/* WARNING, this function is not endian safe! */
static void blit_8i_to_32_ab(__u32 *dst, __u8 *src, __u32 *pal, int width)
{
while (width--)
{
#define BLEND(x, y, a) (y + (((x-y) * a)>>8))
__u32 srccol = pal[*src++];
__u32 dstcol = *dst;
unsigned char sb = srccol & 0xFF;
unsigned char sg = (srccol >> 8) & 0xFF;
unsigned char sr = (srccol >> 16) & 0xFF;
unsigned char sa = (srccol >> 24) & 0xFF;
unsigned char db = dstcol & 0xFF;
unsigned char dg = (dstcol >> 8) & 0xFF;
unsigned char dr = (dstcol >> 16) & 0xFF;
unsigned char da = (dstcol >> 24) & 0xFF;
da = BLEND(0xFF, da, sa) & 0xFF;
dr = BLEND(sr, dr, sa) & 0xFF;
dg = BLEND(sg, dg, sa) & 0xFF;
db = BLEND(sb, db, sa) & 0xFF;
#undef BLEND
*dst++ = db | (dg << 8) | (dr << 16) | (da << 24);
}
}
static void tegra_dc_blend_parallel(struct tegra_dc *dc,
struct tegra_dc_blend *blend)
{
int win_num = dc->gen1_blend_num;
unsigned long mask = BIT(win_num) - 1;
tegra_dc_io_start(dc);
while (mask) {
int idx = get_topmost_window(blend->z, &mask, win_num);
tegra_dc_writel(dc, WINDOW_A_SELECT << idx,
DC_CMD_DISPLAY_WINDOW_HEADER);
tegra_dc_writel(dc, BLEND(NOKEY, FIX, 0xff, 0xff),
DC_WIN_BLEND_NOKEY);
tegra_dc_writel(dc, BLEND(NOKEY, FIX, 0xff, 0xff),
DC_WIN_BLEND_1WIN);
tegra_dc_writel(dc, blend_2win(idx, mask, blend->flags, 0,
win_num), DC_WIN_BLEND_2WIN_X);
tegra_dc_writel(dc, blend_2win(idx, mask, blend->flags, 1,
win_num), DC_WIN_BLEND_2WIN_Y);
tegra_dc_writel(dc, blend_3win(idx, mask, blend->flags,
win_num), DC_WIN_BLEND_3WIN_XY);
}
tegra_dc_io_end(dc);
}
static inline void T_mdp_render_interpolated_scanline_50_cpp(pixel *destScreen, pixel *mdScreen,
int destPitch, int srcPitch,
int width, int height, pixel mask)
{
destPitch /= sizeof(pixel);
srcPitch /= sizeof(pixel);
// TODO: Figure out why the interpolation function is using the line
// below the source line instead of using the current source line.
for (int y = 0; y < height; y++)
{
pixel *SrcLine = &mdScreen[y * srcPitch];
pixel *DstLine1 = &destScreen[(y * 2) * destPitch];
pixel *DstLine2 = &destScreen[((y * 2) + 1) * destPitch];
for (int x = 0; x < width; x++)
{
pixel C = *(SrcLine);
pixel R = *(SrcLine + 1);
pixel D = *(SrcLine + srcPitch);
pixel DR = *(SrcLine + srcPitch + 1);
*DstLine1++ = C;
*DstLine1++ = BLEND(C, R, mask);
*DstLine2++ = (BLEND(C, D, mask) >> 1) & mask;
*DstLine2++ = ((BLEND(BLEND(C, R, mask), BLEND(D, DR, mask), mask)) >> 1) & mask;
SrcLine++;
}
}
}
static inline void
hline32rgba(Imaging im, int x0, int y0, int x1, int ink)
{
int tmp;
unsigned int tmp1, tmp2;
if (y0 >= 0 && y0 < im->ysize) {
if (x0 > x1)
tmp = x0, x0 = x1, x1 = tmp;
if (x0 < 0)
x0 = 0;
else if (x0 >= im->xsize)
return;
if (x1 < 0)
return;
else if (x1 >= im->xsize)
x1 = im->xsize-1;
if (x0 <= x1) {
UINT8* out = (UINT8*) im->image[y0]+x0*4;
UINT8* in = (UINT8*) &ink;
while (x0 <= x1) {
out[0] = BLEND(in[3], out[0], in[0], tmp1, tmp2);
out[1] = BLEND(in[3], out[1], in[1], tmp1, tmp2);
out[2] = BLEND(in[3], out[2], in[2], tmp1, tmp2);
x0++; out += 4;
}
}
}
}
static u32 blend_topwin(u32 flags)
{
if (flags & TEGRA_WIN_FLAG_BLEND_COVERAGE)
return BLEND(NOKEY, ALPHA, 0xff, 0xff);
else if (flags & TEGRA_WIN_FLAG_BLEND_PREMULT)
return BLEND(NOKEY, PREMULT, 0xff, 0xff);
else
return BLEND(NOKEY, FIX, 0xff, 0xff);
}
static inline void
point32rgba(Imaging im, int x, int y, int ink)
{
unsigned int tmp1, tmp2;
if (x >= 0 && x < im->xsize && y >= 0 && y < im->ysize) {
UINT8* out = (UINT8*) im->image[y]+x*4;
UINT8* in = (UINT8*) &ink;
out[0] = BLEND(in[3], out[0], in[0], tmp1, tmp2);
