Generative AI

Adobe working to offer GenAI tools from OpenAI and others


Adobe is considering ways to offer third-party generative AI tools from OpenAI, Runway, and others inside Premiere Pro.

The creative content software vendor on Monday previewed ways video editors could use video generation models from the third-party vendors to generate B-roll background video footage in Adobe’s Premiere Pro video editing platform.

The vendor also showcased a preview of GenAI workflows coming to Premiere Pro that it plans to power with a new video model for Firefly, Adobe’s generative AI model for image generation.

Other than the B-roll generator, Adobe said coming tools for GenAI video workflows include Object Addition and Removal, Generative Extend, and Text to Video.

Object Addition and Removal enables users to select and track objects, and then remove them. Generative Extend lets users add frames to make clips longer. Text to Video enables users to generate new footage in Premiere Pro with text prompts.

The preview of the new tools comes days after a Bloomberg report revealed that while Adobe claimed to train its Firefly only on Adobe stock images, it also used AI-generated content, including images from AI image-generating software vendor Midjourney.

Adobe has marketed Firefly as a responsible image-generating platform because users can feel secure that they are not misappropriating images from creative artists that have not licensed their work to Adobe. Firefly plans contain a provision that indemnifies users from litigation claims arising from alleged image copyright violations.

Choices for users

Adobe’s decision to expand beyond only using its own stock images enables it to provide choices to users, Futurum Group analyst Keith Kirkpatrick said.

“They want to make sure that their users have access to really powerful tools in some of these apps,” Kirkpatrick said.

While some enterprises creating hundreds or thousands of variations of content in advertising campaigns may not want to use images from OpenAI or other third-party vendors, others might. Moreover, a sole artist or creator may feel comfortable using those tools because they can vet them themselves, Kirkpatrick said.

The addition of third-party tools will also likely make Adobe more competitive in the GenAI market.

“They’re realizing that to remain sort of competitive in terms of a creative perspective, they need to have a more open ecosystem,” Kirkpatrick added.

Retaining trust

While giving creators access to third-party tools could be risky for a vendor that prides itself on developing responsible GenAI tools, it doesn’t devalue Adobe’s credibility, said Constellation Research analyst Liz Miller. Adobe will likely enable users to use the third-party tools at their own risk, she said.

“Knowing Adobe, there will be a clear and obvious separation of what is covered by Firefly’s indemnity clause and a ‘warning’ for users that they are not using Firefly’s commercially safe models that are covered by their indemnity clause,” Miller added.

“This investigation into partnerships is really more about Adobe ensuring that they are meeting the needs of the creators that rely on their tools,” she said. “These creators are using these different tools and different AI models to create and generate. Adobe needs to remain ahead of their users’ needs and provide pathways that connect these creators and tools with the larger processes of creation, collaboration and workflows.”

However, Adobe’s success in retaining the trust of users and enterprises will come from its clear distinction between what is included in its indemnification clause and what’s exempt.

Moreover, it could lead to trust in the Firefly models, especially as enterprises first create their ideas in it, while using third party tools for further exploration, Miller said.

Other than AI-powered video workflow, new audio workflows will be generally available to Premiere Pro users in May. The audio workflows include a feature that let editors create custom audio transitions faster and the ability to use AI to tag audio clips as dialogue, music, or sound effects.

The vendor also pledged to attach digital labels that show what AI models were used to generate content on its platforms.

Adobe also released an AI assistant that helps users understand and condense digital content for $4.99 per month.

Esther Ajao is a TechTarget Editorial news writer and podcast host covering artificial intelligence software and systems.



Source

Related Articles

Back to top button