AMD FidelityFX Super Resolution

FSR2, gave me a good bump in fps with very little change in looks. only thing i notice is that power/telegram wires are now twice as thick!?
1440/high vs 1440/high w/FRS(set to quality, default slider) vs 1440/high w/FRS(set to quality, slider upped to 50%)
1663025581103.png
 
So who benefits from making FSR3 open source? Who doesn't care?
Some game dev could be interested in some custom fork of FSR 3 for their own game custom need, but often it is the ability to compile it on anything now and sometime more so in the future
 
Some game dev could be interested in some custom fork of FSR 3 for their own game custom need, but often it is the ability to compile it on anything now and sometime more so in the future
Also bug fixes and optimization for specific games or platforms is now somebody else’s problem as the code is provided as is and all that jazz.
 
Last edited:
Oof…
This response hurts.

Any chance we can have samples for Linux using Vulkan?

We have no plans to add Linux support to our sample framework, but we are fully supportive of anyone that wants to attempt a Cauldron port to non-Windows platforms and will happily answer questions and help out where we can.


I know it’s only their sample code and such but still.
 
Emulator authors? Indie Developers? People who make custom game engines for games? People trying to add this feature into Linux natively?
Most indi devs are using canned engines, engine development is hard. Lots of math degrees required to build a game engine from scratch, but it at least lets EA, and the other big boys work to get it into their engines if they make some, but last I checked they are all buying in on UE5 which already has it.
 
Most indi devs are using canned engines, engine development is hard. Lots of math degrees required to build a game engine from scratch, but it at least lets EA, and the other big boys work to get it into their engines if they make some, but last I checked they are all buying in on UE5 which already has it.
You do not need "lots of math degrees" to make a game engine. We made our own game engine from scratch for our university final project that lasted 5 months, and I wrote the physics engine for it. All it took was a little knowledge in physics and geometry, and a lot of linear algebra.

Now if you want to develop an engine that you want to make money on by licensing to as many developers as possible, like say Unreal Engine, then making one that robust is hard. An engine made for a specific project is not as hard, especially one small in scope like an indie game.
 
Most indi devs are using canned engines, engine development is hard. Lots of math degrees required to build a game engine from scratch, but it at least lets EA, and the other big boys work to get it into their engines if they make some, but last I checked they are all buying in on UE5 which already has it.
I would agree with Armenius, bit of matrix operation, bit of linear algebra, bit of basic newton, maybe bit of differential-integral basic knowledge, to do the indie level game engine the physic simulation is really basic in indie title usually. For a Flight Simulator or if you want fluid, no-Newtonian stuff it would be something else, but more and more there is interesting free to use physic library like bullet or NVIDIA physx your own game engine can use.

Engine development is as hard as what they want them to do, doing an engine that would rival what everything a modern Call of Duty/Battlefield from scratch can do could be virtually impossible for a small team over decade of expertise ended up with something that do an crazy amount of compute every 1/100th of a second, but to make what a small game want to do... that something else completely.

And by making your own game engine in 2023, it can be quite different how fully from scratch and all of it is being made, it can often involve starting with some or a lot of open source library, would it be the modern std C++/boost and other of the like toolkit to using part of a open source physic engine, directX helper libraries, audio engine, database lib, networking one, vulkan-dx abstraction etc... concentrating the game engine on part that are special to the game.

Like by now, people can call making their own game engine starting with something like sdl:
https://github.com/libsdl-org/SDL

A quite successful game like Banished was written from scratch and it was a one guy team I think.

That said you really do not need FSR to be open source for your game engine to use it, specially not on popular platform-C++ version (most popular linux, windows, ios, android, etc...) would AMD build and ships the libs themselves, over time would AMD stop support, if you work on something not popular or in the future if you go back on a project that is now considered old and not supported anymore, etc... being able to compile become quite nice and avoid a lot of potential trouble.
 
Last edited:
So who benefits from making FSR3 open source? Who doesn't care?

Mods can now replace DLSS Frame Generation with FSR3 for NVIDIA RTX 20/30 GPUs in multiple games
https://videocardz.com/newz/mods-ca...3-for-nvidia-rtx-20-30-gpus-in-multiple-games

I was reading Twitter and stumbled upon it.
https://twitter.com/VideoCardz/status/1736725622968156401


View: https://youtu.be/ibGw5PuG4Xc

Its interesting that with an AMD technology going open source-----much of the internet seems to be focused on the benefits to Nvidia's non-RTX 40 GPUs.

IMO, equally notable, is that mods from this should allow AMD cards to have framegen in games which already have DLSS3 framegen, but don't have FSR3. Which is quite a few, at this point.
 
Still possible that headers and pre-built by AMD dlls-.lib would have been enough, and that the FSR code was not needed here.
 
Back
Top