diff --git a/CMakeLists.txt b/CMakeLists.txt
index c1860a2b53..fcf74005e7 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -1906,6 +1906,8 @@ set(GPU_SOURCES
 	GPU/Common/Draw2D.cpp
 	GPU/Common/Draw2D.h
 	GPU/Common/DepthBufferCommon.cpp
+	GPU/Common/DepthRaster.cpp
+	GPU/Common/DepthRaster.h
 	GPU/Common/TextureShaderCommon.cpp
 	GPU/Common/TextureShaderCommon.h
 	GPU/Common/DepalettizeShaderCommon.cpp
diff --git a/Common/Math/CrossSIMD.h b/Common/Math/CrossSIMD.h
index 6ad03b8322..94b2d3933b 100644
--- a/Common/Math/CrossSIMD.h
+++ b/Common/Math/CrossSIMD.h
@@ -5,4 +5,3 @@
 #pragma once
 
 #include "Common/Math/SIMDHeaders.h"
-
diff --git a/Core/Compatibility.cpp b/Core/Compatibility.cpp
index 3790421980..cd2c50c488 100644
--- a/Core/Compatibility.cpp
+++ b/Core/Compatibility.cpp
@@ -149,6 +149,7 @@ void Compatibility::CheckSettings(IniFile &iniFile, const std::string &gameID) {
 	CheckSetting(iniFile, gameID, "DisableMemcpySlicing", &flags_.DisableMemcpySlicing);
 	CheckSetting(iniFile, gameID, "ForceEnableGPUReadback", &flags_.ForceEnableGPUReadback);
 	CheckSetting(iniFile, gameID, "UseFFMPEGFindStreamInfo", &flags_.UseFFMPEGFindStreamInfo);
+	CheckSetting(iniFile, gameID, "SoftwareRasterDepth", &flags_.SoftwareRasterDepth);
 }
 
 void Compatibility::CheckVRSettings(IniFile &iniFile, const std::string &gameID) {
diff --git a/Core/Compatibility.h b/Core/Compatibility.h
index 8a0e33af4d..4688df37c0 100644
--- a/Core/Compatibility.h
+++ b/Core/Compatibility.h
@@ -112,6 +112,7 @@ struct CompatFlags {
 	bool DisableMemcpySlicing;
 	bool ForceEnableGPUReadback;
 	bool UseFFMPEGFindStreamInfo;
+	bool SoftwareRasterDepth;
 };
 
 struct VRCompat {
diff --git a/GPU/Common/DepthRaster.cpp b/GPU/Common/DepthRaster.cpp
new file mode 100644
index 0000000000..20ac069f82
--- /dev/null
+++ b/GPU/Common/DepthRaster.cpp
@@ -0,0 +1,367 @@
+#include <algorithm>
+
+#include "Common/Math/CrossSIMD.h"
+#include "GPU/Common/DepthRaster.h"
+#include "GPU/Math3D.h"
+#include "Common/Math/math_util.h"
+#include "GPU/Common/VertexDecoderCommon.h"
+
+// TODO: Should respect the scissor rect.
+
+struct ScreenVert {
+	int x;
+	int y;
+	uint16_t z;
+	uint16_t behind;
+};
+
+void DepthRasterRect(uint16_t *dest, int stride, int x1, int y1, int x2, int y2, short depthValue, GEComparison depthCompare) {
+	// Swap coordinates if needed, we don't back-face-cull rects.
+	// We also ignore the UV rotation here.
