So if I render them out in linear space, then filter them using AMD cubemap gen and then read these textures in my shader again, would I need to convert them to linear space when reading again ? I guess they won't be stored in linear space, so I'd need to convert them again to linear when reading. But doesn't that seem kind of doubled ?
Sorry if that sounds confusing. What I mean is when I write these environment maps (being in linear space) to my HDD as a *.DDS file, then filter them and read them again in my lighting shader to use, are they already linear or not ? (meaning do I have to convert them to linear space again?).