in connection with the fire around the new censorship rules (as well as related discussions), I would like to describe a couple of thoughts regarding neuro-images: development and related problems
1 - offtop but i must mention it.
first, create an event with a girl of indeterminate age who carries a gun, is often smeared in blood, and drinks alcohol (they literally added that moment with the civcan to their own dataset to lora). And a month later, banning what you've recently done yourself is an incredibly hypocritical garbage.
2 - as I see the situation with neuro-images in general, development has reached a plateau and qualitative growth is not expected in the near future.
instead of increasing the tools and improving the variety of approaches to work, we have increased resolution and millions of attempts at photorealism.
There is no way to build a composition, work on multiple planes, or work correctly with multiple characters. Should we be surprised at the prevalence of pornography on platform if it is not suitable for something more meaningful?
Of course, you can generate brainrot images on Instagram-slop level. BUT there is no need to build illusions about the progressivity of what is happening here, especially telling someone about art (one local madman even tried to convince someone that comics are ART, LOL)
As in any other industry, it takes time to develop, and in a few years it will become clearer why it's needed - does it all make sense?
ps. Some bad people, God bless them. the cover was blocked without explanation - the post stopped being visible for a while