+	if (x1 > x2) {
+		std::swap(x1, x2);
+	}
+	if (y1 > y2) {
+		std::swap(y1, y2);
+	}
+	if (x1 == x2 || y1 == y2) {
+		return;
+	}
+
+	__m128i valueX8 = _mm_set1_epi16(depthValue);
+
+#if PPSSPP_ARCH(SSE2)
+	for (int y = y1; y < y2; y++) {
+		__m128i *ptr = (__m128i *)(dest + stride * y + x1);
+		int w = x2 - x1;
+
+		switch (depthCompare) {
+		case GE_COMP_ALWAYS:
+			while (w >= 8) {
+				_mm_storeu_si128(ptr, valueX8);
+				ptr++;
+				w -= 8;
+			}
+			break;
+			// TODO: Trailer
+		case GE_COMP_NEVER:
+			break;
+		default:
+			// TODO
+			break;
+		}
+	}
+#elif PPSSPP_ARCH(ARM64_NEON)
+
+#else
+	// Do nothing for now
+#endif
+}
+
+using namespace Math3D;
+struct int2 {
+	int x, y;
+	int2(float a, float b) {
+		x = (int)(a + 0.5f);
+		y = (int)(b + 0.5f);
+	}
+};
+
+// Adapted from Intel's depth rasterizer example.
+// Started with the scalar version, will SIMD-ify later.
+void DepthRasterTriangle(uint16_t *depthBuf, int stride, int x1, int y1, int x2, int y2, const ScreenVert vertsSub[3], GEComparison compareMode) {
+	int tileStartX = x1;
+	int tileEndX = x2;
+
+	int tileStartY = y1;
+	int tileEndY = y2;
+
+	// Convert to whole pixels for now. Later subpixel precision.
+	ScreenVert verts[3];
+	verts[0].x = vertsSub[0].x >> 4;
+	verts[0].y = vertsSub[0].y >> 4;
+	verts[0].z = vertsSub[0].z;
+	verts[1].x = vertsSub[2].x >> 4;
+	verts[1].y = vertsSub[2].y >> 4;
+	verts[1].z = vertsSub[2].z;
+	verts[2].x = vertsSub[1].x >> 4;
+	verts[2].y = vertsSub[1].y >> 4;
+	verts[2].z = vertsSub[1].z;
+
+	// use fixed-point only for X and Y.  Avoid work for Z and W.
+	int startX = std::max(std::min(std::min(verts[0].x, verts[1].x), verts[2].x), tileStartX) & int(0xFFFFFFFE);
+	int endX = std::min(std::max(std::max(verts[0].x, verts[1].x), verts[2].x) + 1, tileEndX);
+
+	int startY = std::max(std::min(std::min(verts[0].y, verts[1].y), verts[2].y), tileStartY) & int(0xFFFFFFFE);
+	int endY = std::min(std::max(std::max(verts[0].y, verts[1].y), verts[2].y) + 1, tileEndY);
+
+	if (endX == startX || endY == startY) {
+		// No pixels
+		return;
+	}
+
+	// Fab(x, y) =     Ax       +       By     +      C              = 0
+	// Fab(x, y) = (ya - yb)x   +   (xb - xa)y + (xa * yb - xb * ya) = 0
+	// Compute A = (ya - yb) for the 3 line segments that make up each triangle
+	int A0 = verts[1].y - verts[2].y;
+	int A1 = verts[2].y - verts[0].y;
+	int A2 = verts[0].y - verts[1].y;
+
+	// Compute B = (xb - xa) for the 3 line segments that make up each triangle
+	int B0 = verts[2].x - verts[1].x;
+	int B1 = verts[0].x - verts[2].x;
+	int B2 = verts[1].x - verts[0].x;
+
+	// Compute C = (xa * yb - xb * ya) for the 3 line segments that make up each triangle
+	int C0 = verts[1].x * verts[2].y - verts[2].x * verts[1].y;
+	int C1 = verts[2].x * verts[0].y - verts[0].x * verts[2].y;
+	int C2 = verts[0].x * verts[1].y - verts[1].x * verts[0].y;
+
+	// Compute triangle area
+	int triArea = A0 * verts[0].x + B0 * verts[0].y + C0;
+	if (triArea <= 0) {
+		// Too small to rasterize or backface culled
+		// NOTE: Just disabling this check won't enable two-sided rendering.
