Stable diffusion - released by a ground breaking company named Stability.AI - is an open source AI image generator that i installed on my MacBook Air M2 over the labor day holiday. In two words, I'm amazed! This technology seems to be going largely unnoticed, which is surprising considering the potential that it has to transform society and the workforce as we know it. Here at BeCloud we understand these revolutionary changes and are scrambling to keep up with the evolving tech revolution happening all around us. Though the technology is in its infancy I believe that we are on the cusp of a tech transformation that will transform how we create and interact with images and AI.
Fast on consumer hardware
On my M2 MacBook Air the images utilized on this blog posting were generated in under ten minutes. With a better setup and more tweaking it would be seconds.
Image generation for marketing, advertising and simulation is currently utilizing technology with origins in the 90s. This software uses basic primitive shapes like spheres, cones and cylinders which can be manipulated by a user to produce the environment they want to see in the final image file. The resulting images are pleasant for visualizing new product designs or environments but they lack realism and detail that we as human beings intuitively understand. To make up for this shortcoming it is necessary to hire an artist or photographer who has extensive knowledge and experience to manually recreate what the computer can’t do. The cost of this is significant and can take months to complete.
Stable diffusion employs a innovative approach by using artificial intelligence to create realistic images of people, places, things and animals without requiring human intervention. The images generated by stable diffusion are lifelike with accurate detail and lighting that are difficult for a human to distinguish from real photographs. The technology is complex but revolutionary in its ability to utilize natural language input from a user and turn that into an accurate visual representation. This means that anybody can conjure up almost any visual reality they can imagine by simply stating what the subject matter is to be.
The AI used to power stable diffusion is not new but what has been done here is unique and important, it allows any business or individual access to a level of visual realism that was previously only available via manual methods, at a fraction of the cost. BeCloud agrees with Stability. AI on this point, we are amazed at the potential of this technology but cannot afford to wait a year or two for the service to mature. With our clients we feel that stable diffusion could be a game changer. The tech is designed in such a way that it does not require high powered supercomputers or expensive photo editing software, the end product is beautiful and professional. What’s more is that the technology is simple to use and anyone can operate it. We believe that this is a key moment in the tech world.
By making Stable Diffusion truly opensource Stability, AI, has opened up pandora's box for users to integrate this AI technology in ways not previously conceived. Becloud is already starting to integrate stable diffusion in some of our existing projects "soon to launch" and we are having internal discussions about what the future might hold and what this technology could deliver.
We have also started some research into application integration for stable diffusion. The low hanging fruit is generating images to use in our blog post and customer websites without requiring a graphic artist. The tech can work with any web site or app that needs gorgeous images and we intend to seize maximum advantage of this technology. We can see many cool uses for stable diffusion (AI generated images); but the real value lies in a much bigger idea; an AI visualization layer that exists between people and the digital world, using stable diffusion as an effortless way for anyone to create eye-catching visuals that help drive engagement on all kind of platforms.
Just as with most tech, it is conceivable that in the near future we may see a day when AI is easier for us to fully use, such as learning to drive cars and fly planes.
I have no clue what the upcoming years will bring but I can tell you one thing; this tech is not going away without a fight.