Adobe Firefly Image 5 brings support for layers, will let creators make custom models

Adobe said on Tuesday that it is launching the latest iteration of its image generation model, Firefly Image 5. The company is also adding more features to the Firefly website, support for more third-party models, the ability to generate speech and soundtracks. Notably, the update allows artists to come up with their own image models using their existing art.

Image 5 model can now work at native resolutions of up to 4 megapixels, a massive increase from the previous gen model, which could natively generate images at 1 megapixels but then would upscale them to 4 megapixels. The new model is also better at rendering humans, the company said.

Image 5 also enables layered and prompt-based editing — the model treats different objects as layers and allows you to edit them using prompts, or use tools like resize and rotate. The company said it makes sure that when you edit these layers, the image’s details and integrity are not compromised.

Adobe’s Firefly site has supported third-party models from AI labs like OpenAI, Google, Runway, Topaz, and Flu to augment its appeal to its creative customer base, and now the company is taking that a step further by letting users create custom models based on their art style. Currently in a closed beta, this feature lets users drag and drop assets, such as images, illustrations and sketches, to create a custom image model based on their style.

The company is also adding some new features to its Firefly website, which was redesigned earlier this year. The site now lets you use the prompt box to switch between generating images or videos, choose which AI model you want to work with, change aspect ratios, and more. The site’s home page now features your files and recent generation history, and you also get shortcuts to other Adobe apps (these were previously housed in a menu).

Adobe has also redesigned the video generation and editing tool to support layers and timeline-based editing. This design change is currently only available in a private beta, and will be rolled out to users eventually.

Firefly is also getting two new audio features: Users can now employ AI prompts to generate entire soundtracks and speech — using models from ElevenLabs — for videos. There’s also a new way to easily come up with prompts: just add keywords and sections by selecting words from a word cloud.

As its competitors like Canva add AI to their platforms, Adobe is trying to cater to new-age creators who are increasingly using AI in their workflows. “We’re thinking of the target audience for Firefly as what we call creators or next-generation creative professionals. I think there are these emergent creatives that are GenAI-oriented. They love to use GenAI in all their workloads,” Alexandru Costin, the company’s VP of generative AI, told TechCrunch over a call.

He added that with Firefly, the company now has more freedom to add new features and play around with the interface as it doesn’t have to adhere to the muscle memory of creative professionals who might be used to certain workflows in Adobe’s existing Creative Cloud tools.