+		// Since it's not that common, let's just queue the triangles with both windings.
+		return;
+	}
+
+	float oneOverTriArea = (1.0f / float(triArea));
+
+	float zz[3];
+	for (int vv = 0; vv < 3; vv++) {
+		zz[vv] = (float)verts[vv].z * oneOverTriArea;
+	}
+
+	int rowIdx = (startY * stride + startX);
+	int col = startX;
+	int row = startY;
+
+	// Calculate slopes at starting corner.
+	int alpha0 = (A0 * col) + (B0 * row) + C0;
+	int beta0 = (A1 * col) + (B1 * row) + C1;
+	int gama0 = (A2 * col) + (B2 * row) + C2;
+
+	// Incrementally compute Fab(x, y) for all the pixels inside the bounding box formed by (startX, endX) and (startY, endY)
+	for (int r = startY; r < endY; r++,
+		row++,
+		rowIdx += stride,
+		alpha0 += B0,
+		beta0 += B1,
+		gama0 += B2)
+	{
+		// Compute barycentric coordinates 
+		int idx = rowIdx;
+		int alpha = alpha0;
+		int beta = beta0;
+		int gama = gama0;
+
+		for (int c = startX; c < endX; c++,
+			idx++,
+			alpha += A0,
+			beta += A1,
+			gama += A2)
+		{
+			int mask = alpha >= 0 && beta >= 0 && gama >= 0;
+			// Early out if all of this quad's pixels are outside the triangle.
+			if (!mask) {
+				continue;
+			}
+			// Compute barycentric-interpolated depth
+			float depth = alpha * zz[0] + beta * zz[1] + gama * zz[2];
+			float previousDepthValue = (float)depthBuf[idx];
+
+			int depthMask;
+			switch (compareMode) {
+			case GE_COMP_EQUAL:  depthMask = depth == previousDepthValue; break;
+			case GE_COMP_LESS: depthMask = depth < previousDepthValue; break;
+			case GE_COMP_LEQUAL: depthMask = depth <= previousDepthValue; break;
+			case GE_COMP_GEQUAL: depthMask = depth >= previousDepthValue; break;
+			case GE_COMP_GREATER: depthMask = depth > previousDepthValue; break;
+			case GE_COMP_NOTEQUAL: depthMask = depth != previousDepthValue; break;
+			case GE_COMP_ALWAYS:
+			default:
+				depthMask = 1;
+				break;
+			}
+			int finalMask = mask & depthMask;
+			depth = finalMask == 1 ? depth : previousDepthValue;
+			depthBuf[idx] = (u16)depth;
+		} //for each column
+	} // for each row
+}
+
+// We ignore lots of primitive types for now.
+void DepthRasterPrim(uint16_t *depth, int depthStride, int x1, int y1, int x2, int y2, void *bufferData,
+	const void *vertexData, const void *indexData, GEPrimitiveType prim, int count, VertexDecoder *dec, u32 vertTypeID, bool clockwise) {
+
+	GEComparison compareMode = gstate.getDepthTestFunction();
+	if (gstate.isModeClear()) {
+		if (!gstate.isClearModeDepthMask()) {
+			return;
+		}
+		compareMode = GE_COMP_ALWAYS;
+	} else {
+		if (!gstate.isDepthTestEnabled() || !gstate.isDepthWriteEnabled())
+			return;
+	}
+
+	switch (prim) {
+	case GE_PRIM_INVALID:
+	case GE_PRIM_KEEP_PREVIOUS:
+	case GE_PRIM_LINES:
+	case GE_PRIM_LINE_STRIP:
+	case GE_PRIM_POINTS:
+		return;
+	default:
+		break;
+	}
+
+	// TODO: Ditch indexed primitives for now, also ditched skinned ones since we don't have a fast way to skin without
+	// running the full decoder.
