FSR4 has been great for me. Performance seems slightly worse than FSR3 but the increase in visual quality is well worth it, particularly because IMO FSR4 on balanced both looks and performs better than FSR3 on quality.
I still avoid it when I can because native rendering will always look best, but this is definitely really nice for steam deck where the hardware is much more limited.
On average there are 4.3 births per second, I’m not sure how that compares to in 1970 but its probably still more than 1 so its quite likely someone was born exactly on unix time 0.
A game made for kids should not allow sexually explicit material on their platform. And, in fact, they don’t, this kind of content is very clearly and explicitly against roblox’s content policy but they fail to enforce it.
This isn’t an age verification problem, its a moderation problem. A massive one. There’s hundreds of games that exist almost entirely for the purpose of grooming ppl (a common one is so called “bathroom simulators” iirc) and roblox frequently doesnt do anything about them until there’s public outcry about ine particular game, no matter how many reports come in or how obvious it is.
True, but I dont want reddit’s shitty NSFW tagging system to distract from the important issue here
It does need better tagging but honestly thats not the issue here. This is because of the new UK law, simple as.
deleted by creator
Yes. People seem to be saying this is not a real video and that the white house generated it with AI. That is what I would say it means to say the video “is AI” - particularly because an AI interpolated video is still a real video.
Whether the interpolation is done with AI or not, I don’t know. AI interpolation is fairly common but there are older methods that produce similar artifacts. Based on some of the interpolated frames I’m leaning slightly towards it being AI interpolation but I don’t know well enough to say.
See some of my later replies. I’ve looked at it frame by frame and it looks exactly like the artifacts you’d see from motion interpolation. They can look very similar to AI artifacts.
I know. But that can totally happen in interpolated frames. Look at the cat video on the Wikipedia page. There are moments when the toy dissapears, when the kittens head splits into two, when one arm morphs into another… stuff like this is totally possible and even common for motion interpolation.
I’m looking at it more and I’m fairly certain this is just interpolated.
Look at this frame:
Seems to me to be pretty normal. No weird hand stuff, fingers look fine.
Now look at the next two frames, particularly the second one:
Totally jank. Fingers are blotchy, there’s a ghost finger, overall looks super fake. But then look at the next two frames:
Second to last has a little bit of weirdness still, but by the last frame it looks totally legit and real. No blotchy fingers, no ghost fingers, overall looks very normal.
That kind of behavior - where some frames look completely fine and in between ones look blotchy and weird - is exactly what you expect from motion interpolation.
Look at the example on a cat video on wikepedia, for instance: https://en.wikipedia.org/wiki/Motion_interpolation
The framerate is much lower on the original video here, but you can see a lot of the same kinds of artifacts. Blotchy limbs, “ghosts” in movment, things morphing in ways they shouldn’t… its very similar to what’s seen in the trump video. And mind you, to my knowledge this video is done with older interpolation methods that had nothing to do with genAI.
I know, im seeing what you mean. I’m saying that could very much be an issue with interpolation. Especially if its AI interpolation. I edited my comment to add a bit more detail but its possible that over half the frames of this video are AI generated while the video itself is still real. Obviously I cant say for sure that that’s what is happening but this very much could be an interpolation artifact. Even in non-genAI interpolation, weird smudginess and morphing of fast moving objects in interpolated frames is a common and well known issue.
I think this may be just interpolation weirdness. Apparently the video was recorded in 24fps but encoded in 60.
See this analysis, I posted it in another comment: https://youtube.com/shorts/hXhDB1fK7ok
Edit: worth noting the interpolarion may be done by a similar AI engine which explains why it looks so similar to AI generated footage. Similar technology as the motion smoothness on TVs or DLSS frame gen. If it truly is 24-60fps and the interpolation is done by AI this would indicate over half the frames are AI generated. Would explain a lot while not necessarily meaning the video as a whole is AI.
I disagree. I recently saw this analysis:
https://youtube.com/shorts/hXhDB1fK7ok
It seems pretty thorough, and this guy specializes in exactly this. Some of the weirdness seems to be due to a miss match between the 24fps the video was filmed in and the 60fps it was encoded in.
Thats why I said “come close” - there are plenty of crimes committed by the ultra wealthy.
However, those aren’t the kind of crimes this kind of mass surveillance is targeting either. They are trying to get rid of petty crime, gang violence, theft… stuff like that. And those kinds of crimes would almost dissapear entirely if you eliminated poverty.
The only way you could actually come close to eliminating all crime would be if you eliminated poverty. But that would make the rich less rich, so not gonna happen.
There was debate around whether they should use lidar. I thought I remembered that the initial plan was lidar but that musk dropped it before any actually got into the hands of consumers. I could be wrong though
This is ancient news, but I still cant believe they ditched lidar. Relying in computer vision alone when you could easily give it more reliable data to work with is just stupid.
Obviously this is subjective, but I use android auto all the time and something similar for a linux phone would be really nice for me. Don’t dismiss them just cause you wouldn’t use them
DNS request will go to the DNS server but from what I can tell the packet wont go anywhere else. At least on my own network, traceroute gives only one entry and its the final destination. I assume this is because the router knows its own public IP and as such knows the packet doesnt need to be sent anywhere else.