Will Adobe Content Credentials or Text-to-Vector Graphics Be More Impactful?
The answer may depend on your time horizon and profession
Adobe announced this week the relaunch of ContentCredentials.org. In addition, the company said Microsoft, Publicis, Leica Camera, and Nikon have adopted the icon of transparency. This initiative aims to create a standard digital mark, metadata, and validation method for determining the provenance of digital content.
Microsoft is adding the technology to Bing Image Creator, Publicis is using it through Adobe products, Leica Camera will add it as a feature for a new device launched this year, and Nikon will also add it to a camera. There are many reasons to applaud this initiative. Creating a record of digital content origin and evolution can be useful in protecting IP, understanding the true source of content, and providing proper attribution to creators. But, does this program have any momentum?
Origin of Content Credentials
In February 2021, Adobe, Arm, BBC, Intel, Microsoft, and Truepic launched a formal coalition for standards development: The Coalition for Content Provenance and Authenticity (C2PA). It is a mutually governed consortium created to accelerate the pursuit of pragmatic, adoptable standards for digital provenance, serving creators, editors, publishers, media platforms, and consumers.
Just to confuse this a bit more, the Content Authenticity Initiative is an evangelism organization that promotes the development of content authenticity standards, technical frameworks, and tools. It supports activities such as the C2PA and Content Credentials. The manifestation of this work is intened to show up in products such as Adobe Creative Suite, Microsoft’s design and AI solutions, and products and apps from a wide variety of companies.
The C2PA has 56 members. The Content Authenticity Initiative has over 200. Content Credenditals has what appears to be five. It is not clear that there truly is a lot of momentum behind market adoption. The addition of four companies, all featured members of the C2PA, is a postivie step, albeit a small one.
The programs are also not exclusive to AI-created assets. It is more generally applied to digital assets.
What Will Adoption Mean?
If this collection of technologies and practices is adopted widely, there will be some clear benefits for creators, buyers, and users. However, it may not have the type of impact that many people would like to see for AI-created art.
Some companies and creators that want to identify the provenance of their digital assets will embrace the approach. This will also serve as a market signal that assets without the mark and metadata are not authentic to the companies and individuals that use it. Of course, there is a loophole of digital assets created before adopting the approach. This will apply to newly created assets.
The bigger loophole is for consumers of digital assets. Since most digital assets will not include content credentials data or markings, those that have this information will not be widely encountered. That means its usefulness will be limited for consumers. The presence of the mark and data may be helpful in establishing provenance, but the absence of these indicators will be far more common, and the status quo will remain.
The biggest loophole will be fraud. Nefarious actors will sometimes not apply content credientials and marks to ensure its assets blend in with the digital sea of uncredentialed content. Other times, they may attempt to recreate the data and marks to fraudulently suggest a false provenance.
All of this aside, the C2PA and Content Credentials could be, but are not necesssarily a solution for people who want all AI-created or AI-assisted content labeled as non-human in origin. This was not the original objective of the initiative.
Text-to-Vector Graphics
The new Firefly Vector Model brings together Adobe's vector graphic and generative AI expertise to power the next generation of high-quality designs and illustrations. It is the world’s first generative AI model for vector graphics and first generative AI model to generate “human quality” vector and pattern outputs…This innovation will speed up inspiration, mood boarding, marketing and advertising graphic creation.
It remains to be seen whether this new model will generate the same output quality as other text-to-image solutions. However, the ability to easily edit outputs could quickly make it a preferred solution for professional designers.
There is widespread agreement about the ability to generate high-quality image outputs from AI diffusion models. However, most outputs are essentially uneditable. That limits the potential application of many AI-generated images. With vector-based images, those limits essentially disappear. This change should accelerate the use of AI-generated images by enterprises, agencies, and other organizations.
At the same time, it may not have much uptake in the consumer market. Few consumers have the skill or experience to manipulate digital assets. That means there is little added benefit from editable outputs as opposed to non-editable. The professional crowd will be the primary user base for vector-based text-to-image.
Short and Long Term
In the near term, the ability for AI models to generate vector images will have more impact on the market. The key friction point at the moment appears to be the limitation of this feature to Adobe Illustrator only. As this solution becomes available in other applications, its adoption will accelerate. By contrast, content credentials will face slower adoption.
Provenance could have a bigger impact over the long term. Of course, that will be after the vector graphic adoption wave among design professionals is complete.