+	if (vertTypeID & (GE_VTYPE_IDX_MASK | GE_VTYPE_WEIGHT_MASK)) {
+		return;
+	}
+
+	bool isThroughMode = (vertTypeID & GE_VTYPE_THROUGH_MASK) != 0;
+
+	// Turn the input data into a raw float array that we can pass to an optimized triangle rasterizer.
+	float *verts = (float *)bufferData;
+	ScreenVert *screenVerts = (ScreenVert *)((uint8_t *)bufferData + 65536 * 8);
+
+	// Simple, most common case.
+	int vertexStride = dec->VertexSize();
+	int offset = dec->posoff;
+	float factor = 1.0f;
+	switch (vertTypeID & GE_VTYPE_POS_MASK) {
+	case GE_VTYPE_POS_8BIT:
+		if (!isThroughMode) {
+			factor = 1.0f / 128.0f;
+		}
+		for (int i = 0; i < count; i++) {
+			const s8 *data = (const s8 *)vertexData + i * vertexStride + offset;
+			for (int j = 0; j < 3; j++) {
+				verts[i * 3 + j] = data[j] * factor;
+			}
+		}
+		break;
+	case GE_VTYPE_POS_16BIT:
+		if (!isThroughMode) {
+			factor = 1.0f / 32768.0f;
+		}
+		for (int i = 0; i < count; i++) {
+			const s16 *data = ((const s16 *)((const s8 *)vertexData + i * vertexStride + offset));
+			for (int j = 0; j < 3; j++) {
+				verts[i * 3 + j] = data[j] * factor;
+			}
+		}
+		break;
+	case GE_VTYPE_POS_FLOAT:
+		for (int i = 0; i < count; i++)
+			memcpy(&verts[i * 3], (const u8 *)vertexData + vertexStride * i + offset, sizeof(float) * 3);
+		break;
+	}
+
+	// OK, we now have the coordinates. Let's transform, we can actually do this in-place.
+	if (!(vertTypeID & GE_VTYPE_THROUGH_MASK)) {
+		// TODO: This is very suboptimal. This should be one matrix multiplication per vertex.
+
+		float viewportX = gstate.getViewportXCenter();
+		float viewportY = gstate.getViewportYCenter();
+		float viewportZ = gstate.getViewportZCenter();
+		float viewportScaleX = gstate.getViewportXScale();
+		float viewportScaleY = gstate.getViewportYScale();
+		float viewportScaleZ = gstate.getViewportZScale();
+		
+		bool allBehind = true;
+
+		for (int i = 0; i < count; i++) {
+			float world[3];
+			float view[3];
+			float proj[4];
+			Vec3ByMatrix43(world, verts + i * 3, gstate.worldMatrix);
+			Vec3ByMatrix43(view, world, gstate.viewMatrix);
+			Vec3ByMatrix44(proj, view, gstate.projMatrix);  // TODO: Include adjustments to the proj matrix?
+
+			float w = proj[3];
+
+			bool inFront = w > 0.0f;
+			screenVerts[i].behind = !inFront;
+			if (inFront) {
+				allBehind = false;
+			}
+
+			// Clip to the w=0 plane.
+			proj[0] /= w;
+			proj[1] /= w;
+			proj[2] /= w;
+
+			// Then transform by the viewport and offset to finally get subpixel coordinates. Normally, this is done by the viewport
+			// and offset params.
+			float screen[3];
+			screen[0] = (proj[0] * viewportScaleX + viewportX) * 16.0f - gstate.getOffsetX16();
+			screen[1] = (proj[1] * viewportScaleY + viewportY) * 16.0f - gstate.getOffsetY16();
+			screen[2] = (proj[2] * viewportScaleZ + viewportZ);
+			if (screen[2] < 0.0f) {
+				screen[2] = 0.0f;
+			}
+			if (screen[2] >= 65535.0f) {
+				screen[2] = 65535.0f;
+			}
+			screenVerts[i].x = screen[0];
+			screenVerts[i].y = screen[1];
+			screenVerts[i].z = screen[2];
+		}
+		if (allBehind) {
+			// Cull the whole draw.
