Adobe launches subscriptions for Firefly AI
Adobe is hoping to capitalize on the early success of its Firefly AI models by launching a new standalone subscription service that gives users access to the company's AI image, vector and video generating models.
This marks Adobe's boldest attempt yet to turn its Firefly AI models into a real product.
The company is also launching a redesigned webpage, firefly.adobe.com, where people can use Adobe's AI models. This includes the new Firefly AI video model, which is rolling out in public beta on the Firefly website and in the Premiere Pro Beta app.
Firefly's Standard plan costs $9.99 per month and provides unlimited access to Adobe's AI image and vector generating features, as well as Adobe's new AI video model. The Standard plan gives users 2,000 credits, which is enough to make 20 five-second AI videos.
Users can also connect Firefly plans to their Creative Cloud accounts to get unlimited AI image and vector generation in Photoshop, Express or other Adobe apps.
Meanwhile, the Pro plan will run users $29.99 a month, and offers enough credits to generate 70 five-second AI videos per month. The company is also working on a "Premium" tier (it hasn't announced pricing for this yet) that lets users create 500 AI videos per month, according to Adobe's VP of Generative AI, Alexandru Costin.
Previously, Adobe offered many of Firefly's AI tools within its existing Creative Cloud subscriptions, letting users try the new tools for no added cost. Users could upgrade to pricier plans if they wanted more access to Firefly, but they didn't have to. That system worked well for Adobe: Firefly's generative fill feature, added to Photoshop in 2023, has become one of the company's most popular new features of the last decade.
Now, Adobe wants to see if users will also pay up for its Firefly AI models.
The Firefly video model lets you turn text or images into a five-second, AI-generated video. There are controls on a side panel for changing the camera angles, camera movement, aspect ratio, and other features that creative professionals might want to customize.
The new Firefly offerings will compete directly with OpenAI's Sora, Runway's Gen-3 Alpha, and other AI video models that already have dedicated webpages and subscription plans. Google DeepMind's AI video model, Veo, seems to be a legitimate contender in the space as well, but it's still in private beta.
Part of Adobe's pitch to creative professionals is that Firefly was trained on a dataset of licensed videos, without any brand logos or NSFW content (something the company paid quite a bit to do). That means, according to Adobe, creatives should be able to use the Firefly AI models without worrying about legal troubles.
"We think the key differentiator for us is that we're the only IP-friendly, commercially-safe video model," Costin said in an interview with TechCrunch. "We want to differentiate with deep understanding of customer problems."
Adobe has also tried to ship AI tools that solve problems for creative professionals instead of just generating random AI videos.
For example, one of Firefly's AI video features, Generative Extend, lets users extend any clip's video and background noise by a few seconds. This is one of the more practical AI video tools on the market; other AI models just let you create new videos from scratch, or animate photos.
Costin says Adobe is working on another AI video tool to help with pre-production. The tool, which has yet to be announced, would help get creatives aligned on the same vision by creating a rough sketch of what a scene, or string of scenes, would look like.
However, Adobe needs to walk a fine line with generative AI. Many professionals who have used Adobe's apps for decades are upset about the rise of generative AI tools in their industries. The technology poses a threat to their livelihoods as they risk having their work automated away to an AI model — like the ones Adobe is building.
