Introduction:
AI image producer Fans of Stable Diffusion already number in the thousands, and now there’s a platform for fans to share their work with others exploring the new AI technology to create their models. A startup company named Civitai, Andreessen Horowitz backs Civitai, which means community, has developed a website where users can upload their own AI picture models based on Stable Diffusion for other users to find and their own AI photos, the result of their labour, for others to peruse and appreciate.
According to the CEO of Civitas, Justin Maier, the company’s concept originated from his observation that there was a need for a location where individuals could share their models and others could find them. He claimed that although some people posted their photographs, others could not replicate their efforts to create their own images.
Maier became interested in Midjourney after concluding a job at Microsoft, where he worked as a web development project contractor.
Simultaneously, an open-source version of Stable Diffusion was released, enabling nearly identical AI picture-generating capabilities. As the Stable Diffusion community expanded, users learned new ways to interact with the model by adding new ideas or urging others to do so. One such approach was to use their photos to create AI-generated selfies.
Maier began to see that others were sharing their models, which allowed you to do fascinating things like generate in a punk or synth wave manner or combine models to create new concepts. However, these were shared online, not in an organized community, but on sites like Reddit or Discord.
This inspired him to start Civitai (formerly known as Model Share), an online community that arranged this material so users could discover it later.
He first started Civitai by contacting model creators and requesting permission to publish their creations on the platform. Initially, the range of models was limited to about 40 or 50. But eventually, the website started to expand.
“People were ecstatic to be acknowledged for their efforts. According to Maier, We became the preferred location for sharing these items by roughly January. “People were posting to other websites, such as HuggingFace…yet, it wasn’t image-focused,” he clarifies. Maier continues, “We established ourselves as the de facto standard for exchanging your model, different AI resources, and your created images.”
When the website reached 100,000 registered users by January 2023, Maier realized it had the potential to grow from a community effort into a business. After being “officially” established for three months, Civitai had one million registered users. There are about 3 million registered users, and each month, there are about 12 to 13 million unique visits.
“It’s grown wildly quickly; it’s bigger than we anticipated in January,” remarks Maier.
Andreessen Horowitz backs Civitai:
Andreessen Horowitz backs Civitai (Image Source: techcrunch.com)
Just 10,000 original authors out of the 3 million members upload new models monthly. However, since Citivai made it easier to get started by allowing users to train new models on the website, that number has increased by about 25% in the past month. More people are using the models and producing content.
However, Maier thinks there might be a future market for ethical AI work, i.e., using AI imagery in applications, films, music videos, and other creative pursuits. However, the immediate goal is to develop a user-facing mobile app that serves as a storehouse for the AI imagery and supplements the main website’s experience.
The business now intends to enable users to make money from their work. This will include more direct forms of monetization, such as paying for one-time image generation or model access and linking users with brands looking to employ AI to develop original concepts. However, everything on the website is currently free to use. (The business uses Cloudflare’s R2 to reduce download expenses.)
To enable them to sell their image to people who wish to use it for commercials and other purposes, Maier adds, “I’d like to start working with brands, real people, and IP.” Furthermore, we can configure licensing and permissions to limit the uses for which it can be used effectively. That, in my opinion, will be crucial as we consider how things develop in this AI era,” he continues.
Though that will come later, Maier notes that the business hopes to expand to additional modalities outside AI picture models.
