Adobe Adobe Releases New Firefly Generative AI Models and Web App; Integrates Firefly Into Creative Cloud and Adobe Express

Connect with our team and fellow users to exchange ideas, share your creations, stay updated with the latest features and announcements, and provide feedback. Commercial viability of Firefly is, obviously, a huge part of Adobe’s plans for the software, it just seems like the steps the company is taking are out of order. Adobe is jumping into the generative AI game with the launch of a new family of AI models called Firefly. She joined IDG in 2016 after graduating with a degree in English and American Literature from the University of Kent. Trueman covers collaboration, focusing on videoconferencing, productivity software, future of work and issues around diversity and inclusion in the tech sector.

One example during a demo at the Summit used simple text prompts to transform a static image of a field in springtime full of flowers into the same image during a winter storm. For its marketing and social media tools, meanwhile, Adobe imagines scenarios such as being able to upload a mood board to help with content Yakov Livshits creation and original customizable content. For one thing, it will include context-aware image generation to allow users to experiment with concepts. In illustration, artwork and graphic design, Firefly might also generate custom vectors, brushes and textures from commands or based on a simple sketch.

Adobe opens up its Firefly generative AI model to businesses

Adobe plans to make each design editable using tools that users are already accustomed to, making the process simple to navigate for most users. Those with a paid Creative Cloud plan, whether it is an all-apps or single-app plan, will be given a monthly allotment of Generative Credits. After these are consumed, users are subject to slower content generation unless they buy additional credits. Adobe Firefly-powered features are now available in several Creative Cloud apps, including Generative Fill and Generative Expand in Photoshop, Generative Recolor in Illustrator and Text to Image and Text Effects in Adobe Express.

Data augumentation is a process of generating new training data by applying various image transformations such as flipping, cropping, rotating, and color jittering. The goal is to increase the diversity of training data and avoid overfitting, which can lead to better performance of machine learning models. Adobe is taking a measured approach to rolling out features, starting with text-to-image, to give the market for generative AI a chance to settle down as the company experiments with new concepts. The company seeks to engage with the creative community during the beta, hoping to gather feedback to shape future product iterations. The future of Firefly is largely experimental and, at this point, hypothetical as Adobe works through the needs and concerns of users, brands and creators. Future Firefly models will leverage a variety of assets, technology, and training data from Adobe and others as other models are implemented, with the company saying it will continue to prioritize countering potential harmful bias.

Adobe wants to ensure trust and transparency in AI-generated content

Adobe believes that the full power of technology cannot be realized without great imagination to fuel it. Through the beta process, the company will engage with the creative community and customers as it evolves this transformational technology and begins integrating it into its applications. The first applications that will benefit from Adobe Firefly integration will be Adobe Express, Adobe Experience Manager, Adobe Photoshop and Adobe Illustrator. The general availability of Firefly for Enterprise brings groundbreaking generative AI capabilities to Adobe GenStudio and Express for Enterprise. In addition, Adobe is working with Enterprise customers to enable them to customize models using their own assets and brand-specific content. Customers will also get access to Firefly APIs, embedding the power of Firefly into their own ecosystems and automation workflows.

Video Generation involves deep learning methods such as GANs and Video Diffusion to generate new videos by predicting frames based on previous frames. Video Generation can be used in various fields, such as entertainment, sports analysis, and autonomous driving. Speech Generation can be used in text-to-speech conversion, virtual assistants, and voice cloning.

Yakov Livshits
Founder of the DevEducation project
A prolific businessman and investor, and the founder of several large companies in Israel, the USA and the UAE, Yakov’s corporation comprises over 2,000 employees all over the world. He graduated from the University of Oxford in the UK and Technion in Israel, before moving on to study complex systems science at NECSI in the USA. Yakov has a Masters in Software Development.

A world that lets you merge the best of application workflows with the best of generative AI — where your creative applications act as a creative co-pilot and help you instantly iterate through hundreds of variations of your work, all in your unique style. Because of generative AI, the conversation between creator and computer will transform into something more natural, intuitive and powerful in the years ahead. Beyond unresolved questions around artist and platform compensation, one of the more pressing Yakov Livshits issues with generative AI is its tendency to replicate images, text and more — including copyrighted content — from the data that was used to train it. AI like the first Firefly model “learn” to generate new images from text prompts by “training” on existing images, which often come from data sets that were scraped together by trawling public image hosting websites. Some experts suggest that training models using public images, even copyrighted ones, will be covered by fair use doctrine in the U.S.