out[1] = BLEND(in[3], out[1], in[1], tmp1, tmp2);
out[2] = BLEND(in[3], out[2], in[2], tmp1, tmp2);
}
}
static u32 blend_2win(int idx, unsigned long behind_mask, u32* flags, int xy)
{
int other;
for (other = 0; other < DC_N_WINDOWS; other++) {
if (other != idx && (xy-- == 0))
break;
}
if (BIT(other) & behind_mask)
return blend_topwin(flags[idx]);
else if (flags[other])
return BLEND(NOKEY, DEPENDANT, 0x00, 0x00);
else
return BLEND(NOKEY, FIX, 0x00, 0x00);
}
static u32 blend_3win(int idx, unsigned long behind_mask, u32* flags)
{
unsigned long infront_mask;
int first;
infront_mask = ~(behind_mask | BIT(idx));
infront_mask &= (BIT(DC_N_WINDOWS) - 1);
first = ffs(infront_mask) - 1;
if (!infront_mask)
return blend_topwin(flags[idx]);
else if (behind_mask && first != -1 && flags[first])
return BLEND(NOKEY, DEPENDANT, 0x00, 0x00);
else
return BLEND(NOKEY, FIX, 0x0, 0x0);
}
void CVisualizerSpectrum::Draw()
{
ASSERT(m_pFftObject != NULL);
static const float SCALING = 250.0f;
static const int OFFSET = 0;
static const float DECAY = 3.0f;
float Step = 0.2f * (float(FFT_POINTS) / float(m_iWidth)) * m_iBarSize; // // //
float Pos = 2; // Add a small offset to remove note on/off actions
int LastStep = 0;
for (int i = 0; i < m_iWidth / m_iBarSize; i++) { // // //
int iStep = int(Pos + 0.5f);
float level = 0;
int steps = (iStep - LastStep) + 1;
for (int j = 0; j < steps; ++j)
level += float(m_pFftObject->GetIntensity(LastStep + j)) / SCALING;
level /= steps;
// linear -> db
level = (20 * logf(level / 4.0f)) * 0.8f;
if (level < 0.0f)
level = 0.0f;
if (level > float(m_iHeight))
level = float(m_iHeight);
if (iStep != LastStep) {
if (level >= m_fFftPoint[iStep])
m_fFftPoint[iStep] = level;
else
m_fFftPoint[iStep] -= DECAY;
if (m_fFftPoint[iStep] < 1.0f)
m_fFftPoint[iStep] = 0.0f;
}
level = m_fFftPoint[iStep];
for (int y = 0; y < m_iHeight; ++y) {
COLORREF Color = BLEND(0x6060FF, 0xFFFFFF, (y * 100) / int(level + 1));
if (y == 0)
Color = DIM(Color, 90);
if (m_iBarSize > 1 && (y & 1)) // // //
Color = DIM(Color, 40);
for (int x = 0; x < m_iBarSize; ++x) { // // //
if (m_iBarSize > 1 && x == m_iBarSize - 1)
Color = DIM(Color, 50);
m_pBlitBuffer[(m_iHeight - y - 1) * m_iWidth + i * m_iBarSize + x + OFFSET] = y < level ? Color : BG_COLOR;
}
}
LastStep = iStep;
Pos += Step;
}
}
static void tegra_dc_set_blending(struct tegra_dc *dc, struct tegra_dc_blend *blend)
{
unsigned long mask = BIT(DC_N_WINDOWS) - 1;
while (mask) {
int idx = get_topmost_window(blend->z, &mask);
tegra_dc_writel(dc, WINDOW_A_SELECT << idx,
DC_CMD_DISPLAY_WINDOW_HEADER);
tegra_dc_writel(dc, BLEND(NOKEY, FIX, 0xff, 0xff),
DC_WIN_BLEND_NOKEY);
tegra_dc_writel(dc, BLEND(NOKEY, FIX, 0xff, 0xff),
DC_WIN_BLEND_1WIN);
tegra_dc_writel(dc, blend_2win(idx, mask, blend->flags, 0),
DC_WIN_BLEND_2WIN_X);
tegra_dc_writel(dc, blend_2win(idx, mask, blend->flags, 1),
DC_WIN_BLEND_2WIN_Y);
tegra_dc_writel(dc, blend_3win(idx, mask, blend->flags),
DC_WIN_BLEND_3WIN_XY);
}
}
static inline void
paste_mask_L(Imaging imOut, Imaging imIn, Imaging imMask,
int dx, int dy, int sx, int sy,
int xsize, int ysize, int pixelsize)
{
/* paste with mode "L" matte */
int x, y;
unsigned int tmp1;
if (imOut->image8) {
for (y = 0; y < ysize; y++) {
UINT8* out = imOut->image8[y+dy]+dx;
UINT8* in = imIn->image8[y+sy]+sx;
UINT8* mask = imMask->image8[y+sy]+sx;
for (x = 0; x < xsize; x++) {
*out = BLEND(*mask, *out, *in, tmp1);
out++, in++, mask++;
}
}
} else {
for (y = 0; y < ysize; y++) {
UINT8* out = (UINT8*) (imOut->image32[y + dy] + dx);
