James Cameron Joins Stability AI Board of Directors

Is that something you'd really have use for? In case of images it's easier to immediately judge the effect of a prompt change and its direction. LLMs just blurt out a blob of text that you have to read and analyze every time. Even when rerolling the output for the same prompt, you get a very different-looking result. I find comparing a large number of prompt-output pairs very exhausting. Meanwhile, scrolling through a set of outputted images is a breeze, and the output quality is comparable at first glance.

LLM prompts can also be much longer, consisting of multiple sets/layers of instructions, etc. You just slap a new sentence at the end to modify the behavior. Meanwhile, for image generators it feels like every word counts, and can sometimes take the output in an unintended direction, hence the need for surgical control over word weights.

Not everything is NSFW and the user base does push the technology.

Stability does create the best results though, so if I had significant investment capital, I certainly would be heavily interested. For technology reasons at first more than a business plan. Although there is an enormous potential for the latter here, that I believe can justify investments as well. Might be a more long time invest though.