When I went team red for the first time earlier this year, I really scrutinized zoomed in screenshots to compare the upscaling for FSR and DLSS. With FSR 3, I couldn’t see any difference compared to DLSS. Older FSR versions yeah, but at least for me not a problem any more.

I think the size may have more to do with the team not splitting up the Unreal build paks. Haven’t checked how well it’s actually split up, but I can say that changing even one small thing could result in a giant update if that build has like one pak file with all the things in it. There are ways to configure it in the build but it’s not a magic toggle either. Worked with a studio handing off UE builds before that didn’t build the game in a split friendly way and it made every upload to S3 take forever cause there were only like two really giant paks.

Also makes me wonder, does Steam not do diff patch style updates for changes within individual files? If not, that could save a ton of bandwidth.

Idk, seems he’s salty about losing so much money because he acted creepy and unprofessional towards women. Could go either way, curious how this plays out.

To anyone who didn’t read past the headline: he said he needed more sleep. He was testing it obsessively.

Oh yeah, I read about that! Not really ML, but pretty much what I’d like more games to have.

Oh that’s really interesting; I hadn’t considered racing games as a genre to benefit from this type of machine learning. I guess I figured there’s not so much to AI there that it’s necessary, at least when we already know the “ideal lap line” for cars to follow, but yeah it gets a lot harder when considering other drivers on the track and a huge array of unique car models with their own handling and performance characteristics.

I wouldn’t mind an AI using unorthodox strategies, but yeah that’s a good point that fine tuning it to be fun is a big challenge. Speaking of “non-player-like behavior”, I wonder if AI could be used to find multiplayer exploits sooner, though the problem there is you don’t really have much training data besides QA and playtesters before a full release.

I’m not into fighting games, but that’s pretty neat! I hope the industry follows suit if people like how it works in Street Fighter 6.

Yep typical ratings are oriented relative to the “average” consumer. But no person is actually THE average.

It’s a useful place to find out if something totally sucks though. That’s how I use it. 60+? Probably good, at least for some audiences. Less than that? Only if you’re already hyped or a fan of whatever thing it’s related to.

Slappers only, no oddjob

And this will lead to games with more animations and better animations, not a loss of animation jobs. Might lower the skill ceiling a bit, but there are always going to be bespoke animations requiring much more manual work, especially for non-humanoids and objects. Animators will be able to focus their time more broadly.

All i read is “Damn, they’re a super capable team.”

It’s a joke, dummy