But Adobe is convinced this is where the puck is going in the creative world.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
an hour ago
- Yahoo
OpenAI Careens Toward Messy Divorce From Microsoft
Quick, someone call TMZ: There's an even messier celebrity breakup than Trump-Musk with countless billions at stake. Relations between OpenAI and its largest investor, Microsoft, are continuing to fray as the ascendant artificial intelligence firm struggles to get approval from its investor on the fine points of reorganizing into a for-profit public benefit corporation. According to a report in The Wall Street Journal last week, OpenAI is now even considering a 'nuclear option' to sever ties for good. READ ALSO: Drones Steal the Paris Air Show and Berkshire Slips Amid Concern Retiring CEO May Take 'Buffett Premium' With Him At the heart of the matter is how large a stake Microsoft will own in OpenAI's public benefit corporation, a subsidiary of the ChatGPT maker that will still be controlled by the nonprofit parent. According to a recent Reuters report, OpenAI wants the Big Tech player to hold a 33% stake while relinquishing its rights to future profits. Microsoft hasn't agreed, and the two sides are at loggerheads over the matter, though it's far from their first fight. Microsoft already loosened its grip on the AI firm in January, allowing some key terms of their agreement to change so that OpenAI could tap data centers outside of the Microsoft Azure infrastructure. That resulted in 'Stargate,' a high-profile $500 million data center joint venture between OpenAI, Softbank, and Oracle (with Microsoft, Nvidia, and Arm serving as 'technology partners'). But opening the door to more independence may be objectionable to Microsoft: Earlier this month, Reuters reported OpenAI is now planning to tap Google's cloud services to meet its growing need for computing capacity. In another line-crossing move, The Information reported last week that OpenAI has been offering a suite of ChatGPT enterprise tools at discounts of up to 20%, directly undercutting sales of competing Microsoft services like Copilot. In other words, tensions are at an all-time high, and now both sides are throwing around fighting words at a time when level-headed communication is crucial: According to the WSJ, OpenAI executives have discussed a 'nuclear option' of formally accusing Microsoft of antitrust violations if it can't come to an agreement with the Windows-maker over transition terms. According to a Financial Times report published Wednesday, Microsoft is prepared to walk away from negotiations altogether and simply ride out its existing commercial contract with OpenAI, which is set to last until 2030. That would leave OpenAI stuck with its current structure, which means it'd lose out on half of the $40 billion investment SoftBank committed to making in April, which was contingent on a successful restructuring to for-profit by the end of the year. Burn Book: Microsoft isn't the only Big Tech firm OpenAI has waded into a blood feud with. The company now finds itself openly at war with Meta, which is offering $100 million signing bonuses to poach OpenAI talent as it seeks to bolster its AI efforts. OpenAI founder Sam Altman last week jabbed back by saying, 'I don't think that [Meta's] great at innovation.' At this rate, we have to imagine Altman isn't exactly the most popular player at our imagined weekly poker night among Silicon Valley bigwigs (actually, let's be honest, they're probably playing Magic: The Gathering). This post first appeared on The Daily Upside. To receive delivering razor sharp analysis and perspective on all things finance, economics, and markets, subscribe to our free The Daily Upside newsletter. Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
WIRED
2 hours ago
- WIRED
Gear News This Week: Adobe Wants to Make iPhone Photos Better, and TCL Brings Flexibility to Atmos
Plus: JLab's latest Bluetooth speakers start at $20, Wyze tries to make amends with new security upgrades, and more. Courtesy of Philips; JLab All products featured on WIRED are independently selected by our editors. However, we may receive compensation from retailers and/or from purchases of products through these links. The brains behind the computational photography that vaulted Google's Pixel phone camera to fame have a new camera app for you to try, this time from Adobe. Marc Levoy and Florian Kainz left Google several years ago, and Adobe's Project Indigo seems to be the fruit of their labor. The iPhone-only app is available on the App Store for the iPhone 12 Pro and Pro Max, iPhone 13 Pro and Pro Max, plus iPhone 14 and newer. (It's free and doesn't require an Adobe account.) Like the computational photography techniques pioneered with the Pixel's camera, the Project Indigo camera app captures a burst of photos and combines them to deliver better dynamic range with low noise. The research paper written by Kainz and Levoy claims it produces a more natural 'SLR-like' look, closer to that of a professional camera than a smartphone. It does this by underexposing the image while combining up to 32 frames, much more than most phones. This supposedly creates a delay after pressing the shutter button, but that's the sacrifice to get better