That sounds super cool

Thats super cool, although I'll probably never bother given that image edit models exist

Nah, this isn't scary at all. 😂😶‍🌫️

Wow. Mystifying and chilling all at once. 😶‍🌫️

I think it's like this one that unlocks half speed tensor cores on the GPU.

What exactly is it unlocking?

Sorry about that. I mistook you for someone else.

I disagree with this moderation. Every realistic video in this demo still has downright silly errors, and however far it is from gloopy Wool Smoot taking bites out of his ramen scarf, the goofs are hard to miss. Pointing that out in a week-old thread with zero comments is not the disingenuous verbal abuse or outright bullying that defines trolling.

If you're pushing any model to the least compute it will tolerate, you're gonna get some amusing shortcomings.

Video models have come a long way since people joked about AI models being unable to animate Will Smith Eating Spaghetti!

They still can't, if the two-forks video at top of page is any indication. And the joke's been going long enough that hallucinated Will Smith looks old.

Pretty cool to have you here. Feel free to post any updates.

Yes, it's absolutely recoverable if you know your way around the OS, but now that Linux is getting more popular I suspect there are an increasing number of people who wouldn't know how. The reason I didn't write the exact commands I had to run to activate the unlocker was that I wanted to reduce the risk of someone copy-pasting commands they don't understand into a terminal, and then being unable to repair their install if it stops booting.

The install procedure does involve changing some kernel boot parameters, so if you make a mistake it's possible that you soft-brick your OS.

I mean...just interrupt the boot process at GRUB, and edit the kernel boot parameters to whatever you want, boot into the OS, and then change them back persistently.

Why tie it to a model? Just have the model as a parameter or config option.

Also, it didn't even dawn on me how easy it would be to use a MCP to tie together the main text-based LLM to a pure text-to-image generator.

"Show me some titties."

"Bigger."

"Bigger!"

"Do you know what 'big' means motherfucker???“

Kel.jpeg.

Lol!!! I might install it on an old laptop just to see if it sounds like a jet engine hahaha

Dunno, but I just realized it needs a chromium-based browser.

What are the systems spec requirements?

Looks pretty cool, though I'm guessing this takes way more vram than I have. What I always wonder about this kind of tech is what ways of making it more game-like would actually work, it's hard to think about because it's unclear what the limitations are

cocktailpeanut/breadboard.me was a good tool for this but it seems to have been abandoned as quickly as it was made.

thank you

Flux2 Klein 9B works really well for this.

Ask it to preserve all facial features and hair and only change the expression to what you want.

I've heard some video models can be used as an edit model to do this. You might also want to check Krea 2.

I should have named this sub something else. Stable Diffusion models are so dead rn :D

oof, windows+nvidia only.