No that would slightly defeat the purpose of generating them. No reason that you can't use a pool of source textures with splatting and colourisation, then another pool of detail textures on top of that perhaps to further improve uniqueness. Or something.

'Nuff said....

Procedural yes, static no. I've even said this to you Coolhand on more than one occasion 😉

Having just read a couple of the other threads I can sort of see why. There's already 2 to 4 people testing the game on hardware without decent shader support, mostly laptops using those f@#king Intel chipsets, so whilst taking all of the terrain generation and rendering over to shaders would be fricking sweet... it would stop people from playing 🙁

The alternative is to maintain separate shader & fixed function pipeline (FFP) systems, which takes more developer time and you might even see gameplay diverging between them. For example, I'd be quite interested in porting the terrain generation & rendering to the GPU, including updating the physics to give more precise terrain collision for landings etc. Now that's not going to work on machines using the FFP, so they'll get the CPU based system we currently have which will mean not only will planets look different but because of the different generation ability / features of CPU vs GPU you'll actually get different terrains. They'd probably be broadly similar, but I do mean "broadly".

I have spent most of today reacquainting myself with texture splatting and I might go onto that after I've done my current... thing

Yeah but I think that was Durandal 😉

Oh yeah 🙂

But I was referring to collhand's message just up there.. *points up* 🙂

But anyway... 🙂 Have you had any thoughts on how you might do this?

Thats a shame. Is there a way around that? Like how system and secotr/galaxy generation was all moved from floats and doubles to int32 to get around precision issues between the platforms..?

I wasn't thinking of it as a technical limitation in the same way but more of a sheer processing power difference. Without trying it my best guess is that you'd still be able to do the same thing, land in the same place, but that you might use algorithms that are better suited to GPU so the results themselves will be superficially different.

I.e; we could all land in mexico, but those with GPU would all look the same, and those with CPU would all look the same, but GPU and CPU generated landscapes would be different in more than just the amount of detail.

Without trying to implement it I couldn't begin to guess. Perhaps the CPU system will translate perfectly to GPU and this speculation is worthless after all 😀

As for texture splatting, yes I've had a few ideas and done some reading, it's possible to do it via FFP too. Not as well, but still possible.