DLSS vs. FSR vs. XeSS

[sc94597]

JRPGfan said:

sc94597 said:

I mean, Nvidia is now an AI research and development company before it is anything else. It's one of the few companies that is a powerhouse in fundamental research (probably only bested by Google Deep-mind, although they specialize in different domains.) They're going to sustain their advantage in DLSS and neural-rendering for the foreseeable future. AMD is finally catching up to be somewhat competitive, and Intel has been more or less close to AMD (up until recently ahead of them, now behind.) 

Your thinking of old FSR back before they had it based on AI models, from when it was just algorithms. 
Ei. FSR2 and 3.... back then, what you wrote was true.

Nowadays its not.

Its like DLSS transformer model > FSR 4 >>>>>>>>>>>>>>>> XeSS (intel).

DLSS and FSR are like neck and neck, and intel is far behind.

I am actually not thinking of FSR 1-3, but rather how each company has handled their DL suites in general. There is much more to the DLSS suite and Nvidia's plans for deep-learning assisted graphics than what we have currently, and it's only recently (a week ago) that AMD has caught up in parity with what Nvidia has had for years. AMD's answer to ray reconstruction - a feature Nvidia has had for three years (since the Lovelace launch) - just released a week ago, as an example.  

If you read Nvidia's white-papers you basically get a preview of what they will release with each DLSS iteration.

https://research.nvidia.com/publications