*We have followed through on our plans and made small reductions in the PC installation size over the last few patches while still adding new content. While this was a good start, our short term fixes have not been enough to keep up with all of the new content in the latest patch. The longer term goal has always been to bring the PC installation size much closer in line with the console versions. We are happy to report that, thanks to our partners at Nixxes, we have reached that goal much sooner than expected._
By completely de-duplicating our data, we were able to reduce the PC installation size from ~154GB to ~23GB, for a total saving of ~131GB (~85%). We have completed several rounds of internal QA and are ready to roll this out to early adopters as a public technical beta. Our testing shows that for the small percentage of players still using mechanical hard disk drives, mission loading times have only increased by a few seconds in the worst cases. This is live NOW!*
Fuck fake frames. All my homies use real frames.
They are still better anti aliasing than TAA.
I never said anything about frame generation.
Then what the heck do you think FSR3+ and DLSS use?
You’re way too rude for somebody this unaware of the topic at hand.
FSR and DLSS are at their core temporal upscalers. They take motion vectors, subpixel samples from jittering objects, and a low resolution scene, and using shaders for FSR or AI models for DLSS, interpolate the existing pixels to fill the entire target resolution. That’s it. This is not frame generation, and they don’t use anything, whatever you meant by that.
You can then, on top of the regular upscaling, enable frame generation to enable an entirely different path that holds frames in the buffer and creates intermediary frames. Those are the fake frames you complained about.
One can use both FSR and DLSS without no frame generation whatsoever, and both were originally created without any type of frame generation to begin with. At the present, Helldivers already uses FSR without frame generation - just for upscaling - but it’s FSR 1.0 (previously called FidelityFX), a matrix based spatial scalar that only looks at one central pixel and tries to apply weights to determine how to fill in the neighbors. This looks horrendous. FSR 2.x and onwards, and DLSS, use the full temporal mechanism I described.
That’s “what the heck” I think DLSS does.