Warn about the high performance cost of JINC2 and xBR filtering
These options (especially xBR) have a very high cost. In my experience, xBR has a greater impact on the framerate than enabling 4× SSAA or PGXP in CPU mode (whose quality increase is much more obvious). These options are also essentially unusable on mobile, regardless of how powerful your device is. As a result, this adds warnings both in the option names and the description. Bilinear filtering is not affected by this performance cost, as GPUs can perform this in hardware without having to emulate filtering through a shader.
This commit is contained in:
@ -886,9 +886,9 @@ static constexpr auto s_texture_filter_names =
|
||||
make_array("Nearest", "Bilinear", "BilinearBinAlpha", "JINC2", "JINC2BinAlpha", "xBR", "xBRBinAlpha");
|
||||
static constexpr auto s_texture_filter_display_names =
|
||||
make_array(TRANSLATABLE("GPUTextureFilter", "Nearest-Neighbor"), TRANSLATABLE("GPUTextureFilter", "Bilinear"),
|
||||
TRANSLATABLE("GPUTextureFilter", "Bilinear (No Edge Blending)"), TRANSLATABLE("GPUTextureFilter", "JINC2"),
|
||||
TRANSLATABLE("GPUTextureFilter", "JINC2 (No Edge Blending)"), TRANSLATABLE("GPUTextureFilter", "xBR"),
|
||||
TRANSLATABLE("GPUTextureFilter", "xBR (No Edge Blending)"));
|
||||
TRANSLATABLE("GPUTextureFilter", "Bilinear (No Edge Blending)"), TRANSLATABLE("GPUTextureFilter", "JINC2 (Slow)"),
|
||||
TRANSLATABLE("GPUTextureFilter", "JINC2 (Slow, No Edge Blending)"), TRANSLATABLE("GPUTextureFilter", "xBR (Very Slow)"),
|
||||
TRANSLATABLE("GPUTextureFilter", "xBR (Very Slow, No Edge Blending)"));
|
||||
|
||||
std::optional<GPUTextureFilter> Settings::ParseTextureFilterName(const char* str)
|
||||
{
|
||||
|
||||
Reference in New Issue
Block a user