How feasible is a port to OpenGL 4.5 ?

Is this something a modder group can tackle on their own?

Is this something a modder group could maybe tackle with dev hints and pointers?

Is this something only the devs could accomplish and if so, UBER (@jables ) how likely are you to do this?

Is this something that's plain impossible?

____________________________________________________________________

If we want to jet-propel this game into the future of modding it would be great that PA support more aesthetic features. i believe OpenGL 4.3 supports tessellation and a number of other interesting features.

this would allow not only modders but also devs to do some really cool stuff.

of course I have no intention of keeping the people with OpenGL 3.0 max from playing this would simply be an option that you can set in the settings (requirering a relaunch) that would enable those who have more powerfull graphic cards (AND REALLY NOT THAT MUCH MORE POWERFUL) to benefit from all this.

Modders might (untested) be able to use more advanced shader features, as long as they don't require any engine cooperation. The engine itself is pretty much off limits to anyone except Uber.

This is really off-topic and I'm sorry about that, but I'm having really strange images in my head now... Not unpleasant, mind you.

Why 4.3 and not 4.5 or later? 4.4 added bindless texturing, and 4.5 added even more ways to avoid binding and generally be cleanly atomic (I guess before Khronos gave up and decided to make a new API to ditch binding in favour of direct state manipulation and atomicity once and for all).

OSX currently only supports up to OpenGL 4.1.

you're right. apparently 4.5 is the latest current version but I can't tell what version one could currently code in at grand max.

@jables @mkrater @paleoludic @forrestthewoods

Not possible at all, too much work to maintain two different engines, OSX is slowing us down. Blame Apple

This topic was discussed already many times.

OpenGL versions require hardware and driver support. I don't think there are that many people with OpenGL 4.5 support out there.

There is a program in the bin directory of PA called glewinfo. It will spit out a glewinfo.txt file with all OpenGL versions and all extensions glew knows and could find. These aren't exactly the OpenGL levels the driver may support, but it gives a good hint of what is there.

No.
No.
Yes, and approaching zero percent chance.

And Yes.



Well, okay, no, it's possible for someone who is really persistent to decompile the executable, pick apart the code to figure out the rendering system, write their own executable patch that overwrites it, and replace all of the shaders, but it's not plausible. If they're good enough to do that it'd probably just be easier to write a whole new game engine from scratch that can emulate PA's client behavior and netcode.

But really I'm not really sure what you want to add with OpenGL 4.5 support that you can't do with the current OpenGL 3.1 support.

well I don't know...

I'm no expert could you enlighten me?
is there no benefit?

There's likely some minor performance improvements possible, but there's a lot we could of done still in OpenGL 3.1 that we didn't simply because there wasn't a need. There are other bottlenecks than rendering.

Tessellation is a nice idea, but almost no developers actually use it since it's more efficient to just have a higher poly model when needed and PA's art style is a lot of hard edges.

Image Load Store might let us use a screen buffer or two less, but I think would be slower.

Sparse Textures are a hardware level implementation of virtual texturing, but I don't think it supports the way we use them so wouldn't be useful.

Bindless Textures means some slightly reduced CPU usage, but our game is already fairly stingy on the number of textures used and sorts draw calls in a way to reduce texture binds so it would be negligible.

There's a bunch of other stuff to reduce cpu usage that again our engine and even game visual style wouldn't take advantage of.

@bgolus I'm not really looking at it from that perspective.

I'm looking at it from the point of view of trying to present PA in a beauty contest, if you get the image, where it would look as true to life as possible.

doesn't PA currently have the incapacity to do things such as tessellation, motion blur, high detail models with tons of curved and round surfaces?

I'm really looking at it from a modder trying to change PA's aesthetics's point of view.

Those things exist? :O

They do, but don't expect readable code out of it

I'm more interested in Direct X 12 since it looks like you can get up to 20% performance boost. Not sure how that work or if this is actually true since the demo from Ashes of Singularity about DX12 performance does not convinced me, from that video I can tell for sure that PA performance better than that game.

Well DirectX in PA will not happen because DX only runs on Windows.
On the other hand - do we really need that much more rendering performance in PA?
My computer is four years old and can handle a drawing a few thousand units just fine.
Note that newer DX and OpenGL versions do nothing to improve how fast the simulation actually runs.

@ace63 has already a good answer but if "next-gen" rendering API is required I personally think Vulkan would be the way to go.

Vulkan™ on Intel® Graphics at SIGGRAPH 2015 -

Well, I think bigger planets has to do with better rendering performance, I'm not expert so I might be wrong.