UINT8* in = (UINT8*) (imIn->image32[y + sy] + sx);
UINT8* mask = (UINT8*) (imMask->image8[y+sy] + sx);
for (x = 0; x < xsize; x++) {
UINT8 a = mask[0];
out[0] = BLEND(a, out[0], in[0], tmp1);
out[1] = BLEND(a, out[1], in[1], tmp1);
out[2] = BLEND(a, out[2], in[2], tmp1);
out[3] = BLEND(a, out[3], in[3], tmp1);
out += 4; in += 4; mask ++;
}
}
}
}
static inline void
fill_mask_RGBA(Imaging imOut, const UINT8* ink, Imaging imMask,
int dx, int dy, int sx, int sy,
int xsize, int ysize, int pixelsize)
{
/* fill with mode "RGBA" matte */
int x, y, i;
unsigned int tmp1, tmp2;
if (imOut->image8) {
sx = sx*4+3;
for (y = 0; y < ysize; y++) {
UINT8* out = imOut->image8[y+dy]+dx;
UINT8* mask = (UINT8*) imMask->image[y+sy]+sx;
for (x = 0; x < xsize; x++) {
*out = BLEND(*mask, *out, ink[0], tmp1, tmp2);
out++, mask += 4;
}
}
} else {
dx *= pixelsize;
sx = sx*4 + 3;
for (y = 0; y < ysize; y++) {
UINT8* out = (UINT8*) imOut->image[y+dy]+dx;
UINT8* mask = (UINT8*) imMask->image[y+sy]+sx;
for (x = 0; x < xsize; x++) {
for (i = 0; i < pixelsize; i++) {
*out = BLEND(*mask, *out, ink[i], tmp1, tmp2);
out++;
}
mask += 4;
}
}
}
}
static inline void
paste_mask_L(Imaging imOut, Imaging imIn, Imaging imMask,
int dx, int dy, int sx, int sy,
int xsize, int ysize, int pixelsize)
{
/* paste with mode "L" matte */
int x, y, i;
unsigned int tmp1, tmp2;
if (imOut->image8) {
for (y = 0; y < ysize; y++) {
UINT8* out = imOut->image8[y+dy]+dx;
UINT8* in = imIn->image8[y+sy]+sx;
UINT8* mask = imMask->image8[y+sy]+sx;
for (x = 0; x < xsize; x++) {
*out = BLEND(*mask, *out, *in, tmp1, tmp2);
out++, in++, mask++;
}
}
} else {
for (y = 0; y < ysize; y++) {
UINT8* out = (UINT8*) imOut->image[y+dy]+dx*pixelsize;
UINT8* in = (UINT8*) imIn->image[y+sy]+sx*pixelsize;
UINT8* mask = (UINT8*) imMask->image[y+sy]+sx;
for (x = 0; x < xsize; x++) {
for (i = 0; i < pixelsize; i++) {
*out = BLEND(*mask, *out, *in, tmp1, tmp2);
out++, in++;
}
mask++;
}
}
}
}
BLEND BlendFromString(const char* str)
{
if (!str)
return BLEND_ONE;
for (int i = 1; i <= 19; i++)
{
if (_stricmp(str, STR_BLEND[i - 1]) == 0)
{
return BLEND(i);
}
}
Logger::Log(FB_ERROR_LOG_ARG, FormatString("%s is not valid Blend.", str).c_str());
return BLEND_ONE;
}
void WebPBlendAlpha(WebPPicture* pic, uint32_t background_rgb) {
const int red = (background_rgb >> 16) & 0xff;
const int green = (background_rgb >> 8) & 0xff;
const int blue = (background_rgb >> 0) & 0xff;
int x, y;
if (pic == NULL) return;
if (!pic->use_argb) {
const int uv_width = (pic->width >> 1); // omit last pixel during u/v loop
const int Y0 = VP8RGBToY(red, green, blue, YUV_HALF);
// VP8RGBToU/V expects the u/v values summed over four pixels
const int U0 = VP8RGBToU(4 * red, 4 * green, 4 * blue, 4 * YUV_HALF);
const int V0 = VP8RGBToV(4 * red, 4 * green, 4 * blue, 4 * YUV_HALF);
const int has_alpha = pic->colorspace & WEBP_CSP_ALPHA_BIT;
if (!has_alpha || pic->a == NULL) return; // nothing to do
for (y = 0; y < pic->height; ++y) {
// Luma blending
uint8_t* const y_ptr = pic->y + y * pic->y_stride;
uint8_t* const a_ptr = pic->a + y * pic->a_stride;
for (x = 0; x < pic->width; ++x) {
const int alpha = a_ptr[x];
if (alpha < 0xff) {
y_ptr[x] = BLEND(Y0, y_ptr[x], a_ptr[x]);
}
}
// Chroma blending every even line
if ((y & 1) == 0) {
uint8_t* const u = pic->u + (y >> 1) * pic->uv_stride;
uint8_t* const v = pic->v + (y >> 1) * pic->uv_stride;
uint8_t* const a_ptr2 =
(y + 1 == pic->height) ? a_ptr : a_ptr + pic->a_stride;
for (x = 0; x < uv_width; ++x) {
// Average four alpha values into a single blending weight.