image quality. The app has manual controls, including shutter speed and ISO, and can capture photos in RAW and JPEG. Adobe's products are already the default for many creatives in post-production workflows, but if the company can craft a camera app that can produce better results than the native camera in smartphones, that could put one's entire workflow through Adobe's ecosystem. The app is still experimental, and the team is hoping to add more features, like Portrait mode, an Android app, and video recording. Wyze Tightens Security With VerifiedView We were big fans of Wyze's budget security camera wares, but those low, low prices came at a cost. We stopped testing and recommending Wyze cameras in early 2024 after a series of security incidents. (The final straw was the exposure of 13,000 camera feeds to customers who didn't own them.) The company has been working hard to beef up its security since then and win back customer trust. The latest announcement to that end is VerifiedView, which tags all your videos with your unique user ID. Before anyone can view, download, or share that content, Wyze verifies that the user ID on the content matches the account that's trying to access it. In short, it's a fail-safe to prevent anyone from being able to watch videos from other people's cameras, and it should make a recurrence of that last breach impossible. Over the last few months, Wyze has pulled in security consultants for extensive penetration testing, improved password requirements, enabled two-factor authentication by default, improved cloud security, beefed up encryption, and rolled out tools to detect suspicious logins. Wyze cofounder Dave Crosby told WIRED that Wyze has also reduced its reliance on third-party tools, which were partly to blame for that last breach. The company now has a bug bounty program and a transparent reporting policy, plus Crosby says every employee has completed cybersecurity training. It's a shame it took a serious breach for this to happen, but these actions are encouraging, and the company has managed to stay incident-free for almost 18 months now. We plan to resume testing Wyze devices soon. — Simon Hill Peak Design Finally Refreshes Its Travel Tripod I've been regularly using Peak Design's Travel Tripod for 5 years, so I'm excited to see the company finally debuting a successor: the Pro Tripod. It's now on Kickstarter with an expected October ship date. The Pro Tripod comes in various versions: Pro Lite ($800), Pro ($900), and Pro Tall ($1,000). They all can hold 40 pounds (as opposed to 20), are taller, and remain travel-friendly and portable. It was difficult to shoot video with the original Travel Tripod, but that's remedied with the Pro Tripod's fluid panning capability via the redesigned Pro Ball Head. There's a separate Tilt Mod you can add for full pan and tilt functionality. The differences between the new tripods aren't vast. The Pro Lite shaves some weight (it's 3.7 pounds) but doesn't get as stiff; the Pro Tall has a higher deploy height and can be stiffer, but is heavier and larger (4.5 pounds). The Pro blends a bit of both (4.2 pounds). The tripods have carbon fiber legs with a CNC-machined anodized aluminum center hub. They look like great upgrades all-around, only if you can stomach the leap in price. JLab's New Bluetooth Speakers Are Crazy Cheap JLab is taking its talents for crafting quality earbuds at shockingly low prices into the portable speaker market with four new budget Bluetooth models. The lineup starts with the $20 Pop Party Speaker, a hangable oval with dual two-inch drivers, dual passive radiators, customizable RGB lighting via the JLab app, and around eight hours of battery life. Stepping up to the $30 Go Party Speaker gets you slightly improved water resistance (IPX6 vs IPX5) and twice the battery life in a tubular design reminiscent of JBL's Flip speakers. The larger JBuds Party ($70) offers 30 watts of power to make it 'one of the most powerful speakers at its price,' according to JLab, though, unlike most speakers at this level, it's not fully dunkable, offering just IPX6 water resistance. Finally, because every brand needs a karaoke speaker, there's the $150 Epic Party with a 360-degree soundstage, four 2.5-inch drivers, a 5.25-inch woofer, and up to 16 hours of claimed battery life. Like its siblings, the Epic boasts RGB lighting and includes both a 3.5-mm aux connection and a quarter-inch jack for adding a microphone. All four speakers are available this week, and we'll be checking them out soon to see how they stack up to the best Bluetooth speakers we've tested. — Ryan Waniata TCL Z100 Speaker Makes Atmos Sound More Flexible More than 18 months after it was first teased at IFA 2023, TCL has confirmed the launch of the Z100—the world's first Dolby Atmos FlexConnect speaker. The aim here is to deliver Atmos sound wirelessly and flexibly, with a range of configurations available—from two to four speakers. As they are wireless, the Z100 speakers can be placed wherever is convenient in the room, and will work along with the sound from compatible TCL TVs as well as any other Z100s you own, creating a 5.1.2-channel or 7.1.4-channel Atmos sound. Each speaker packs four drivers, including one upwards firing, to deliver 170 