And, of course, all generative AI content should be tagged in CAI’s Content Credentials. Barring a major setback on the copyright or licensing front, Adobe plans to forge ahead with Firefly, eventually introducing models that not only generate images and text but illustrations, graphic designs, 3D models and more. Costin was adamant that it’s a major area of investment for Adobe, whose last big gamble — the $20 billion acquisition of startup Figma — is on the cusp of being blocked by a Department of Justice lawsuit, reportedly. On the second point, Costin says that Firefly models were Yakov Livshits trained using “carefully curated” and “inclusive” image datasets and that Adobe employs a range of techniques to detect and block toxic content, including automated and human moderation and filters. History has shown that these sorts of measures can be bypassed, but Costin suggests that it’ll be a carefully guided — if imperfect — learning process. It’s an expansion of the generative AI tools Adobe introduced in Photoshop, Express and Lightroom during its annual Max conference last year, which let users create and edit objects, composites and effects by simply describing them.

How Long Did It Take Top Apps to Reach 100 Million Downloads?

Or even how it might help the company’s creative community get the most value from their Firefly experience, both monetarily and creatively. The beta version of Firefly is not-for-commercial-use, web-only and supported on Chrome, Safari and Bing. It is currently not available on tablets or mobile, although those devices will eventually be supported. Firefly in general availability (GA) will first be integrated into Adobe Experience Manager, Express, Photoshop and Illustrator. Eventually, Firefly will be integrated across all Adobe products into customers’ content creation workflows. No, generative credits don’t roll over to the next month because the cloud-based computational resources are fixed and assume a certain allocation per user in a given month.

• Adobe includes credits to use Firefly in varying amounts depending on which Creative Cloud subscription plan you’re paying for, but it’s raising subscription prices in November.
• Firefly is smart enough to get the crab’s reflection mostly right, though if you look closely, imperfections are evident.
• Some people are concerned about the ethics of using generative AI technologies, especially those technologies that simulate human creativity.
• On the second point, Costin says that Firefly models were trained using “carefully curated” and “inclusive” image datasets and that Adobe employs a range of techniques to detect and block toxic content, including automated and human moderation and filters.
• Creative Cloud All Apps includes 1,000 monthly credits, while a single-app plan doles out 500 credits per cycle.
• By combining capabilities in Adobe’s Sensei AI such as data analytics and behavior predictions with Firefly image generation, brands can make the AI “conversation” even richer, enabling them to extend and leverage the full view of the customer.

The CAI’s size — roughly 900 members — makes it more likely that Adobe’s proposals will gain some sort of traction. Artists could find themselves in a situation where they’re forced to use multiple opt-out tools to prevent their artwork from being trained on. But content creators who choose will be able to opt out of training, Adobe says, by attaching a “do not train” credentials tag to their work. On a technical level, the first Firefly model isn’t dissimilar to text-to-image AI like OpenAI’s DALL-E 2 and Stable Diffusion. Both can transfer the style of one image to another and generate new images from text descriptions. Adobe said Firefly has been designed to serve users with a wide array of skill sets and technical backgrounds, supporting text prompts in over 100 languages.

With the move to public release, something else is changing about Adobe Firefly aside from who can use it — how people use it. Firefly underpins popular new features like Generative Fill and Generative Expand in Photoshop. Firefly also helps users retouch and restore damaged photos, which will help preserve precious memories for many. Costin explained that the team had looked at a token-based system, but the feedback from early testers was that this was too hard to explain to customers.

We want your feedback on how best to shape the future of generative AI for creativity. We’ll be introducing future models tailored toward different skill-levels and use-cases, making use of a variety of assets, tech and training data from Adobe and others. We’re planning to include models that creators can train on their own personal style and brand language — which will be hugely beneficial to individual creators and enterprise creative teams. We will also train models on a broader variety of content sources, while working to counter potential bias or other harms in generated content. Generative AI will create a lot of opportunities and raise a lot of questions, and we know the best way to move forward is in partnership with all of you.