First time posting but have been lurking out here reading up on lolcows, degenerates and the like for years off and on now. Also been waiting a long time for the exposure and felting of Quarter Pounder. So, I decided to pull the trigger on creating an account. That way I could contribute to it in my own little way. Which takes me to why I am posting. I've been working with AI locally for some time now on my PC and somehow got sidetracked with everything going on with him from projects I should be working on and ended up doing multiple generations of a song about the gay ops flaggot cuck with trying to get versions of it I like. I have included a few of them here that weren't half bad but overall there is definitely a need for more tweaking of things. The working title is Cuck Flaggot.
I was also thinking of doing an AI generated album cover for it using reference images of him. The problem I ran into is, I couldn't find at least of him where you see the Cornering's pear shaped bulbous body in full view for Flux 2 to have some idea of his shape with doing a quick search on Google for one. I did find a couple of them showing his receding hairline though. When I get time, I will try to find one and maybe generate one.
Also, for those who are interested, everything I use is Free and Open Source Software or licensed in a permissive enough way that it shouldn't be a problem for most use cases average people would use it for. I am actually surprised with seeing people out here using KDE and Kdenlive that someone using Linux or even Windows, or Mac too hasn't been using it on their own computers to meme the shit out of grifting retards like Jer here, not to mention others worthy of threads. I'd be willing to go into more detail if you want to look at using AI yourselves. Just let me know and I can point you in the right direction.
Because AI video gen tools fucking suck, lol. Why do you think his new videos all have extremely simple motion, almost no lip-sync, and really repetitive movements? Because these are the easiest ways to not have the thing sperg the fuck out at you, causing you to have to waste a bunch of tokens on another take to try to get usable footage. The telltale mark of an AI-generated video is a fuckton of cuts, and that's done because you're often getting at best a few frames of useful material from any given prompt, so you have to frankenstein them together. Sometimes you god-roll, but that's one among thousands and thousands.
When it comes to token services and the companies that run them, think of them like DSP and Quarter Pounder here. Like both of them, the AI companies have their pay piggies who are willing to throw large sums of money at them. They also have regular fans that will throw a few bucks their ways as well. The AI companies are no different and I have a theory as to why AI tools in general suck with token based AI services.
It always seems from what I've seen with talking to people who use these services is that the AI companies quietly tweak their AI model along with start limiting the available context and training to it which leads to more bad generations happening. The more of them there are, the more money they make. It is all one big scam.
Don't get me wrong, AI tools in general definitely have room for improvement. As do all of the different models out there for image, video and audio creation. I'll be the first to admit that. The same with there being a human factor involved too with prompts and trying to explain to it exactly what you want. Depending on what you are trying to accomplish, you really have to dumb it down sometimes, almost to the level of Jeremy on occasion.
With doing AI locally, unless you are talking some action movie type stuff, I haven't had much issue with motion. I can also create videos with synced lips too using audio clips with one AI model which is something that I don't think is easy to do with the pay pig token services. I could be wrong where I don't use them. I can even do long form videos as well with another model I use with a workflow that generates multiple consistent clips and then pieces them together into one long video.
It's legitimately news to me that AI-slop videos aren't free to make, and that makes this AI shit feel all the more dystopian
I would say that pretty much all of the AI-Slop that Indian and Eastern European grifters and scammers produce to content farm is made with Free and Open Source Software. They aren't paying for anything beyond electricity to produce their AI slop content, if they are even paying for that where in both regions of the world electricity theft is a big thing.