* always render normal BG/OBJ graphics (even if they're not getting displayed, they can still be fed to the capture unit)

* fix 3D viewport calculation
* keep track of the clearbuffer attributes before VBlank

4 files changed, 41 insertions, 32 deletions

diff --git a/src/GPU2D.cpp b/src/GPU2D.cpp
index 40f1c6e..e1419bf 100644
--- a/src/GPU2D.cpp
+++ b/src/GPU2D.cpp
@@ -368,6 +368,25 @@ void GPU2D::DrawScanline(u32 line)
     u32 dispmode = DispCnt >> 16;
     dispmode &= (Num ? 0x1 : 0x3);
 
+    // always render regular graphics
+    DrawScanline_Mode1(line, dst);
+
+    // capture
+    if ((Num == 0) && (CaptureCnt & (1<<31)))
+    {
+        u32 capwidth, capheight;
+        switch ((CaptureCnt >> 20) & 0x3)
+        {
+        case 0: capwidth = 128; capheight = 128; break;
+        case 1: capwidth = 256; capheight = 64;  break;
+        case 2: capwidth = 256; capheight = 128; break;
+        case 3: capwidth = 256; capheight = 192; break;
+        }
+
+        if (line < capheight)
+            DoCapture(line, capwidth, dst);
+    }
+
     switch (dispmode)
     {
     case 0: // screen off
@@ -377,10 +396,7 @@ void GPU2D::DrawScanline(u32 line)
         }
         break;
 
-    case 1: // regular display
-        {
-            DrawScanline_Mode1(line, dst);
-        }
+    case 1: // regular display, already taken care of
         break;
 
     case 2: // VRAM display
@@ -426,22 +442,6 @@ void GPU2D::DrawScanline(u32 line)
         break;
     }
 
-    // capture
-    if ((!Num) && (CaptureCnt & (1<<31)))
-    {
-        u32 capwidth, capheight;
-        switch ((CaptureCnt >> 20) & 0x3)
-        {
-        case 0: capwidth = 128; capheight = 128; break;
-        case 1: capwidth = 256; capheight = 64;  break;
-        case 2: capwidth = 256; capheight = 128; break;
-        case 3: capwidth = 256; capheight = 192; break;
-        }
-
-        if (line < capheight)
-            DoCapture(line, capwidth, dst);
-    }
-
     // master brightness
     if (dispmode != 0)
     {
diff --git a/src/GPU3D.cpp b/src/GPU3D.cpp
index 41baf34..e06ff90 100644
--- a/src/GPU3D.cpp
+++ b/src/GPU3D.cpp
@@ -57,6 +57,10 @@
 // (the idea is that each position matrix has an associated vector matrix)
 //
 // TODO: check if translate works on the vector matrix? seems pointless
+//
+// viewport Y coordinates are upside-down
+//
+// clear color/depth/bitmap/etc registers (04000350/04000354) are double-buffered
 
 
 namespace GPU3D
@@ -237,6 +241,7 @@ Polygon* RenderPolygonRAM;
 u32 RenderNumPolygons;
 
 u32 ClearAttr1, ClearAttr2;
+u32 RenderClearAttr1, RenderClearAttr2;
 
 u32 FlushRequest;
 u32 FlushAttributes;
@@ -1634,10 +1639,11 @@ void ExecuteCommand()
             break;
 
         case 0x60: // viewport x1,y1,x2,y2
+            // note: viewport Y coordinates are upside-down
             Viewport[0] = ExecParams[0] & 0xFF;
-            Viewport[1] = (ExecParams[0] >> 8) & 0xFF;
+            Viewport[1] = 191 - (ExecParams[0] >> 24);
             Viewport[2] = ((ExecParams[0] >> 16) & 0xFF) - Viewport[0] + 1;
-            Viewport[3] = (ExecParams[0] >> 24) - Viewport[1] + 1;
+            Viewport[3] = (191 - ((ExecParams[0] >> 8) & 0xFF)) - Viewport[1] + 1;
             break;
 
         case 0x70: // box test
@@ -1725,6 +1731,9 @@ void VBlank()
         RenderPolygonRAM = CurPolygonRAM;
         RenderNumPolygons = NumPolygons;
 
+        RenderClearAttr1 = ClearAttr1;
+        RenderClearAttr2 = ClearAttr2;
+
         CurRAMBank = CurRAMBank?0:1;
         CurVertexRAM = &VertexRAM[CurRAMBank ? 6144 : 0];
         CurPolygonRAM = &PolygonRAM[CurRAMBank ? 2048 : 0];
diff --git a/src/GPU3D.h b/src/GPU3D.h
index be121bf..5c29b76 100644
--- a/src/GPU3D.h
+++ b/src/GPU3D.h
@@ -59,7 +59,7 @@ typedef struct
 extern u32 DispCnt;
 extern u32 AlphaRef;
 extern s32 Viewport[4];
-extern u32 ClearAttr1, ClearAttr2;
+extern u32 RenderClearAttr1, RenderClearAttr2;
 
 bool Init();
 void DeInit();
diff --git a/src/GPU3D_Soft.cpp b/src/GPU3D_Soft.cpp
index f15761c..dc94130 100644
--- a/src/GPU3D_Soft.cpp
+++ b/src/GPU3D_Soft.cpp
@@ -781,12 +781,12 @@ void RenderPolygon(Polygon* polygon)
 
 void RenderFrame(Vertex* vertices, Polygon* polygons, int npolys)
 {
-    u32 polyid = (ClearAttr1 >> 24) & 0x3F;
+    u32 polyid = (RenderClearAttr1 >> 24) & 0x3F;
 
     if (DispCnt & (1<<14))
     {
-        u8 xoff = (ClearAttr2 >> 16) & 0xFF;
-        u8 yoff = (ClearAttr2 >> 24) & 0xFF;
+        u8 xoff = (RenderClearAttr2 >> 16) & 0xFF;
+        u8 yoff = (RenderClearAttr2 >> 24) & 0xFF;
 
         for (int y = 0; y < 256*192; y += 256)
         {
@@ -818,16 +818,16 @@ void RenderFrame(Vertex* vertices, Polygon* polygons, int npolys)
     else
     {
         // TODO: confirm color conversion
-        u32 r = (ClearAttr1 << 1) & 0x3E; if (r) r++;
-        u32 g = (ClearAttr1 >> 4) & 0x3E; if (g) g++;
-        u32 b = (ClearAttr1 >> 9) & 0x3E; if (b) b++;
-        u32 a = (ClearAttr1 >> 16) & 0x1F;
+        u32 r = (RenderClearAttr1 << 1) & 0x3E; if (r) r++;
+        u32 g = (RenderClearAttr1 >> 4) & 0x3E; if (g) g++;
+        u32 b = (RenderClearAttr1 >> 9) & 0x3E; if (b) b++;
+        u32 a = (RenderClearAttr1 >> 16) & 0x1F;
         u32 color = r | (g << 8) | (b << 16) | (a << 24);
 
-        u32 z = ((ClearAttr2 & 0x7FFF) * 0x200) + 0x1FF;
+        u32 z = ((RenderClearAttr2 & 0x7FFF) * 0x200) + 0x1FF;
 		if (z >= 0x10000 && z < 0xFFFFFF) z++;
 
-		polyid |= ((ClearAttr1 & 0x8000) >> 7);
+		polyid |= ((RenderClearAttr1 & 0x8000) >> 7);
 
         for (int i = 0; i < 256*192; i++)
         {