// TODO(skal): might lead to visible contouring. Can we do better?
const int alpha =
a_ptr[2 * x + 0] + a_ptr[2 * x + 1] +
a_ptr2[2 * x + 0] + a_ptr2[2 * x + 1];
u[x] = BLEND_10BIT(U0, u[x], alpha);
v[x] = BLEND_10BIT(V0, v[x], alpha);
}
if (pic->width & 1) { // rightmost pixel
const int alpha = 2 * (a_ptr[2 * x + 0] + a_ptr2[2 * x + 0]);
u[x] = BLEND_10BIT(U0, u[x], alpha);
v[x] = BLEND_10BIT(V0, v[x], alpha);
}
}
memset(a_ptr, 0xff, pic->width);
}
} else {
INLINE void bitmap_4_draw(INT32 firstpix, INT32 iwidth, UINT32 *src, INT32 xpos, UINT8 flags, INT32 dxpos)
{
if (firstpix & 7)
{
UINT32 pixsrc = src[firstpix / 8];
while (firstpix & 7)
{
int pix = (pixsrc >> (4 * (~firstpix & 7))) & 0x0f;
if ((!(flags & 4) || pix) && (UINT32)xpos < 360)
{
if (!(flags & 2))
scanline[xpos] = clutbase[BYTE_XOR_BE(pix)];
else
BLEND(scanline[xpos], clutbase[BYTE_XOR_BE(pix)]);
}
xpos += dxpos;
firstpix++;
}
}
inline static void
_blend_u8_c (guint8 * dest, const guint8 * src,
gint src_stride, gint dest_stride, gint src_width, gint src_height,
gint dest_width, gint b_alpha)
{
gint i, j;
gint src_add = src_stride - src_width;
gint dest_add = dest_stride - dest_width;
for (i = 0; i < src_height; i++) {
for (j = 0; j < src_width; j++) {
*dest = BLEND (*dest, *src, b_alpha);
dest++;
src++;
}
src += src_add;
dest += dest_add;
}
}
inline void jaguar_state::bitmap_4_draw(UINT16 *scanline, INT32 firstpix, INT32 iwidth, UINT32 *src, INT32 xpos, UINT8 flags, INT32 dxpos, UINT16 *clutbase)
{
if (firstpix & 7)
{
UINT32 pixsrc = src[firstpix >> 3];
while (firstpix & 7)
{
int pix = (pixsrc >> ((~firstpix & 7) << 2)) & 0x0f;
if ((!(flags & 4) || pix) && (UINT32)xpos < 760)
{
if (!(flags & 2))
scanline[xpos] = clutbase[BYTE_XOR_BE(pix)];
else
BLEND(scanline[xpos], clutbase[BYTE_XOR_BE(pix)]);
}
xpos += dxpos;
firstpix++;
}
}
void CNoiseEditor::OnPaint()
{
ASSERT(m_pBackDC != NULL);
CPaintDC dc(this);
DrawBackground(m_pBackDC, m_iItems, false, 0);
DrawRange(m_pBackDC, m_iItems, 0);
// Return now if no sequence is selected
if (!m_pSequence) {
PaintBuffer(m_pBackDC, &dc);
return;
}
// Draw items
int Count = m_pSequence->GetItemCount();
if (!Count) {
PaintBuffer(m_pBackDC, &dc);
return;
}
int StepWidth = GetItemWidth();
int StepHeight = m_GraphRect.Height() / m_iItems;
// Draw items
for (int i = 0; i < Count; i++) {
// Draw noise frequency
int item = m_pSequence->GetItem(i) & 0x1F;
int x = m_GraphRect.left + i * StepWidth + 1;
int y = m_GraphRect.top + StepHeight * (m_iItems - item);
int w = StepWidth;
int h = StepHeight;//* item;
if (m_iCurrentPlayPos == i)
DrawPlayRect(m_pBackDC, x, y, w, h);
else
DrawRect(m_pBackDC, x, y, w, h);
// Draw switches
item = m_pSequence->GetItem(i);
int Offset = h * 36 - 1;
if (item & S5B_MODE_SQUARE) {
static const COLORREF BUTTON_COL = COMBINE(0, 160, 160);
int y = Offset;
int h = 9;
m_pBackDC->FillSolidRect(x, y, w, h, BUTTON_COL);
m_pBackDC->Draw3dRect(x, y, w, h, BLEND(BUTTON_COL, 0xFFFFFF, 80), BLEND(BUTTON_COL, 0x000000, 80));
}
else {
static const COLORREF BUTTON_COL = COMBINE(50, 50, 50);
int y = Offset;
int h = 9;
m_pBackDC->FillSolidRect(x, y, w, h, BUTTON_COL);
