fixup! fixed the deswizzling logic. Patch by @yuriks
This commit is contained in:
parent
1e5a32fce3
commit
3b6c037ba2
2 changed files with 58 additions and 20 deletions
|
@ -849,21 +849,25 @@ void RasterizerOpenGL::ReloadColorBuffer() {
|
|||
std::unique_ptr<u8[]> temp_fb_color_buffer(new u8[fb_color_texture.width * fb_color_texture.height * bytes_per_pixel]);
|
||||
|
||||
// Directly copy pixels. The OpenGL internal formats match the 3DS ones, so no conversion is necessary.
|
||||
using VideoCore::CopyTextureAndUntile;
|
||||
switch (bytes_per_pixel) {
|
||||
case 4:
|
||||
VideoCore::CopyTextureAndTile<u32>((u32*)temp_fb_color_buffer.get(), (u32*)color_buffer, fb_color_texture.width, fb_color_texture.height);
|
||||
CopyTextureAndUntile<u32>(reinterpret_cast<u32*>(temp_fb_color_buffer.get()),
|
||||
reinterpret_cast<u32*>(color_buffer),
|
||||
fb_color_texture.width, fb_color_texture.height);
|
||||
break;
|
||||
case 3:
|
||||
VideoCore::CopyTextureAndTile<u24_be>((u24_be*)temp_fb_color_buffer.get(), (u24_be*)color_buffer, fb_color_texture.width, fb_color_texture.height);
|
||||
CopyTextureAndUntile<u24_be>(reinterpret_cast<u24_be*>(temp_fb_color_buffer.get()),
|
||||
reinterpret_cast<u24_be*>(color_buffer),
|
||||
fb_color_texture.width, fb_color_texture.height);
|
||||
break;
|
||||
case 2:
|
||||
VideoCore::CopyTextureAndTile<u16>((u16*)temp_fb_color_buffer.get(), (u16*)color_buffer, fb_color_texture.width, fb_color_texture.height);
|
||||
break;
|
||||
case 1:
|
||||
VideoCore::CopyTextureAndTile<u8>(temp_fb_color_buffer.get(), color_buffer, fb_color_texture.width, fb_color_texture.height);
|
||||
CopyTextureAndUntile<u16>(reinterpret_cast<u16*>(temp_fb_color_buffer.get()),
|
||||
reinterpret_cast<u16*>(color_buffer),
|
||||
fb_color_texture.width, fb_color_texture.height);
|
||||
break;
|
||||
default:
|
||||
LOG_ERROR(Render_OpenGL, "Unimplemented pixel size %u bytes per pixel", bytes_per_pixel);
|
||||
UNREACHABLE();
|
||||
}
|
||||
|
||||
state.texture_units[0].texture_2d = fb_color_texture.texture.handle;
|
||||
|
@ -968,21 +972,25 @@ void RasterizerOpenGL::CommitColorBuffer() {
|
|||
state.Apply();
|
||||
|
||||
// Directly copy pixels. The OpenGL internal formats match the 3DS ones, so no conversion is necessary.
|
||||
using VideoCore::CopyTextureAndTile;
|
||||
switch (bytes_per_pixel) {
|
||||
case 4:
|
||||
VideoCore::CopyTextureAndTile<u32>((u32*)color_buffer, (u32*)temp_gl_color_buffer.get(), fb_color_texture.width, fb_color_texture.height);
|
||||
CopyTextureAndTile<u32>(reinterpret_cast<u32*>(color_buffer),
|
||||
reinterpret_cast<u32*>(temp_gl_color_buffer.get()),
|
||||
fb_color_texture.width, fb_color_texture.height);
|
||||
break;
|
||||
case 3:
|
||||
VideoCore::CopyTextureAndTile<u24_be>((u24_be*)color_buffer, (u24_be*)temp_gl_color_buffer.get(), fb_color_texture.width, fb_color_texture.height);
|
||||
CopyTextureAndTile<u24_be>(reinterpret_cast<u24_be*>(color_buffer),
|
||||
reinterpret_cast<u24_be*>(temp_gl_color_buffer.get()),
|
||||
fb_color_texture.width, fb_color_texture.height);
|
||||
break;
|
||||
case 2:
|
||||
VideoCore::CopyTextureAndTile<u16>((u16*)color_buffer, (u16*)temp_gl_color_buffer.get(), fb_color_texture.width, fb_color_texture.height);
|
||||
break;
|
||||
case 1:
|
||||
VideoCore::CopyTextureAndTile<u8>(color_buffer, temp_gl_color_buffer.get(), fb_color_texture.width, fb_color_texture.height);
|
||||
CopyTextureAndTile<u16>(reinterpret_cast<u16*>(color_buffer),
|
||||
reinterpret_cast<u16*>(temp_gl_color_buffer.get()),
|
||||
fb_color_texture.width, fb_color_texture.height);
|
||||
break;
|
||||
default:
|
||||
LOG_ERROR(Render_OpenGL, "Unimplemented pixel size %u bytes per pixel", bytes_per_pixel);
|
||||
UNREACHABLE();
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
@ -37,9 +37,9 @@ void DumpTGA(std::string filename, short width, short height, u8* raw_data);
|
|||
|
||||
/// Lookup table for the offsets used to convert an image to Morton order.
