I asked AI to clean up the scene and place it in natural light. This time, it failed miserably. It looks like a cartoon.
Sydney Sweeney did an energetic topless scene on Euphoria (s3e7)
Other Crap
Uncle Scoopy's world-weary musings about naked celebrities, sports, humor and other important, manly things.
I would not want to be the guy who would have to do that sex scene. I would blow a load way too early.
I don’t think the actor is supposed to blow a load at all.
Doesn’t even look like they were ever in the same shot together.
Exactly, they probably weren’t even in the same room. Now, regarding the scene… In this season alone, Sidney Sweeney already has 3 scenes that will certainly be among the best nude scenes of the year in our competition here, lol, just in this season of Euphoria. Of course, I don’t know if Scop will combine these 3 scenes as if they were one, or if he will only include one, etc.
To get that POV shot looking up at Sydney riding… You know that’s the camera operator on the bottom and not the male actor, right? Just like the reverse shot of him on the bed – he’s being ridden by the camera operator. The two actors might not even have been on the set at the same time.
This scene reminded me of those “depraved” sex scenes from some old romantic comedies; it’s much more comical than sexy lol, because of the exaggerated way the actors act, the faces, the absurd dialogue, lol. It was definitely Sam Levinson and Sidney’s intention in this scene to be somewhat comical.
Wow, your previous AI experiment was scary good, but this one is scary bad, or maybe funny bad.
Agree. Either I gave foolish instructions this time, or the damn thing is totally inconsistent. Probably a little of both. I have noticed that the range runs the full gamut from brilliant to awful, but I might be able to eliminate the awful by redoing the results with different instructions. But that is costly and time-consuming, and these models don’t allow feedback, so I normally only use the best stuff and discard the rest. I just posted this one to show how bad it was.
In a perfect world, I could talk to the model like I talk to Chat GPT, where I would say , “That’s an incorrect answer. You need to dig deeper,” or “That guy’s hair is too red.” Unfortunately, ChatGPT, even in its “Pro” version at a pricey $100 a month, will not handle adult material or anything it suspects to be copyrighted. The Chinese models have no such compunctions, indicating to me that the Chinese will eventually win this race by giving consumers more options and avoiding timidity.
That feeds into my theory about what Scoop’s previous AI experiment actually was. I.e., it was not a video created by AI but a previously unknown outtake found and slightly altered by AI. AI is very good at finding things that are already there, even if it has to break into a ‘private’ databank to do it. It can also alter existing things and stitch different existing things together. I am as yet unconvinced that it is currently at the level of creating videos ‘whole cloth’ of entirely realistic-looking humans. I am probably wrong, but I am holding onto my theory until it is proved incorrect.
Wasn’t that experiment on an older film? I wonder if it’s partly a thing of, which older things shot on actual film, the material to brighten is fundamentally there already for it to work with, but with something new shot entirely digital, there’s no more depth for it to work with, nothing THERE for it brighten, and thus it become cartoonish.
These chat models have a parameter called temperature. Its purpose is to make its responses more human-like by varying its takes. Which is a cheap way to fake moods & impulses. The temperature is cranked up to keep the responses from being robotic & repetitive. It’s good enough to fool people. It’s hard to detect their inhumanity. So the inconsistency is by design.
Yep – scary bad is an apt description. Frankly I think that non-AI techniques might be better for brightening a cap like this. Also considering previous successes it could be interesting to try a prompt like “re-create this scene as closely as possible but with better lighting”.