just subscribers but users of free services as well, like Express and other mobile apps, as well as PDF applications and other online services.

What followed was the usual parade of speakers, highlighting all of the newest developments in Adobe’s Creative Cloud applications and now the Firefly platform. I detailed many of those new features and products in Part 1 of my MAX coverage, but agentic AI (or Adobe’s AI Assistant) came up throughout that presentation. Using Adobe’s AI assistant will consume Firefly credits. However, once applications support MCP, users will also be able to control them through other AI tools, including locally run agentic models, offering an alternative for those who prefer not to use a pay-as-you-go model. The entire keynote is available to watch here if you want the full experience.

After the keynote came the Creative Park, where Adobe and its partners offered hands-on and interactive demonstrations of new technology, while smaller sessions and lectures were held throughout the convention center.

The first session I checked out was from AJ Bleyer, who I first discovered the last time MAX was held in Miami. His presentation was about filmmaking and highlighting a film he had created in Miami the night before. Most presentation content is prepared weeks or months in advance, but Bleyer was able to create a polished product with AI effects using footage from the night before. I was impressed enough to make sure I always fit his presentations into my schedule. He talked about how to use AI for previz and location scouting and showed how to use brush-for-prompt to edit and steer AI generation in Photoshop, which I had not seen before. He advocated for using AI to demonstrate why something costs a certain price. (“If you want it to look like this, then that is what it costs.”) He also uses AI to demonstrate an image in context, like on a billboard, or on a screen with someone watching it, to show clients the actual end result they are looking for. (Check out postPerspective’s podcast with Bleyer as a guest.)

I also attended a photography-focused session with Travis Keyes. He discussed working under time constraints and making the most of the limitations you encounter. He also emphasized the connection between the artist and the subject and the stories each brings to the experience.

Adobe’s Anton Knoblach then gave me a demo on the new generative AI tools. Since I am already familiar with most of those options, we did a deep dive into the brand-new Generative Remove tool. Under the hood, Adobe expands the user’s mask slightly and then generates new pixels with a customized Firefly model to cover the object in question. It then creates a new video asset with alpha matte and composites that on top of the original clip, preserving the rest of the original pixels. That composite appears as a single clip in the timeline, but the video of just the patch can be found in the Premiere generated assets folder, proving that the rest of the pixels will be untouched.

In addition to the panels, I like to explore Creative Park’s booths, which included various Adobe teams and their industry partners, like Nvidia, Dell, HP and others. I designed a custom character via AI in the Nvidia booth, and Nvidia had it (and a crowd of others) walking around a digital billboard. Nvidia also kindly made a metal print of it for me to take home. This was to show off the local AI functionality of the recently released RTX Spark systems. There was a Disney Imagineering booth, where people could try out the customized Firefly-based tools that Disney uses to design ride features. And in the Logitech booth, there was an appliance-sized model of the new “Creator Edition” of its MX Master 4 mouse.

On the evening of the first day, Adobe hosted an event at the Fillmore Theater, where it showed three new short films. After the screening, there was a discussion with the films’ directors. The first short was shot on real film, which surprised me. The second felt more like a trailer for a feature film than a stand-alone story. That short has since led to the longer film, which is now in production.

The third film was the most complete story, highlighting the futility of living for affirmation on social media, which, given the audience, was interesting. The conference itself draws more influencers and social media gurus every year, which is a shift from the corporate designers and advertising creatives that I believe comprised the main audience a decade ago. I think the inward focus of that population was on full display at the conference, with all of the selfie sticks and people recording themselves instead of what was around them. Many people’s primary goal of attending the conference appeared to be to learn new ways to grow follower count, which doesn’t seem like a particularly healthy focus. It definitely causes me to more closely evaluate how I want to raise my 10-year-old son in this age of social media. It is important to be able to communicate in that medium and to be a good storyteller, but you have to be able to engage in that space without letting it define your identity. I witnessed the full spectrum of those possible results when looking around the show.

Day 2
The second day of the show always starts off with the Creativity Keynote. It features no new products or technologies; it’s just a series of presentations from — and conversations with —successful artists, from movie directors to designers. This year’s presentation was hosted by Adobe’s Jason Levine, and the first guest was Drex Lee, who is known for his epic one-shot short videos. He shoots them on his phone, with no support gear aside from a selfie stick and his body. (His dexterity and skill replace the traditional production crew required for intricately choreographed shots.)

Next up was Siobhán Maroney, who shared the winding path of her career into creative design. Her key takeaways were to connect with other people and always strive to deliver more than they expect. Levine then interviewed filmmaker Omar Kamara after showing a clip from his most recent short. Kamara talked about the importance of community and the support that you can draw from the voices around you. The event concluded with a conversation between Adobe’s Stacy Martinet and fashion designer Jonathan Anderson. Fashion is pretty far removed from my world, so the closest connection I could make to that conversation was about how effort and perseverance pay off.

