gamers are now realizing that dlss5 uses ai models for post processing … which it did since dlss v1.
Gamers are now realizing that NVidia wants to do more than just upscaling with the newest version of DLSS.
Ironically, DLSS1 models were trained on each game individually to maximize the visual improvement without changing the style and overall feel of the output - exactly what NV originally claimed DLSS5 would do.
DLSS5 was supposed to be an algorithm to enhance a scene’s lighting; it was supposed to be aware of 3d geometry, material properties and not alter an asset’s textures.
But what a surprise, Jensen lied through his teeth. It’s just a dumb screen space image enhancer that isn’t aware of anything except a couple rough motion vectors. It applies its Instagram selfie filter to everything in the scene with zero regard to the devs artistic intent.
And you get all that awesomeness for the tiny price of two 5090s.
So no, gamers are not angry because it took them a decade to realize DLSS is driven by “ai”.
@atimkos
I like Ramprover’s post, but your example is spot-on.
Ironically, DLSS1 models were trained on each game individually to maximize the visual improvement without changing the style and overall feel of the output - exactly what NV originally claimed DLSS5 would do.
But what a surprise, Jensen lied through his teeth. It’s just a dumb screen space image enhancer that isn’t aware of anything except a couple rough motion vectors. It applies its Instagram selfie filter to everything in the scene with zero regard to the devs artistic intent.
And you get all that awesomeness for the tiny price of two 5090s.
I like Ramprover’s post, but your example is spot-on.
Edited
it’s both.
Edited
Wasn’t DLSS more of a frame optimizer rather than an image enhancer?
DLSS5 be like “who ordered some wrinkles?”
Edited