m_pBackDC->Draw3dRect(x, y, w, h, BLEND(BUTTON_COL, 0xFFFFFF, 80), BLEND(BUTTON_COL, 0x000000, 80));
}
if (item & S5B_MODE_NOISE) {
static const COLORREF BUTTON_COL = COMBINE(160, 0, 160);
int y = Offset + 11;
int h = 9;
m_pBackDC->FillSolidRect(x, y, w, h, BUTTON_COL);
m_pBackDC->Draw3dRect(x, y, w, h, BLEND(BUTTON_COL, 0xFFFFFF, 80), BLEND(BUTTON_COL, 0x000000, 80));
}
else {
static const COLORREF BUTTON_COL = COMBINE(50, 50, 50);
int y = Offset + 11;
int h = 9;
m_pBackDC->FillSolidRect(x, y, w, h, BUTTON_COL);
m_pBackDC->Draw3dRect(x, y, w, h, BLEND(BUTTON_COL, 0xFFFFFF, 80), BLEND(BUTTON_COL, 0x000000, 80));
}
}
DrawLoopPoint(m_pBackDC, StepWidth);
DrawReleasePoint(m_pBackDC, StepWidth);
DrawLine(m_pBackDC);
PaintBuffer(m_pBackDC, &dc);
}
int ShapeWipeMain::process_realtime(VFrame *incoming, VFrame *outgoing)
{
unsigned char *pattern_row;
int col_offset;
unsigned char threshold;
unsigned char value;
int i,j,k;
int opacity;
init_shapes();
load_configuration();
int w = incoming->get_w();
int h = incoming->get_h();
if (strncmp(filename, last_read_filename, BCTEXTLEN) ||
strncmp(shape_name, current_name, BCTEXTLEN) ||
preserve_aspect != last_preserve_aspect)
{
reset_pattern_image();
}
if (!pattern_image)
{
read_pattern_image(w, h);
strncpy(last_read_filename, filename, BCTEXTLEN);
strncpy(current_name, shape_name, BCTEXTLEN);
last_preserve_aspect = preserve_aspect;
}
if (!pattern_image)
{
fprintf(stderr, "Shape Wipe: cannot load shape %s\n", filename);
return 0;
}
if (direction)
{
threshold = (unsigned char)(
(float)PluginClient::get_source_position() /
(float)PluginClient::get_total_len() *
(float)(max_value - min_value))
+ min_value;
}
else
{
threshold = (unsigned char)((max_value - min_value) - (
(float)PluginClient::get_source_position() /
(float)PluginClient::get_total_len() *
(float)(max_value - min_value)))
+ min_value;
}
if (antialias)
{
if (direction)
{
/* Top left corner */
opacity = 0;
COMPARE1(0,0);
COMPARE1(0,1);
COMPARE1(1,0);
COMPARE1(1,1);
BLEND(0,0,4.0);
/* Top edge */
for (k = 1; k < w-1; k++)
{
opacity = 0;
COMPARE1(0,k-1);
COMPARE1(0,k);
COMPARE1(0,k+1);
COMPARE1(1,k-1);
COMPARE1(1,k);
COMPARE1(1,k+1);
BLEND(0,k,6.0);
}
/* Top right corner */
opacity = 0;
COMPARE1(0,w-1);
COMPARE1(0,w-2);
COMPARE1(1,w-1);
COMPARE1(1,w-2);
BLEND(0,w-1,4.0);
/* Left edge */
for (j = 1; j < h-1; j++)
{
opacity = 0;
COMPARE1(j-1,0);
COMPARE1(j,0);
COMPARE1(j+1,0);
COMPARE1(j-1,1);
COMPARE1(j,1);
COMPARE1(j+1,1);
BLEND(j,0,6.0);
}
/* Middle */
for (j = 1; j < h-1; j++)
{
for (k = 1; k < w-1; k++)
{
opacity = 0;
COMPARE1(j-1,k-1);
COMPARE1(j,k-1);
COMPARE1(j+1,k-1);
COMPARE1(j-1,k);
COMPARE1(j,k);
COMPARE1(j+1,k);
COMPARE1(j-1,k+1);
COMPARE1(j,k+1);
COMPARE1(j+1,k+1);
BLEND(j,k,9.0);
}
}
/* Right edge */
for (j = 1; j < h-1; j++)
{
opacity = 0;
COMPARE1(j-1,w-1);
COMPARE1(j,w-1);
COMPARE1(j+1,w-1);
COMPARE1(j-1,w-2);
COMPARE1(j,w-2);
COMPARE1(j+1,w-2);
BLEND(j,w-1,6.0);
}
/* Bottom left corner */
opacity = 0;
COMPARE1(h-1,0);
COMPARE1(h-1,1);
COMPARE1(h-2,0);
COMPARE1(h-2,1);
BLEND(h-1,0,4.0);
/* Bottom edge */
for (k = 1; k < w-1; k++)
{
opacity = 0;
COMPARE1(h-1,k-1);
COMPARE1(h-1,k);
COMPARE1(h-1,k+1);
