Stability AI announces multiple feature updates for Stable Diffusion.

Stability AI has announced several new powerful features for its Stable Diffusion platform. These updates not only provide exciting new capabilities for text-to-image conversion, but also enter the realm of 3D content creation. The most notable enhancement is the brand-new Stable 3D model. Up until now, Stable Diffusion has primarily been used for generating 2D images. The Stable 3D model will change this by offering functionality that can be used for any type of 3D content creation, including graphic design and even video game development. In addition to venturing into the realm of 3D content generation, Stability AI has also introduced a Sky Replacer tool, which, as the name suggests, is used to replace the sky in 2D images. The Stable Diffusion platform also provides Stable Fine-Tuning, aimed at helping businesses expedite the image refinement process for specific use cases. Furthermore, the company will integrate invisible watermarks into the images generated by the Stability AI API for content authentication. These new updates are designed to assist businesses in creative development as generative artificial intelligence becomes increasingly integrated into regular workflows. Emad Mostaque, CEO of Stability AI, stated in an interview, "This is a way to provide the necessary tools for creative storytellers to have an additional level of control over images." Stable Diffusion adds functionality in the fiercely competitive GenAI field Stability AI's advancements come at a time when the text-to-image generation market is becoming increasingly competitive. Adobe has entered this market with its Firefly tool, tightly integrated with the company's design software. Midjourney is continuously adding new features to its technology, helping designers generate images. Meanwhile, OpenAI recently released its improved DALL-E 3 model for generating text within images. Mostaque is well aware of his competitors and is striving to differentiate Stability AI from other companies in various ways. In particular, he emphasizes that his company is no longer just about models, but about achieving creative workflows. He points out that the new Sky Replacer and fine-tuning features are additional steps beyond the core foundational models for generating images. Sky Replacer is not only a feature, but also a focus for enterprise use cases The concept of replacing image backgrounds is not new. In non-generative AI applications, backgrounds are often replaced using techniques such as green screens and chroma keying. Mostaque states that Stability AI is building automated workflows based on these classic techniques to enable rapid and efficient background replacement for enterprise users. Changing the background color of the sky is not just about adding creativity; it has very specific and practical use cases. Mostaque says, "Sky Replacer is great for the real estate industry." He points out that users want to have different backgrounds with different lighting effects. Fundamentally, he emphasizes that all of this is about providing control, as organizations have their own workflows for generating images and content. What Stability AI is doing is building optimized workflows to help achieve the desired control for different use cases. He says, "Sky Replacer is the first in a series of very professional and enterprise-specific tools that we are about to launch. We have gained relevant experience over the past six to twelve months." Stable 3D expands the application scope of Stable Diffusion The new Stable 3D model adds additional 3D datasets and vectorization capabilities by extending the diffusion models used in Stable Diffusion. "I'm very excited about the ability to create an entire 3D world," says Mostaque. He explains that Stable 3D is built on the work of Stable Diffusion and Stability AI on Objaverse-XL, one of the world's largest open 3D datasets. Building and rendering 3D images has long been a resource-intensive process, but Mostaque is optimistic about Stable 3D being more efficient than traditional 3D image generation methods. He emphasizes that Stable AI is still in its early stages, but he is optimistic about the gradual development and expansion of this technology. Stable 3D will initially be offered as a private preview. He says, "It's very efficient compared to classical 3D model creation. Things that used to take a long time to build can now yield preliminary results quickly." Watermarks and content credentials Stability AI is now integrating invisible watermarks and content credentials into its API. Content credentials are part of a multi-vendor industry effort that Adobe and other companies are involved in, aiming to provide information about the authors of content. Mostaque states that adding invisible watermarks and content credentials is a responsible practice. This is also part of a broader effort by Stability AI to bring authenticity to generated content. "We are pioneering many initiatives, as well as some other announcements we will make in this field, along with additional research, because we want to know what is real and what is fake. This also helps us build attribution and other mechanisms for future releases," says Mostaque.