watts of total power. The Z100 is currently compatible with the latest generation of TCL QD-Mini LED TVs, like the QM6K, and will also double up as a Bluetooth speaker (without Atmos). It'll be available first in France starting July 2025, with further European and US launches to follow. — Verity Burns Philips Hue Debuts the Wall Washer The latest smart lighting release from Philips Hue is the Hue Play Wall Washer. This sleek, black, aluminum device is just 6 inches tall and is designed to sit on your TV cabinet or sideboard and, as the name suggests, wash your wall in colorful light. Employing a new ColorCast projection technology, the Wall Washer sprays light at a wide angle and supports multiple simultaneous colors and a ton of lighting effects. You can sync it with your existing Hue lights and systems, including the Hue HDMI Sync box and the TV and PC Sync apps. Folks can set the intensity, speed, brightness, 3D positioning, and direction of the light it emits in the Hue app. Available this month, a single wall washer costs a rather spicy $220, or you can snag a two-pack for $385. Select markets already have access to the built-in AI assistant in the Hue app, which you can ask to generate scenes based on your mood or activity, but it's rolling out to the UK in July and globally (including the US) in August. — Simon Hill
Yahoo
3 hours ago
- Yahoo
Why OpenAI engineers are turning down $100 million from Meta, according to Sam Altman
When you buy through links on our articles, Future and its syndication partners may earn a commission. OpenAI CEO Sam Altman says competitors, particularly Mark Zuckerberg's Meta, have been trying to poach OpenAI engineers with sky-high compensation packages. 'They started making these, like, giant offers to people on our team. You know, like $100 million signing bonuses and more than that in compensation per year,' Altman said this week on the Uncapped podcast, hosted by his brother, Jack Altman. Altman said he was glad to see that those enticing offers haven't worked on OpenAI's best people. He assumes this is because they looked at the two paths, Meta and OpenAI, and concluded that the latter has a better shot at delivering on superintelligence and will eventually become the more valuable company. Amid the digs, Altman said Meta is missing the one thing that truly matters in AI: a culture of real innovation. 'There are many things I respect about Meta as a company, but I don't think they're great at innovation,' said Altman, when discussing Meta's attempts to lure OpenAI engineers. He explained that by trying to recruit OpenAI staff with massive guaranteed compensation packages, Meta is essentially building a culture that prioritizes money over the work and mission. He believes that focusing on money rather than purpose and product is a recipe for the wrong kind of culture. Altman contrasted this with OpenAI's approach, which he said attracts and retains talent by aligning financial incentives with a shared sense of purpose and innovative work. 'The special thing about OpenAI is we've managed to build a culture that is good at innovation, and I think we understand a lot of things they don't know about what it takes to succeed at that,' he explained further. Drawing a parallel to past tech rivalries, Altman recalled hearing Zuckerberg discuss how Google tried to enter the social media space in the early days of Facebook. However, to those at Facebook, it was clear that it wasn't going to work for Google. Altman said he now feels similarly about Meta's approach to AI, suggesting that Meta is making an error by trying to replicate OpenAI's success directly. He even discussed how he believes many people at Meta simply copy OpenAI. Altman explained this with an example of how many other companies' chat apps resemble ChatGPT, down to the UI mistakes. He drew from his own experience to argue that the copy-and-paste strategy is fundamentally flawed, and that trying to go where your competitor already is, instead of building a culture around innovation, rarely works. When asked why he thinks Meta sees OpenAI as such a competitor, Altman mentioned how an ex-Meta employee once told him that Meta views ChatGPT as a Facebook replacement. He explained that the user experience with ChatGPT felt different, like one of the few tech products that didn't feel 'somewhat adversarial.' He contrasted this with Google, which he said has started showing worse search results, and with Meta's products, which try to hack users' brains to keep them scrolling. Instead of doing either, ChatGPT simply tries to help users with whatever questions they may have, and even help them feel better. Beyond discussing Meta, the Altman brothers talked about a wide range of topics related to the future of AI, OpenAI's strategy, and even Sam's personal reflections. Altman made a 'crazy claim' that AI will discover new science, and that humanoid robots are one of his dreams — something he thinks will be achievable within the next 5 to 10 years. An internal OpenAI doc reveals exactly how ChatGPT may become your "super-assistant" very soon OpenAI CEO Sam Altman replies to artists irate over their stolen work ChatGPT's Sam Altman threatened to "Uno reverse" Facebook over AI app — he might be dead serious