COMPARE1(h-2,k-1);
COMPARE1(h-2,k);
COMPARE1(h-2,k+1);
BLEND(h-1,k,6.0);
}
/* Bottom right corner */
opacity = 0;
COMPARE1(h-1,w-1);
COMPARE1(h-1,w-2);
COMPARE1(h-2,w-1);
COMPARE1(h-2,w-2);
BLEND(h-1,w-1,4.0);
}
else
{
/* Top left corner */
opacity = 0;
COMPARE2(0,0);
COMPARE2(0,1);
COMPARE2(1,0);
COMPARE2(1,1);
BLEND(0,0,4.0);
/* Top edge */
for (k = 1; k < w-1; k++)
{
opacity = 0;
COMPARE2(0,k-1);
COMPARE2(0,k);
COMPARE2(0,k+1);
COMPARE2(1,k-1);
COMPARE2(1,k);
COMPARE2(1,k+1);
BLEND(0,k,6.0);
}
/* Top right corner */
opacity = 0;
COMPARE2(0,w-1);
COMPARE2(0,w-2);
COMPARE2(1,w-1);
COMPARE2(1,w-2);
BLEND(0,w-1,4.0);
/* Left edge */
for (j = 1; j < h-1; j++)
{
opacity = 0;
COMPARE2(j-1,0);
COMPARE2(j,0);
COMPARE2(j+1,0);
COMPARE2(j-1,1);
COMPARE2(j,1);
COMPARE2(j+1,1);
BLEND(j,0,6.0);
}
/* Middle */
for (j = 1; j < h-1; j++)
{
for (k = 1; k < w-1; k++)
{
opacity = 0;
COMPARE2(j-1,k-1);
COMPARE2(j,k-1);
COMPARE2(j+1,k-1);
COMPARE2(j-1,k);
COMPARE2(j,k);
COMPARE2(j+1,k);
COMPARE2(j-1,k+1);
COMPARE2(j,k+1);
COMPARE2(j+1,k+1);
BLEND(j,k,9.0);
}
}
/* Right edge */
for (j = 1; j < h-1; j++)
{
opacity = 0;
COMPARE2(j-1,w-1);
COMPARE2(j,w-1);
COMPARE2(j+1,w-1);
COMPARE2(j-1,w-2);
COMPARE2(j,w-2);
COMPARE2(j+1,w-2);
BLEND(j,w-1,6.0);
}
/* Bottom left corner */
opacity = 0;
COMPARE2(h-1,0);
COMPARE2(h-1,1);
COMPARE2(h-2,0);
COMPARE2(h-2,1);
BLEND(h-1,0,4.0);
/* Bottom edge */
for (k = 1; k < w-1; k++)
{
opacity = 0;
COMPARE2(h-1,k-1);
COMPARE2(h-1,k);
COMPARE2(h-1,k+1);
COMPARE2(h-2,k-1);
COMPARE2(h-2,k);
COMPARE2(h-2,k+1);
BLEND(h-1,k,6.0);
}
/* Bottom right corner */
opacity = 0;
COMPARE2(h-1,w-1);
COMPARE2(h-1,w-2);
COMPARE2(h-2,w-1);
COMPARE2(h-2,w-2);
BLEND(h-1,w-1,4.0);
}
}
else
{
switch(incoming->get_color_model())
{
case BC_RGB_FLOAT:
SHAPEWIPE(float, 3)
break;
case BC_RGB888:
case BC_YUV888:
SHAPEWIPE(unsigned char, 3)
break;
case BC_RGBA_FLOAT:
SHAPEWIPE(float, 4)
break;
case BC_RGBA8888:
case BC_YUVA8888:
SHAPEWIPE(unsigned char, 4)
break;
case BC_RGB161616:
case BC_YUV161616:
SHAPEWIPE(uint16_t, 3)
break;
case BC_RGBA16161616:
case BC_YUVA16161616:
SHAPEWIPE(uint16_t, 4)
break;
}
}
return 0;
}
void gPixmap::blit(const gPixmap &src, ePoint pos, const gRegion &clip, int flag)
{
for (unsigned int i=0; i<clip.rects.size(); ++i)
{
eRect area=eRect(pos, src.size());
area&=clip.rects[i];
area&=eRect(ePoint(0, 0), size());
if ((area.width()<0) || (area.height()<0))
continue;
eRect srcarea=area;
srcarea.moveBy(-pos.x(), -pos.y());
if ((surface->data_phys && src.surface->data_phys) && (gAccel::getInstance()))
if (!gAccel::getInstance()->blit(surface, src.surface, area.topLeft(), srcarea, flag))
continue;
if ((surface->bpp == 8) && (src.surface->bpp==8))
{
__u8 *srcptr=(__u8*)src.surface->data;
__u8 *dstptr=(__u8*)surface->data;
srcptr+=srcarea.left()*src.surface->bypp+srcarea.top()*src.surface->stride;
dstptr+=area.left()*surface->bypp+area.top()*surface->stride;
for (int y=0; y<area.height(); y++)
{
if (flag & (blitAlphaTest|blitAlphaBlend))
{
// no real alphatest yet
int width=area.width();
unsigned char *src=(unsigned char*)srcptr;
unsigned char *dst=(unsigned char*)dstptr;
// use duff's device here!