+			return;
+		}
+	} else {
+		for (int i = 0; i < count; i++) {
+			screenVerts[i].x = (int)verts[i * 3 + 0] << 4;
+			screenVerts[i].y = (int)verts[i * 3 + 1] << 4;
+			screenVerts[i].z = (u16)clamp_value(verts[i * 3 + 2], 0.0f, 65535.0f);
+		}
+	}
+
+	// Then we need to stitch primitives from strips, etc etc...
+	// For now we'll just do it tri by tri. Later let's be more efficient.
+	
+	switch (prim) {
+	case GE_PRIM_RECTANGLES:
+		for (int i = 0; i < count / 2; i++) {
+			uint16_t z = screenVerts[i + 1].z;  // depth from second vertex
+			// We remove the subpixel information here.
+			DepthRasterRect(depth, depthStride, screenVerts[i].x >> 4, screenVerts[i].y >> 4, screenVerts[i + 1].x >> 4, screenVerts[i + 1].y >> 4,
+				z, compareMode);
+		}
+		break;
+	case GE_PRIM_TRIANGLES:
+		for (int i = 0; i < count / 3; i++) {
+			if (screenVerts[i * 3].behind || screenVerts[i * 3 + 1].behind || screenVerts[i * 3 + 2].behind) {
+				continue;
+			}
+			DepthRasterTriangle(depth, depthStride, x1, y1, x2, y2, screenVerts + i * 3, compareMode);
+		}
+		break;
+	case GE_PRIM_TRIANGLE_STRIP:
+	{
+		int wind = 2;
+		for (int i = 0; i < count - 2; i++) {
+			int i0 = i;
+			int i1 = i + wind;
+			wind ^= 3;
+			int i2 = i + wind;
+			if (screenVerts[i0].behind || screenVerts[i1].behind || screenVerts[i2].behind) {
+				continue;
+			}
+			ScreenVert v[3];
+			v[0] = screenVerts[i0];
+			v[1] = screenVerts[i1];
+			v[2] = screenVerts[i2];
+			DepthRasterTriangle(depth, depthStride, x1, y1, x2, y2, v, compareMode);
+		}
+		break;
+	}
+	}
+}
diff --git a/GPU/Common/DepthRaster.h b/GPU/Common/DepthRaster.h
new file mode 100644
index 0000000000..01fa60e257
--- /dev/null
+++ b/GPU/Common/DepthRaster.h
@@ -0,0 +1,12 @@
+#pragma once
+
+#include "Common/CommonTypes.h"
+#include "GPU/ge_constants.h"
+
+// Specialized, very limited depth-only rasterizer.
+// Meant to run in parallel with hardware rendering, in games that read back the depth buffer
+// for effects like lens flare.
+// So, we can be quite inaccurate without any issues, and skip a lot of functionality.
+
+class VertexDecoder;
+void DepthRasterPrim(uint16_t *dest, int stride, int x1, int x2, int y1, int y2, void *bufferData, const void *vertexData, const void *indexData, GEPrimitiveType prim, int count, VertexDecoder *decoder, u32 vertexTypeID, bool clockwise);
diff --git a/GPU/Common/DrawEngineCommon.h b/GPU/Common/DrawEngineCommon.h
index 0f8ab8a751..595ab929aa 100644
--- a/GPU/Common/DrawEngineCommon.h
+++ b/GPU/Common/DrawEngineCommon.h
@@ -158,6 +158,11 @@ public:
 		_dbg_assert_(numDrawVerts_ == 0 && numDrawInds_ == 0);
 	}
 
+	// temporary hack
+	uint8_t *GetTempSpace() {
+		return decoded_ + 12 * 65536;
+	}
+
 protected:
 	virtual bool UpdateUseHWTessellation(bool enabled) const { return enabled; }
 	void UpdatePlanes();
diff --git a/GPU/Common/VertexDecoderCommon.h b/GPU/Common/VertexDecoderCommon.h
index 2793ad80ee..4b905f03c9 100644
--- a/GPU/Common/VertexDecoderCommon.h
+++ b/GPU/Common/VertexDecoderCommon.h
@@ -122,7 +122,7 @@ public:
 // Reads decoded vertex formats in a convenient way. For software transform and debugging.
