Wisdom news on October 11, in the early hours of this morning, the world's largest design software leader Adobe threw out three main new models in the Adobe Firefly creative generative AI model series: Adobe Firefly Image 2 (Firefly II imaging model), Firefly Vector Model (Firefly vector model) and Firefly Design Model (Firefly Design Model).
Adobe Firefly Image 1 and 2 generate image effect comparison
Among them, the Firefly vector model is the world's first generative AI model focused on generating vector graphics, which brings vector graphics and generative AI expertise directly into the Adobe Illustrator workflow through text-to-vector graphics capabilities.
Adobe also announced more than 100 major AI innovations and updates to Creative Cloud, including vector graphics software Adobe Illustrator, image editing software Adobe Photoshop, online photo editor Adobe Lightroom, video editing software Adobe Premiere Pro, and graphics video processing software Adobe After Effects and image processing software such as Adobe Stock are being transformed by the Adobe Firefly model and other AI-driven features.
For example, if you want to intelligently fill an image in Photoshop, you only need to select the extended image size and click Generate.
David Wadhwani, president of Adobe's digital media business, said that in March Adobe launched a beta version of the Firefly image model, and to date, users have generated more than 3 billion images using the tool. And the Firefly-powered generative fill technology in Photoshop has increased user adoption by 10 times compared to previous versions. It can be seen that AI is deeply integrated into the core of the creative workflow.
Unified use of all apps in Creative Cloud is priced at $54.99 per month and Photoshop at $20.99 per month.
Other pricing references: https://www.adobe.com/creativecloud/plans.html
Free trial link: https://www.adobe.com/creativecloud/features.html
The wave of generative AI not only makes AI startups such as Midjourney and Stability AI stand in the spotlight of capital, but also makes old software companies such as Adobe revival their sensitivity to technology and product upgrade iteration. In the past year, Adobe's stock price has risen by nearly 90%, and as of today's local time, Adobe's market value reached $242.5 billion (1,768.8 billion yuan).
1. Automatically match the same style to generate, copyright risk pictures cannot be referenced
Adobe Firefly Image 2 is a second-generation imaging model that enhances the rendering quality of images by improving human skin, hair, eyes, hands, and other body structures in images, resulting in higher quality images for users.
The Firefly Image 2 model has three new features: Generative Match, Photo Settings, and Prompt Guidance.
Generate matching means that the Firefly Image 2 model can generate similar images based on user-provided reference images. Users can select an image from a preselected list in the tool, or upload a reference image themselves, and the Firefly Image 2 model can meet the appropriate brand guidelines or copy the style of an existing image, saving them time designing from scratch and keeping the style of the image consistent.
Finally, Adobe stores thumbnails of reference images on its servers, prompts users for messages based on the Generate Match feature, and then asks them to confirm that they have permission to use the uploaded image and other terms of use.
When an image uploaded by a user is at copyright risk, or if the image is marked as "protected content", the Firefly Image 2 model cannot replicate its style:
In the photo setup function, the Firefly Image 2 model can generate more fidelity details such as skin pores and leaves, and for higher quality images, users can also control depth of field, motion blur, field of view and generation.
At the same time, users can apply and adjust photo settings, similar to manual camera lens control. Among them, Auto mode automatically selects "Photo" or "Art" as the image generation style, and then the user selects the appropriate photo settings to apply.
In terms of hint guidance, the Firefly Image 2 model improves the understanding of text cues and can identify more landmarks and cultural symbols.
During the model authoring process, users can receive suggestions for improvement prompts to spark new creative ideas while reducing the number of rebuilds. Based on this feature, the user can determine how to prompt effectively, or whether it should be rewritten.
In addition, users can select specific elements such as key terms, colors, shapes, etc. to avoid generating images that they don't need.
Finally, during the image saving stage, users can share and save images directly from Firefly and fine-tune the image with their favorite tips. Thanks to its sharing feature, users can share their creations and let other users know how to create specific images, and images in the shared library can be opened and used by others.
Second, the world's first vector graphics generation model, gradient, layering, tiling can be realized
Adobe claims that the Adobe Firefly vector model is the world's first vector graphics generative AI model and the first generative AI model to generate "human quality" vector and pattern output.
Adobe vector graphics software Illustrator is now available in beta of its text-to-vector feature, where creative researchers can generate vector graphics that can be edited and organized into groups.
There are five new features for vector models that can generate matching images, vector gradients, grouped layered output, seamless tiling, and output precise geometry.
Among them, the Firefly vector model feature ensures that vector graphics added to the user's artboard perfectly match the existing style. This allows the model to produce high-quality vector output for scenes, themes, icons, and patterns.
Adobe officially says that in general, gradients are a challenge in the vectorized world. Firefly vector models can now generate editable vector gradients that simplify geometry without the need for multiple solid fill paths to depict the same image.