while (width--)
{
if (!*src)
{
src++;
dst++;
} else
*dst++=*src++;
}
} else
memcpy(dstptr, srcptr, area.width()*surface->bypp);
srcptr+=src.surface->stride;
dstptr+=surface->stride;
}
} else if ((surface->bpp == 32) && (src.surface->bpp==32))
{
__u32 *srcptr=(__u32*)src.surface->data;
__u32 *dstptr=(__u32*)surface->data;
srcptr+=srcarea.left()+srcarea.top()*src.surface->stride/4;
dstptr+=area.left()+area.top()*surface->stride/4;
for (int y=0; y<area.height(); y++)
{
if (flag & blitAlphaTest)
{
int width=area.width();
unsigned long *src=(unsigned long*)srcptr;
unsigned long *dst=(unsigned long*)dstptr;
while (width--)
{
if (!((*src)&0xFF000000))
{
src++;
dst++;
} else
*dst++=*src++;
}
} else if (flag & blitAlphaBlend)
{
// uh oh. this is only until hardware accel is working.
int width=area.width();
// ARGB color space!
unsigned char *src=(unsigned char*)srcptr;
unsigned char *dst=(unsigned char*)dstptr;
#define BLEND(x, y, a) (y + ((x-y) * a)/256)
while (width--)
{
unsigned char sa = src[3];
unsigned char sr = src[2];
unsigned char sg = src[1];
unsigned char sb = src[0];
unsigned char da = dst[3];
unsigned char dr = dst[2];
unsigned char dg = dst[1];
unsigned char db = dst[0];
dst[3] = BLEND(0xFF, da, sa);
dst[2] = BLEND(sr, dr, sa);
dst[1] = BLEND(sg, dg, sa);
dst[0] = BLEND(sb, db, sa);
#undef BLEND
src += 4; dst += 4;
}
} else
memcpy(dstptr, srcptr, area.width()*surface->bypp);
srcptr+=src.surface->stride/4;
dstptr+=surface->stride/4;
}
} else if ((surface->bpp == 32) && (src.surface->bpp==8))
{
__u8 *srcptr=(__u8*)src.surface->data;
__u8 *dstptr=(__u8*)surface->data; // !!
__u32 pal[256];
for (int i=0; i<256; ++i)
{
if (src.surface->clut.data && (i<src.surface->clut.colors))
pal[i]=(src.surface->clut.data[i].a<<24)|(src.surface->clut.data[i].r<<16)|(src.surface->clut.data[i].g<<8)|(src.surface->clut.data[i].b);
else
pal[i]=0x010101*i;
#if defined(__sh__)
//printf("%s:%d col = %08X\n", __FUNCTION__, __LINE__, pal[i]);
if((pal[i]&0xFF000000) >= 0xE0000000)
pal[i] = 0xFF000000;
#endif
pal[i]^=0xFF000000;
}
srcptr+=srcarea.left()*src.surface->bypp+srcarea.top()*src.surface->stride;
dstptr+=area.left()*surface->bypp+area.top()*surface->stride;
for (int y=0; y<area.height(); y++)
{
int width=area.width();
unsigned char *psrc=(unsigned char*)srcptr;
__u32 *dst=(__u32*)dstptr;
if (flag & blitAlphaTest)
blit_8i_to_32_at(dst, psrc, pal, width);
else if (flag & blitAlphaBlend)
blit_8i_to_32_ab(dst, psrc, pal, width);
else
blit_8i_to_32(dst, psrc, pal, width);
srcptr+=src.surface->stride;
dstptr+=surface->stride;
}
} else
eWarning("cannot blit %dbpp from %dbpp", surface->bpp, src.surface->bpp);
}
}
mlib_status
__mlib_ImageBlendRGBA2ARGB(
mlib_image *dst,
const mlib_image *src)
{
mlib_type type;
mlib_u8 *sl, *dl;
mlib_s32 slb, dlb, nchan, width, height;
mlib_s32 i, j, ii, off;
P_TYPE *sp, *dp;
P_TYPE ss, aa, ds, dd, d_h, d_l;
P_TYPE mzero, const255, mask64, d_half;
MLIB_IMAGE_CHECK(dst);
MLIB_IMAGE_CHECK(src);
MLIB_IMAGE_FULL_EQUAL(dst, src);
MLIB_IMAGE_GET_ALL_PARAMS(dst, type, nchan, width, height, dlb, dl);