 class VertexReader {
 public:
-	VertexReader(u8 *base, const DecVtxFormat &decFmt, int vtype) : base_(base), data_(base), decFmt_(decFmt), vtype_(vtype) {}
+	VertexReader(const u8 *base, const DecVtxFormat &decFmt, int vtype) : base_(base), data_(base), decFmt_(decFmt), vtype_(vtype) {}
 
 	void ReadPos(float pos[3]) const {
 		// Only DEC_FLOAT_3 is supported.
@@ -297,8 +297,8 @@ public:
 	}
 
 private:
-	u8 *base_;
-	u8 *data_;
+	const u8 *base_;
+	const u8 *data_;
 	DecVtxFormat decFmt_;
 	int vtype_;
 };
diff --git a/GPU/GPU.vcxproj b/GPU/GPU.vcxproj
index 5cb3ea9e62..c27d083549 100644
--- a/GPU/GPU.vcxproj
+++ b/GPU/GPU.vcxproj
@@ -346,6 +346,7 @@
   </ItemDefinitionGroup>
   <ItemGroup>
     <ClInclude Include="..\ext\xbrz\xbrz.h" />
+    <ClInclude Include="Common\DepthRaster.h" />
     <ClInclude Include="Common\ReplacedTexture.h" />
     <ClInclude Include="Common\TextureReplacer.h" />
     <ClInclude Include="Common\TextureShaderCommon.h" />
@@ -468,6 +469,7 @@
   <ItemGroup>
     <ClCompile Include="..\ext\xbrz\xbrz.cpp" />
     <ClCompile Include="Common\DepthBufferCommon.cpp" />
+    <ClCompile Include="Common\DepthRaster.cpp" />
     <ClCompile Include="Common\ReplacedTexture.cpp" />
     <ClCompile Include="Common\TextureReplacer.cpp" />
     <ClCompile Include="Common\TextureShaderCommon.cpp" />
diff --git a/GPU/GPU.vcxproj.filters b/GPU/GPU.vcxproj.filters
index 610ba94cbe..1529b974c1 100644
--- a/GPU/GPU.vcxproj.filters
+++ b/GPU/GPU.vcxproj.filters
@@ -279,6 +279,9 @@
     <ClInclude Include="Debugger\State.h">
       <Filter>Debugger</Filter>
     </ClInclude>
+    <ClInclude Include="Common\DepthRaster.h">
+      <Filter>Common</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="Math3D.cpp">
@@ -554,6 +557,9 @@
     <ClCompile Include="Debugger\State.cpp">
       <Filter>Debugger</Filter>
     </ClCompile>
+    <ClCompile Include="Common\DepthRaster.cpp">
+      <Filter>Common</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <FxCompile Include="..\assets\shaders\tex_4xbrz.csh">
diff --git a/GPU/GPUCommonHW.cpp b/GPU/GPUCommonHW.cpp
index 9b5389c875..f5383cf6be 100644
--- a/GPU/GPUCommonHW.cpp
+++ b/GPU/GPUCommonHW.cpp
@@ -13,6 +13,7 @@
 #include "GPU/Common/DrawEngineCommon.h"
 #include "GPU/Common/TextureCacheCommon.h"
 #include "GPU/Common/FramebufferManagerCommon.h"
+#include "GPU/Common/DepthRaster.h"
 
 struct CommonCommandTableEntry {
 	uint8_t cmd;
@@ -1039,6 +1040,10 @@ void GPUCommonHW::Execute_Prim(u32 op, u32 diff) {
 	if (passCulling) {
 		if (!drawEngineCommon_->SubmitPrim(verts, inds, prim, count, decoder, vertTypeID, true, &bytesRead)) {
 			canExtend = false;