After that keynote event, I had a conversation with Adobe’s Jason Druss about Premiere’s new Color Mode and related tools. Color Mode is extremely powerful, both in terms of how it can affect the visual media and how those changes can be managed across a long series of clips. But there will be a learning curve to using the new tools effectively, and older workflows and paradigms will require adjustment. And not all of the tools that editors are going to be looking for are there yet. I predict that it will be a rough transition for many users, but the eventual result should be a vastly superior process once they can fully embrace it. But Adobe is going to need to re-introduce simpler options for people who need to make basic adjustments and aren’t ready to dive into the full toolset.

Adobe had set up its own collection of “Creator Cards” (like baseball cards) for its users, with people making their own custom cards and getting cards of Adobe featured artists. I am not sure how well these cards are going to catch on, but I have a framed one— or, technically, my son does now, and he is super-excited about it. So there is that.

After that I was able to fit in a session about compositing in Premiere from John Knowles.  It was a good presentation and packed in a lot of good ideas about how to fully use Premiere’s capabilities. The new ideas I took away from the talk were 1) using transitions on adjustment layers (for multiple simultaneous transitions, or more likely, in my case, transitioning multiple composited layers), 2) adding effects directly into the vector graphics stack (to apply to objects below but not above it in the stack), and 3) the Logo Cutout effect (for removing black or white backgrounds), which I had probably heard of but forgotten about. All of those are potentially useful tools when doing projects within Premiere, and the little things like this are what allow people to use the software more effectively and become an expert at a tool.

Following that session came MAX Sneaks, where Adobe showed off the new technologies it is working on that aren’t ready to be included in its software yet. The event featured Terry Crews for comic relief, along with a number of Adobe engineers sharing their recent work. Obviously, AI played a big role here, but it was mostly about generative AI instead of agentic AI. The developments in that regard are about giving users more control over the generative AI process, whether that’s creating personalities for generated characters with Project Screen Test, or controlling the timing of events in a generated video with Project On Cue. Project Scene Sweep was the most powerful. Engineers demonstrated changing a character’s hairstyle and outfit throughout a series of shots while keeping the new look consistent. It is quite mind-blowing. It will probably take a while before the technology can be generalized enough for more people to use it, but it is interesting to see what is possible.

The immediate application of that technology could be to maintain consistency and fix mistakes in continuity. Was one character missing a piece of wardrobe in one take? Project Scene Sweep  can change what that character was wearing to match the other shots. Scenarios like that are where I see this technology being used first, as opposed to totally restyling an entire cast throughout a show. Why would anyone need to do that? But turning the entire cast into aliens? That would be an interesting application.

Surprisingly, the project that got the strongest reaction from the audience was a simpler one, Project Deep Brush. It allows you to paint on an image in context. The shape of the brush conforms to the shape and texture of the objects in the picture, and you can paint on surfaces without affecting the objects in the foreground of those surfaces. It probably got such a strong reception because people could imagine using it immediately without a learning curve. It will just be an intelligent enhancement of the brush tool in Photoshop, and it is already available for users to try on Adobe Labs.

Adobe capped off Wednesday with the MAX Bash, which was held in Miami’s LoanDepot Park, the stadium for the Marlins baseball team. A baseball stadium seems like a strange place for a party, and it was. The food was served in the shops on the upper concourse, and there was a dance party down on the field itself. It seemed like there were fewer things to explore and experience than in past Bash parties, both in terms of unique foods to try and also creative endeavors to observe or take part in. There were a couple artists creating murals, some orbiting selfie stick stations, some local tricked-out cars… but not the wide variety of creative and artistic displays I have seen before.

Thursday is usually quieter, as attendees are recovering from Bash or have already departed. It was a good time to visit booths that had been too crowded before and talk with Adobe representatives about upcoming features. I attended a session about Kling’s generative AI models, where they talked about the recently released Kling 4.0 video model and announced an upcoming Kling 4.0 image model that can create pictures in up to 8K resolution.

And then I was done and headed back home. On my way out, I snapped a picture of this community wall, which had been available to write on all week as a bit of a communal creativity project.


Mike McCarthy is an online editor/workflow consultant with over 15 years of experience on feature films and commercials. He has been working on new solutions for tapeless workflows, DSLR filmmaking and multi-screen and surround video experiences. Check out his site.

Follow by Email

Facebook

fb-share-icon

Twitter

Follow Me

Tweet

YouTube

LinkedIn

Share

Instagram

来源:postPerspective · postperspective.com