GS/HW: Multiple fixes

bin/resources/GameIndex.yaml

@@ -32786,6 +32786,8 @@ SLPM-65080:
   name: "Tokimeki Memorial 3"
   region: "NTSC-J"
   compat: 5
+  gsHWFixes:
+    textureInsideRT: 1 # Fixes character portraits.
 SLPM-65081:
   name: "Grandia II"
   region: "NTSC-J"

pcsx2/GS/Renderers/HW/GSRendererHW.cpp

@@ -4210,10 +4210,10 @@ __ri void GSRendererHW::EmulateTextureSampler(const GSTextureCache::Target* rt,
 	m_conf.ps.wms = (wms & 2 || target_region) ? wms : 0;
 	m_conf.ps.wmt = (wmt & 2 || target_region) ? wmt : 0;
 
-	// Depth + bilinear filtering isn't done yet (And I'm not sure we need it anyway but a game will prove me wrong)
-	// So of course, GTA set the linear mode, but sampling is done at texel center so it is equivalent to nearest sampling
+	// Depth + bilinear filtering isn't done yet. But if the game has just set a Z24 swizzle on a colour texture, we can
+	// just pretend it's not a depth format, since in the texture cache, it's not.
 	// Other games worth testing: Area 51, Burnout
-	if (psm.depth && m_vt.IsLinear())
+	if (psm.depth && m_vt.IsLinear() && tex->GetTexture()->IsDepthStencil())
 		GL_INS("WARNING: Depth + bilinear filtering not supported");
 
 	// Performance note:
@@ -4293,22 +4293,14 @@ __ri void GSRendererHW::EmulateTextureSampler(const GSTextureCache::Target* rt,
 		}
 
 		// Depth format
-		if (tex->m_texture->GetType() == GSTexture::Type::DepthStencil)
+		if (tex->m_texture->IsDepthStencil())
 		{
 			// Require a float conversion if the texure is a depth format
 			m_conf.ps.depth_fmt = (psm.bpp == 16) ? 2 : 1;
 
 			// Don't force interpolation on depth format
 			bilinear &= m_vt.IsLinear();
 		}
-		else if (psm.depth)
-		{
-			// Use Integral scaling
-			m_conf.ps.depth_fmt = 3;
-
-			// Don't force interpolation on depth format
-			bilinear &= m_vt.IsLinear();
-		}
 
 		const GSVector4 half_pixel = RealignTargetTextureCoordinate(tex);
 		m_conf.cb_vs.texture_offset = GSVector2(half_pixel.x, half_pixel.y);
@@ -4384,9 +4376,11 @@ __ri void GSRendererHW::EmulateTextureSampler(const GSTextureCache::Target* rt,
 	m_conf.cb_ps.HalfTexel = GSVector4(-0.5f, 0.5f).xxyy() / WH.zwzw();
 	if (complex_wms_wmt)
 	{
+		// Add 0.5 to the coordinates because the region clamp is inclusive, size is exclusive. We use 0.5 because we want to clamp
+		// to the last texel in the image, not halfway between it and wrapping around.
 		const GSVector4i clamp(m_cached_ctx.CLAMP.MINU, m_cached_ctx.CLAMP.MINV, m_cached_ctx.CLAMP.MAXU, m_cached_ctx.CLAMP.MAXV);
-		const GSVector4 region_repeat(GSVector4::cast(clamp));
-		const GSVector4 region_clamp(GSVector4(clamp) / WH.xyxy());
+		const GSVector4 region_repeat = GSVector4::cast(clamp);
+		const GSVector4 region_clamp = (GSVector4(clamp) + GSVector4::cxpr(0.0f, 0.0f, 0.5f, 0.5f)) / WH.xyxy();
 		if (wms >= CLAMP_REGION_CLAMP)
 		{
 			m_conf.cb_ps.MinMax.x = (wms == CLAMP_REGION_CLAMP && !m_conf.ps.depth_fmt) ? region_clamp.x : region_repeat.x;

pcsx2/GS/Renderers/HW/GSTextureCache.cpp

@@ -1864,8 +1864,7 @@ GSTextureCache::Target* GSTextureCache::LookupTarget(GIFRegTEX0 TEX0, const GSVe
 	}
 	else if (!is_frame && !GSConfig.UserHacks_DisableDepthSupport)
 	{
-
-		int rev_type = (type == DepthStencil) ? RenderTarget : DepthStencil;
+		const int rev_type = (type == DepthStencil) ? RenderTarget : DepthStencil;
 
 		// Depth stencil/RT can be an older RT/DS but only check recent RT/DS to avoid to pick
 		// some bad data.
@@ -1878,13 +1877,35 @@ GSTextureCache::Target* GSTextureCache::LookupTarget(GIFRegTEX0 TEX0, const GSVe
 			if (bp != t->m_TEX0.TBP0 || !t->m_valid_rgb)
 				continue;
 
-			// Probably an old target, get rid of it.
-			if (possible_clear && GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp != GSLocalMemory::m_psm[TEX0.PSM].bpp)
+			const GSLocalMemory::psm_t& t_psm_s = GSLocalMemory::m_psm[t->m_TEX0.PSM];
+			if (t_psm_s.bpp != psm_s.bpp)
 			{
-				InvalidateSourcesFromTarget(t);
-				i = rev_list.erase(i);
-				delete t;
-				continue;
+				bool remove_target = possible_clear;
+
+				// If we have a BW change, and it's not a multiple of 2 (for a shuffle), the game's going to get a jigsaw
+				// puzzle of pages and can't be expecting to have legitimate data. Tokimeki Memorial 3 reuses a BW 17
+				// buffer as BW 10, and if we don't discard the BW 17 buffer, the BW 10 buffer ends up twice the size.
+				const u32 shuffle_bw = (psm_s.bpp > t_psm_s.bpp) ? (TEX0.TBW / 2u) : (TEX0.TBW * 2u);
+				if (t->m_TEX0.TBW != TEX0.TBW && t->m_TEX0.TBW != shuffle_bw)
+				{
+					// But we'll make sure the whole existing target's actually being drawn over to be safe.
+					const u32 end_block = GSLocalMemory::GetUnwrappedEndBlockAddress(TEX0.TBP0, TEX0.TBW, TEX0.PSM, draw_rect);
+					if (end_block >= t->UnwrappedEndBlock())
+					{
+						GL_CACHE("Not converting %s at %x TBW %u with end block of %x when we're drawing through %x",
+							to_string(rev_type), t->m_TEX0.TBP0, t->m_TEX0.TBW, t->UnwrappedEndBlock(), end_block);
+						remove_target = true;
+					}
+				}
+
+				// Probably an old target, get rid of it.
+				if (remove_target)
+				{
+					InvalidateSourcesFromTarget(t);
+					i = rev_list.erase(i);
+					delete t;
+					continue;
+				}
 			}
 
 			if (t->m_age == 0)