lool

Me neither. From what I can tell, the author isn’t either if you pay attention to the end of the article.

I believe the absolute worst case scenario is that we grow to rely on it so much that it ends up controlling us indirectly, leading to any number of weird consequences.

Are you inpainting the eyes at all? Eyes often get screwed up due to autoencoder artifacts. Some models that use a custom autoencoder like AnythingV3,4,5,etc. don’t have this problem, probably because their training data contained lots of images of people with large eyes. You could try training a custom autoencoder, or use a model with one.

Or just inpaint the eyes, and make sure that the inpaint area is set to “only masked”.

They will P-hack the results like usual until a conclusion they favor is reached. See my other comment about why they are doing this to begin with.

Remember the days when it was considered common sense to never share your real life information online?

Those days are going to come back because of this.

Tinfoil hat tiem!

Its all to protect the viability of blackmail. Anyone who has been blackmailed via photography, video, or audio will be able to use AI-generated content to plausibly deny the legitimacy of the blackmail material. Those who rely on blackmail for power, wealth, and influence are incentivized to buy as much time as they can. Blackmailed individuals can be leveraged to buy time in various ways, one of which is on display in the headline of this article. Now that many past social taboos have been normalized, what can you blackmail people with in the current year?

2. Open source was a pretty contentious issue, and a hard one to discuss due to the varying degrees of access that are possible. A system with NDA for top universities/institutions and tiers of access according to how trustworthy they are seemed to be something pretty acceptable.

In case you needed any more convincing to start being a data hoarder, new arguments for it surface every day.

Its not even about protecting children. These people are some of the last people who would care about protecting children.

Now is probably the time to start scraping every model, LoRA, etc. you might want to use just in case.

I should just clarify, I don’t care about alignment whatsoever. I don’t really care if you disagree; it will only hurt you in the long run.

Even if it were copyrightable, AI art is going to become so abundant that it will be too worthless to enforce copyright law over.

Yep :c The enshittification of youtube will continue well into the future.

The only thing I am hopeful about is that youtube is fundamentally not profitable. That ship will have to sink eventually.

It wasn’t a warning, it was a vision. Look up who the Huxley family really are.

heehee

I was trying to make more lolis wearing pantyhose ❤️ but then it gave me this picture instead which i thought was cute c:

Setting the generation dimensions during initial generation to something other than the dimensions the model was trained at can sometimes give you weird results though, like duplicates of what your prompt specifies.

You can always crop generated images to get the dimensions you want. Or do some inpainting tricks if you want to extend an image sideways one way or the other.

Of course, you can upscale smaller images using the Ultimate SD Upscale script. What size would you need for a profile banner though? Higher resolutions need more RAM if the initial upscale preprocessor is used (optional, but it gives better quality.)

Once I get more system RAM, I will make wallpaper versions of these

me too! if i had a living room

Oh I did. I run it in public mode and leave it on whenever I have to go to work, so I can do gens during downtime.

Sometimes my computer disconnects from the internet in the middle of the day though, and I don’t have access to an ethernet connection :c