The innovative AI Startup Stability AI, which is renowned for its text-to-image model Stable Diffusion has made it to the headlines with the announcement of its new project, StableStudio. It is an open-source version of its commercial AI-powered design software, DreamStudio. This unique and interesting move is an attempt to move closer towards their aim of collaborative development and outperform the speed at which things move in this generative AI and art field compared to individual companies and their closed-door efforts. This move is also considered a leveraging move after recent investments from tech giants like Google, Microsoft, and Amazon.
Stability AI has published an essay in which it discussed its opinions on potential future developments for generative AI.
They contend that the growth of generative AI would be more effectively facilitated by an open, cooperative, and community-driven development. They described their desire to collaborate with the larger community to develop the finest user interface possible, giving consumers complete control over the generative AI’s creative potential.
Started as an animation studio for the open-source generative AI art model Disco Diffusion, DreamStudio has gradually set a new goal and focus on image generation with the introduction of Stable Diffusion. This new shift in focus has put DreamStudio in close and tight competition with other rival generative image platforms like Midjourny and NightCafe.
While StableStudio and DreamStudio share many similarities, there are some important differences. Stable Studio does not share DreamStudio brandings or Stability-specific account features such as billing and API management. In addition to that, the backend API calls have been replaced by a plug-in system.
Even though StableStudio shares a vision of collaborative development, some skeptics view this release of StableStudio as an attempt by Stability AI to outsource the development of DreamStudio to the open-source community. While this perspective isn’t entirely baseless and illogical, it is also a fact that Stability AI is under a lot of pressure to monetize its diverse efforts, spanning art, animation, biomedicine, and generative audio.
CEO of Stability AI, Emad Mostaque, hinted at going public with Stability AI with an initial public offering (IPO). Interestingly, according to the recent posts on Stability Ai, despite raising over $100 million in venture capital in October last year with a valuation exceeding $1 billion, it has needed to be faster to generate revenue and is rapidly burning its cash reserves.
This new approach comes with an existential challenge for Stability AI. Unlike Stable Diffusion, which the company did not inherently develop but partnered up with research organizations to create and build upon, Stability AI primarily focused on providing cloud access to the computational power required for training AI models rather than developing its models.
This approach is evolving for good. There were some key highlight developments several weeks ago, such as Stability AI released a suite of text-generating AI models aimed at competing with systems like OpenAI’s GPT-4 and ChatGPT. On top of that, Stability AI unveiled Stable Diffusion XL (SDXL), an enhanced version of the original model that introduces significant improvements, such as the generation of hands.
This bold move by Stability AI, open-sourcing their project, aligns strategically with the company’s ongoing efforts to secure more funding. As time evolves, let us see how this move turns out for them.