|
||||
static const u8 morton_lut[64] = {
|
||||
0, 1, 4, 5, 16, 17, 20, 21,
|
||||
2, 3, 6, 7, 18, 19, 22, 23,
|
||||
8, 9, 12, 13, 24, 25, 28, 29,
|
||||
0, 1, 4, 5, 16, 17, 20, 21,
|
||||
2, 3, 6, 7, 18, 19, 22, 23,
|
||||
8, 9, 12, 13, 24, 25, 28, 29,
|
||||
10, 11, 14, 15, 26, 27, 30, 31,
|
||||
32, 33, 36, 37, 48, 49, 52, 53,
|
||||
34, 35, 38, 39, 50, 51, 54, 55,
|
||||
|
@ -57,13 +57,15 @@ static inline u32 MortonInterleave(u32 x, u32 y) {
|
|||
}
|
||||
|
||||
/**
|
||||
* Copies the texture data from the source address to the destination address,
|
||||
* applying a Morton-order transformation while copying.
|
||||
* Copies the texture data from the source address to the destination address, applying a
|
||||
* Morton-order transformation while copying.
|
||||
*
|
||||
* @param T Type of the source and destination pointers, the swizzling process depends on the size
|
||||
* of this parameter.
|
||||
* @param dst Pointer to which the texture will be copied.
|
||||
* @param src Pointer to the source texture data.
|
||||
* @param width Width of the texture, should be a multiple of 8.
|
||||
* @param height Height of the texture, should be a multiple of 8.
|
||||
* @param T Type of the source and destination pointers, the swizzling process depends on the size of this parameter.
|
||||
*/
|
||||
template<typename T>
|
||||
static inline void CopyTextureAndTile(T* dst, const T* src, unsigned int width, unsigned int height) {
|
||||
|
@ -83,6 +85,34 @@ static inline void CopyTextureAndTile(T* dst, const T* src, unsigned int width,
|
|||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Copies texture data while undoing the transformation applied by `CopyTextureAndTile`.
|
||||
*
|
||||
* @param T Type of the source and destination pointers, the swizzling process depends on the size
|
||||
* of this parameter.
|
||||
* @param dst Pointer to which the texture will be copied.
|
||||
* @param src Pointer to the source texture data.
|
||||
* @param width Width of the texture, should be a multiple of 8.
|
||||
* @param height Height of the texture, should be a multiple of 8.
|
||||
*/
|
||||
template<typename T>
|
||||
static inline void CopyTextureAndUntile(T* dst, const T* src, unsigned int width, unsigned int height) {
|
||||
for (unsigned int y = 0; y + 8 <= height; y += 8) {
|
||||
for (unsigned int x = 0; x + 8 <= width; x += 8) {
|
||||
T* line = &dst[y * width + x];
|
||||
|
||||
for (unsigned int yy = 0; yy < 8; ++yy) {
|
||||
for (unsigned int xx = 0; xx < 8; ++xx) {
|
||||
line[xx] = src[morton_lut[yy * 8 + xx]];
|
||||
}
|
||||
line += width;
|
||||
}
|
||||
|
||||
src += 8 * 8;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Calculates the offset of the position of the pixel in Morton order
|
||||
*/
|
||||
|
|
Loading…
Reference in a new issue