slb = mlib_ImageGetStride(src);
sl = mlib_ImageGetData(src);
if (type != MLIB_BYTE || nchan != 4) {
return (MLIB_FAILURE);
}
mzero = _mm_setzero_si128();
const255 = _mm_set1_epi32(0x00ff00ff);
mask64 = _mm_set1_epi32(0xffffff00);
d_half = _mm_set1_epi32(0x00800080);
for (j = 0; j < height; j++) {
P_TYPE alp, a0, a1, ralp, s0, s1, d0, d1, drnd;
mlib_m128 s0u, s1u;
sp = (void *)sl;
dp = (void *)dl;
if (!(((mlib_s32)sp | (mlib_s32)dp) & 15)) {
for (i = 0; i < (width / 4); i++) {
ss = _mm_load_si128(sp);
dd = _mm_load_si128(dp);
s0 = _mm_unpacklo_epi8(ss, mzero);
a0 = _mm_shufflelo_epi16(s0, 0xff);
a0 = _mm_shufflehi_epi16(a0, 0xff);
s0 = _mm_shufflelo_epi16(s0, 0x93);
s0 = _mm_shufflehi_epi16(s0, 0x93);
BLEND(d_h, a0, s0, _mm_unpacklo_epi8(dd,
mzero));
s1 = _mm_unpackhi_epi8(ss, mzero);
a1 = _mm_shufflelo_epi16(s1, 0xff);
a1 = _mm_shufflehi_epi16(a1, 0xff);
s1 = _mm_shufflelo_epi16(s1, 0x93);
s1 = _mm_shufflehi_epi16(s1, 0x93);
BLEND(d_l, a1, s1, _mm_unpackhi_epi8(dd,
mzero));
d_h = _mm_packus_epi16(d_h, d_l);
d_h = _mm_or_si128(_mm_and_si128(mask64, d_h),
_mm_andnot_si128(mask64, dd));
_mm_store_si128(dp, d_h);
sp++;
dp++;
}
} else {
for (i = 0; i < (width / 4); i++) {
#if 0
ss = _mm_loadu_si128(sp);
s0 = _mm_unpacklo_epi8(ss, mzero);
s1 = _mm_unpackhi_epi8(ss, mzero);
#else
s0u.m128d = _mm_load_sd((mlib_d64 *)sp);
s1u.m128d = _mm_load_sd((mlib_d64 *)sp + 1);
s0 = _mm_unpacklo_epi8(s0u.m128i, mzero);
s1 = _mm_unpacklo_epi8(s1u.m128i, mzero);
#endif
dd = _mm_loadu_si128(dp);
a0 = _mm_shufflelo_epi16(s0, 0xff);
a0 = _mm_shufflehi_epi16(a0, 0xff);
s0 = _mm_shufflelo_epi16(s0, 0x93);
s0 = _mm_shufflehi_epi16(s0, 0x93);
BLEND(d_h, a0, s0, _mm_unpacklo_epi8(dd,
mzero));
a1 = _mm_shufflelo_epi16(s1, 0xff);
a1 = _mm_shufflehi_epi16(a1, 0xff);
s1 = _mm_shufflelo_epi16(s1, 0x93);
s1 = _mm_shufflehi_epi16(s1, 0x93);
BLEND(d_l, a1, s1, _mm_unpackhi_epi8(dd,
mzero));
d_h = _mm_packus_epi16(d_h, d_l);
d_h = _mm_or_si128(_mm_and_si128(mask64, d_h),
_mm_andnot_si128(mask64, dd));
#if 1
_mm_storeu_si128(dp, d_h);
#else
s0u.m128i = d_h;
s1u.m128i = _mm_shuffle_epi32(d_h, 0x3e);
_mm_store_sd((mlib_d64 *)dp, s0u.m128d);
_mm_store_sd((mlib_d64 *)dp + 1, s1u.m128d);
#endif
sp++;
dp++;
}
}
if (width & 3) {
s0u.m128d = _mm_load_sd((mlib_d64 *)sp);
s1u.m128d = _mm_load_sd((mlib_d64 *)sp + 1);
s0 = _mm_unpacklo_epi8(s0u.m128i, mzero);
s1 = _mm_unpacklo_epi8(s1u.m128i, mzero);
dd = _mm_loadu_si128(dp);
a0 = _mm_shufflelo_epi16(s0, 0xff);
a0 = _mm_shufflehi_epi16(a0, 0xff);
s0 = _mm_shufflelo_epi16(s0, 0x93);
s0 = _mm_shufflehi_epi16(s0, 0x93);
BLEND(d_h, a0, s0, _mm_unpacklo_epi8(dd, mzero));
a1 = _mm_shufflelo_epi16(s1, 0xff);
a1 = _mm_shufflehi_epi16(a1, 0xff);
s1 = _mm_shufflelo_epi16(s1, 0x93);
s1 = _mm_shufflehi_epi16(s1, 0x93);
BLEND(d_l, a1, s1, _mm_unpackhi_epi8(dd, mzero));
d_h = _mm_packus_epi16(d_h, d_l);
d_h = _mm_or_si128(_mm_and_si128(mask64, d_h),
_mm_andnot_si128(mask64, dd));
for (ii = 0; ii < (width & 3); ii++) {
((mlib_s32 *)dp)[ii] = ((mlib_s32 *)&d_h)[ii];
}
}
sl += slb;
dl += dlb;
}
return (MLIB_SUCCESS);
}
