This sent me down a rabbit hole of trying a lot more gibberish/unconventional prompts on Midjourney. There’s something fascinating about seeing such a complex response that’s clearly an unintended side-effect of the training process, even though the results have a pretty conventional “Midjourney look”.
Putting in your username is also an amusing exercise. This is what I get for babelspace The ‘space’ influence is pretty obvious, I think it’s associating babel > babble and babies, which I assume is where the color scheme and youthfulness comes from.
If anyone without a Midjourney subscription wants to see what their username looks like, I would be happy to run it for you.
babelspace
joined 1 year ago
I’ve seen the fedi search link before, it’s a great idea. That said, it’s currently no good for finding things on kbin - for example, try searching AskKbin, which has been quite active - you get almost nothing back. Hopefully it will be better in a few days.
Aww. How did she get that name?
Neat. I've never used terms that were as close to gibberish, but I've used made-up extensions of real root words - for ex. "panthalassan" - in image prompts with stable diffusion and midjourney, with decent results. I doubt that it is any more effective than using real words, but it is fun to do.
I really like what you’ve been posting so far. I’ve been working more with image generating AI -but I would have been interested in following information regarding more general/text focused AI on Reddit, except all the communities I checked were pretty bad 😂.
Some places I think you can promote it -
[email protected]
Any big technology community, there’s multiple on different instances
If we can find an article related to image AI, I can post it to the stable diffusion groups. I’ve been posting at [email protected].
Awesome. Though I notice very little shows up from kbin.social; content I know is there is missing when I search for it. That may have more to do with the recency of the site growth or the cloudflare protection that was up a few days ago.
For example, here is one I just made that gathers general music communities across instances: music groups
Good on you! Hope that a lot more people on kbin take this approach.
Assuming you don’t want to go to effort of training for a single image, I’d try inputting the image of a face in img2img mode with some Controlnet options to selectively retain elements of the image.
I like the idea of this community, and subscribed, but fyi that link doesn't work for me on kbin. This is the right link on here, I believe: https://kbin.social/m/[email protected] - m rather than c.
Click the plus icon to the left of your username at the top right corner of the screen . Last option on the drop-down menu is “create new magazine”.