At the same time, when generating vector graphics, all elements can be logically grouped and layered. For example, in a vector illustration of a tree, it automatically groups all the relevant paths that make up the tree, making it easier for users to adjust, edit, and reuse the final vector output.
Firefly vector models generate vectorized patterns that tile seamlessly, so even if the design pattern is repeated infinitely, there will be no noticeable gaps in between.
Finally, the high-quality vector curves output by the model are smooth and accurate, and can intelligently handle the intersection of multiple curves.
Third, AI automatically generates templates, and provides text and image content in one stop
Firefly Design Model enables instant generation of template designs through the text-to-template feature in Adobe Express, combined with the professional layout technology of Firefly Image Model, Adobe Stock and Adobe Fonts.
In addition, the Firefly design model supports a large number of mainstream scale templates, which are fully editable in Express and are customized for different uses such as print, social, and online advertising. Users can generate images in Firefly and then continue refining hints in Express, enriching titles with text effects supported by Firefly.
The model uses AI to generate templates and then uses its auto-generated text and images to help users quickly choose from thousands of templates and identify layouts that best suit their needs.
In terms of content generation, users can generate text and images based on simple prompts, while selecting text size, color, and position, and cropping, replacing, or rotating the image as needed.
Firefly design model adopted
Instant end-to-end designto create images, text, and tones that harmonize with user prompts.
Fourth, a number of generative AI tools have been updated, and the image editing workflow has been revolutionized
Ely Greenfield, CTO of Digital Media, Adobe, said: "We are rapidly and continuously advancing Firefly innovation, advancing in imaging, design, and vector generation, and these three new models set a new industry standard for output quality and user control. ”
In addition to producing editable high-quality vector graphics, Adobe has updated Retype, a tool that quickly identifies similar fonts in Adobe Fonts, converts static text into editable text, quickly converts images and graphics into realistic product and brand models, and adds "Illustrator" to make it easier to collaborate and exchange feedback between designers." Share for Review".
Adobe Photoshop, a wider range of image editing software, is much more usable based on the Firefly model, and Photoshop subscribers can now experience Photoshop's AI-based capabilities through the web version.
In Adobe Lightroom, an online photo editor, AI improvements include lens blur, HDR optimization, and a simplified editing experience in Lightroom Mobile.
Among them, HDR optimization enables anyone who takes a photo to edit and export it with brighter highlights, deeper shadows, and more vivid colors; The Point Color feature helps photographers achieve precise, accurate, and detailed color refinement, while providing users with content credential support for new file types.
Lightroom's lens blur feature
Adobe also released new AI-based text editing and motion improvements in Premiere Pro and After Effects, allowing users to seamlessly publish video content to social platforms like TikTok and YouTube from tools like Express and Premiere Pro.
The image editing process of image processing software Adobe Stock has also been simplified, allowing users to transform concepts into actual visuals with simple text prompts, and expanding image backgrounds and aspect ratios with a single click.
In addition, to make the design process easier for users, the online design tool Express allows users to use Generative Fill, a creative co-pilot powered by Firefly. Based on this, users can insert, delete or replace objects, people, etc. in any image through text descriptions.
At the same time, based on the Firefly design model, users can also use text descriptions to generate editable templates, and the translation function that comes with the tool supports 45 language options.
Adobe also introduced GenStudio, a new end-to-end solution that provides brands with generative AI-based capabilities to accelerate content ideation, creation, production, and activation, including instant editing of digital content secured for commercial use through Express and Firefly in Adobe Experience Manager Assets, a digital asset management system.
To make your brand identity stand out, Adobe allows companies to customize and fine-tune Firefly using their own style, persona, and objects, and fully integrate custom Firefly models into the company's Creative Cloud workflow and automate production.
Havas, one of the world's six largest advertising and media groups, announced an expanded partnership with Adobe to adopt Adobe GenStudio.
Conclusion: In the era of AI creation, the business path of the old creative software giants is clear
With the momentum of Firefly's full launch and several tool updates, Adobe has demonstrated its innovation in redefining the creative workflow. David Wadhwani, president of Adobe's digital media business, said, "Since releasing our first Firefly image model in March, we've been amazed by the amazing response from the community to show how useful Firefly can be for enhancing experimentation and design, as well as speeding up workflows without compromising creative control. ”
Today, generative AI is reinventing the way people generate and edit images and videos, making it easy to set a specific depth of field for an image, extend the background, and generate a variety of imaginative images in a single sentence, which is a wake-up call for the established creative software giant. Adobe has once again strengthened its deep accumulation in the field of image editing with a series of updates. Compared with the hot Midjourney and Stable Diffusion, Adobe emphasizes the secondary editing ability of images and simplifies the user's use process step by step.
In an interview with the Financial Times in June, Adobe CEO Shantanu Narayen compared current generative AI to previous technology platform shifts, predicting that many venture-backed AI companies have sprung up but lack obvious business models and will eventually face a reshuffle. However, in previous platform transformations, the market will remain uncertain for a long time until the ultimate winner emerges.