+		} else if (PSP_CoreParameter().compat.flags().SoftwareRasterDepth) {
+			DepthRasterPrim((uint16_t *)Memory::GetPointerWrite(gstate.getDepthBufRawAddress() | 0x04000000), gstate.DepthBufStride(),
+				gstate.getScissorX1(), gstate.getScissorY1(), gstate.getScissorX2(), gstate.getScissorY2(), drawEngineCommon_->GetTempSpace(),
+				verts, inds, prim, count, decoder, vertTypeID, false);
 		}
 		onePassed = true;
 	} else {
@@ -1117,6 +1122,10 @@ void GPUCommonHW::Execute_Prim(u32 op, u32 diff) {
 			if (passCulling) {
 				if (!drawEngineCommon_->SubmitPrim(verts, inds, newPrim, count, decoder, vertTypeID, clockwise, &bytesRead)) {
 					canExtend = false;
+				} else if (PSP_CoreParameter().compat.flags().SoftwareRasterDepth) {
+					DepthRasterPrim((uint16_t *)Memory::GetPointerWrite(gstate.getDepthBufRawAddress() | 0x04000000), gstate.DepthBufStride(),
+						gstate.getScissorX1(), gstate.getScissorY1(), gstate.getScissorX2(), gstate.getScissorY2(), drawEngineCommon_->GetTempSpace(),
+						verts, inds, newPrim, count, decoder, vertTypeID, clockwise);
 				}
 				// As soon as one passes, assume we don't need to check the rest of this batch.
 				onePassed = true;
diff --git a/UI/ImDebugger/ImGe.h b/UI/ImDebugger/ImGe.h
index 227b0f0c1e..41dc6a8b1f 100644
--- a/UI/ImDebugger/ImGe.h
+++ b/UI/ImDebugger/ImGe.h
@@ -71,14 +71,14 @@ struct ImGePixelViewer : public PixelLookup {
 	}
 	bool FormatValueAt(char *buf, size_t bufSize, int x, int y) const override;
 
-	uint32_t addr = 0x04000000;
+	uint32_t addr = 0x04110000;
 	uint16_t stride = 512;
 	uint16_t width = 480;
 	uint16_t height = 272;
-	GEBufferFormat format = GE_FORMAT_565;
+	GEBufferFormat format = GE_FORMAT_DEPTH16;
 	bool useAlpha = false;
 	bool showAlpha = false;
-	float scale = 1.0f;
+	float scale = 20.0f;
 
 private:
 	void UpdateTexture(Draw::DrawContext *draw);
diff --git a/UWP/GPU_UWP/GPU_UWP.vcxproj b/UWP/GPU_UWP/GPU_UWP.vcxproj
index 7bb4b346bd..a7ba27a140 100644
--- a/UWP/GPU_UWP/GPU_UWP.vcxproj
+++ b/UWP/GPU_UWP/GPU_UWP.vcxproj
@@ -109,6 +109,7 @@
     </Link>
   </ItemDefinitionGroup>
   <ItemGroup>
+    <ClInclude Include="..\..\GPU\Common\DepthRaster.h" />
     <ClInclude Include="..\..\GPU\Common\ReplacedTexture.h" />
     <ClInclude Include="..\..\GPU\Common\TextureReplacer.h" />
     <ClInclude Include="..\..\GPU\Common\TextureShaderCommon.h" />
@@ -177,6 +178,7 @@
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="..\..\GPU\Common\DepthBufferCommon.cpp" />
+    <ClCompile Include="..\..\GPU\Common\DepthRaster.cpp" />
     <ClCompile Include="..\..\GPU\Common\ReplacedTexture.cpp" />
     <ClCompile Include="..\..\GPU\Common\TextureReplacer.cpp" />
