sc94597 said:
DLSS 3.0 is pretty much defined by optical multi-frame generation. DLSS 3.5 is the ray construction. The fact that Nvidia didn't lock 3.5 to their 4000 series gives more credence to their argument that the OFA in the older cards isn't fast enough for 3.0. If it were purely to sell 4000 series chips they wouldn't backtrack and provide support for 3.5 in the older architectures. Really I think Nvidia should separate the DLSS versions out into different things. DLSS 2.0 should be rebranded DLSS Super Resolution (DLSSSR) & DLAA, DLSS 3.0 should be rebranded DLSS OFG (Optical Frame Generation), DLSS 3.5 should be rebranded DLSS RR (DLSS Ray Reconstruction), and then they could version each one of these as they improve them. |
Feels like Nvidia been working with Nintendo too long lol. The naming convention is so poor that people think its the next version that does everything previous version did and more. They also don't realise you don't have to use all those features, just ones suitable. Like frame interpolation is bad enough in regular video encoding, so games will need work on it for sure. People don't seem to understand that DLSS2.0 basically scales only, therefore it uses less resources then native and it allows the extra resource to actually keep the game steady at say 30fps or 60fps where in a native environment it would not be possible, instead assuming it will be adding extra frames when this will not be the aim for the switch.
Agree with you
Should be as you said and laid out in a common sense way:
DLSS (Parent), with below children you toggle on and off.
- SR 1.0
- OFG 1.0
- RR 1.0
- FUTURE EXTENSION X
- etc