     <ClCompile Include="..\..\GPU\Common\TextureShaderCommon.cpp" />
@@ -261,4 +263,4 @@
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
   <ImportGroup Label="ExtensionTargets">
   </ImportGroup>
-</Project>
+</Project>
\ No newline at end of file
diff --git a/UWP/GPU_UWP/GPU_UWP.vcxproj.filters b/UWP/GPU_UWP/GPU_UWP.vcxproj.filters
index 84b4c5d396..31d14b549f 100644
--- a/UWP/GPU_UWP/GPU_UWP.vcxproj.filters
+++ b/UWP/GPU_UWP/GPU_UWP.vcxproj.filters
@@ -80,6 +80,7 @@
     <ClCompile Include="..\..\GPU\Debugger\GECommandTable.cpp">
       <Filter>Debugger</Filter>
     </ClCompile>
+    <ClCompile Include="..\..\GPU\Common\DepthRaster.cpp" />
   </ItemGroup>
   <ItemGroup>
     <ClInclude Include="..\..\GPU\Common\DepalettizeShaderCommon.h" />
@@ -163,10 +164,11 @@
     <ClInclude Include="..\..\GPU\Debugger\GECommandTable.h">
       <Filter>Debugger</Filter>
     </ClInclude>
+    <ClInclude Include="..\..\GPU\Common\DepthRaster.h" />
   </ItemGroup>
   <ItemGroup>
     <Filter Include="Debugger">
       <UniqueIdentifier>{49bcf7f6-518a-4ecd-af55-bda3a344efe7}</UniqueIdentifier>
     </Filter>
   </ItemGroup>
-</Project>
+</Project>
\ No newline at end of file
diff --git a/android/jni/Android.mk b/android/jni/Android.mk
index 10ab9a5f77..dbd8809788 100644
--- a/android/jni/Android.mk
+++ b/android/jni/Android.mk
@@ -530,6 +530,7 @@ EXEC_AND_LIB_FILES := \
   $(SRC)/GPU/Common/SoftwareTransformCommon.cpp.arm \
   $(SRC)/GPU/Common/ReinterpretFramebuffer.cpp \
   $(SRC)/GPU/Common/DepthBufferCommon.cpp \
+  $(SRC)/GPU/Common/DepthRaster.cpp \
   $(SRC)/GPU/Common/VertexDecoderCommon.cpp.arm \
   $(SRC)/GPU/Common/VertexDecoderHandwritten.cpp.arm \
   $(SRC)/GPU/Common/TextureCacheCommon.cpp.arm \
diff --git a/assets/compat.ini b/assets/compat.ini
index 8a374753ef..0452c5b854 100644
--- a/assets/compat.ini
+++ b/assets/compat.ini
@@ -1228,8 +1228,10 @@ ULJS19067 = true
 ULAS42247 = true
 ULAS42318 = true
 
+[SoftwareRasterDepth]
+
 [DisableFirstFrameReadback]
-# Wipeout Pure: Temporary workaround for lens flare flicker. See #13344
+# Wipeout Pure
 UCUS98612 = true
 UCJS10007 = true
 UCES00001 = true
diff --git a/libretro/Makefile.common b/libretro/Makefile.common
index 804a1d7219..c1cb5a454f 100644
--- a/libretro/Makefile.common
+++ b/libretro/Makefile.common
@@ -543,6 +543,7 @@ SOURCES_CXX += \
 	$(GPUDIR)/Common/TextureScalerCommon.cpp \
 	$(GPUDIR)/Common/SoftwareTransformCommon.cpp \
 	$(GPUDIR)/Common/DepthBufferCommon.cpp \
+	$(GPUDIR)/Common/DepthRaster.cpp \
 	$(GPUDIR)/Common/StencilCommon.cpp \
 	$(GPUDIR)/Software/TransformUnit.cpp \
 	$(GPUDIR)/Software/SoftGpu.cpp \