[ci-skip] 2.4 Release

Redundant calls, blending will be enabled if needed.

.github/ISSUE_TEMPLATE/app_bug_report.yaml

@@ -70,6 +70,8 @@ body:
       - Windows 11
       - Windows 10 (64bit)
       - Linux (64bit) - Specify distro below
+      - macOS 26 (Tahoe)
+      - macOS 15 (Sequoia)
       - macOS 14 (Sonoma)
       - macOS 13 (Ventura)
       - macOS 12 (Monterey)

.github/ISSUE_TEMPLATE/emu_bug_report.yaml

@@ -87,6 +87,8 @@ body:
       - Windows 11
       - Windows 10 (64bit)
       - Linux (64bit) - Specify distro below
+      - macOS 26 (Tahoe)
+      - macOS 15 (Sequoia)
       - macOS 14 (Sonoma)
       - macOS 13 (Ventura)
       - macOS 12 (Monterey)
@@ -104,14 +106,14 @@ body:
     id: cpu
     attributes:
       label: CPU
-      placeholder: "Example: i5-7600"
+      placeholder: "Example: Intel i5 12400F"
     validations:
       required: true
   - type: input
     id: gpu
     attributes:
       label: GPU
-      placeholder: "Example: GTX 1070"
+      placeholder: "Example: Nvidia RTX 4060"
     validations:
       required: true
   - type: textarea

.github/workflows/cron_publish_flatpak.yml

@@ -6,36 +6,45 @@ on:
   workflow_dispatch: # As well as manually.
 
 jobs:
-  check:
-    if: github.repository == 'PCSX2/pcsx2'
-    name: "Check if release is needed"
-    runs-on: ubuntu-latest
-    timeout-minutes: 180
-    outputs:
-      PCSX2_RELEASE: ${{ steps.getinfo.outputs.PCSX2_RELEASE }}
-      FLATHUB_RELEASE: ${{ steps.getinfo.outputs.FLATHUB_RELEASE }}
-    steps:
-      - name: Get latest tag and Flathub release
-        id: getinfo
-        env:
-          GH_TOKEN: ${{ github.token }}
-        run: |
-          PCSX2_RELEASE=$(gh api -H 'Accept: application/vnd.github+json' -H 'X-GitHub-Api-Version: 2022-11-28' /repos/PCSX2/pcsx2/releases | jq -r '.[0].tag_name')
-          FLATHUB_RELEASE=$(curl -L -s https://flathub.org/api/v2/appstream/net.pcsx2.PCSX2 | jq -r '.releases | max_by(.version) | .version')
-          echo "Latest PCSX2 release is: '${PCSX2_RELEASE}'"
-          echo "Latest Flathub release is: '${FLATHUB_RELEASE}'"
-          PCSX2_RELEASE=$(echo $PCSX2_RELEASE | sed 's/[^0-9]*//g')
-          FLATHUB_RELEASE=$(echo $FLATHUB_RELEASE | sed 's/[^0-9]*//g')
-          echo "PCSX2_RELEASE=${PCSX2_RELEASE}" >> "$GITHUB_OUTPUT"
-          echo "FLATHUB_RELEASE=${FLATHUB_RELEASE}" >> "$GITHUB_OUTPUT"
+
+# check is disabled as the flathub api does not give us beta repository information
+# Alternatively we can "flatpak remote-info or parse the appstream directly for the beta repo"
+# Maybe in the future if we don't want to publish the same version twice if we get no commits
+# for 24 hours.
+
+#  check:
+#    if: github.repository == 'PCSX2/pcsx2'
+#    name: "Check if release is needed"
+#    runs-on: ubuntu-latest
+#    timeout-minutes: 180
+#    outputs:
+#      PCSX2_RELEASE: ${{ steps.getinfo.outputs.PCSX2_RELEASE }}
+#      FLATHUB_RELEASE: ${{ steps.getinfo.outputs.FLATHUB_RELEASE }}
+#    steps:
+#      - name: Get latest tag and Flathub release
+#        id: getinfo
+#        env:
+#          GH_TOKEN: ${{ github.token }}
+#        run: |
+#          PCSX2_RELEASE=$(gh api -H 'Accept: application/vnd.github+json' -H 'X-GitHub-Api-Version: 2022-11-28' /repos/PCSX2/pcsx2/releases | jq -r '.[0].tag_name')
+#          FLATHUB_RELEASE=$(curl -L -s https://flathub.org/api/v2/appstream/net.pcsx2.PCSX2 | jq -r '.releases | max_by(.version) | .version')
+#          echo "Latest PCSX2 release is: '${PCSX2_RELEASE}'"
+#          echo "Latest Flathub release is: '${FLATHUB_RELEASE}'"
+#          PCSX2_RELEASE=$(echo $PCSX2_RELEASE | sed 's/[^0-9]*//g')
+#          FLATHUB_RELEASE=$(echo $FLATHUB_RELEASE | sed 's/[^0-9]*//g')
+#          echo "PCSX2_RELEASE=${PCSX2_RELEASE}" >> "$GITHUB_OUTPUT"
+#          echo "FLATHUB_RELEASE=${FLATHUB_RELEASE}" >> "$GITHUB_OUTPUT"
 
   build:
-    needs: check
+#    needs: check
     # outputs are automatically compared as strings. This doesn't work in our favour
     # Use fromJson() to convert them to proper integers...
     # see: https://github.com/github/docs/pull/25870
     # and: https://github.com/orgs/community/discussions/57480
-    #if: fromJson(needs.check.outputs.FLATHUB_RELEASE) < fromJson(needs.check.outputs.PCSX2_RELEASE)
+
+#   if: fromJson(needs.check.outputs.FLATHUB_RELEASE) < fromJson(needs.check.outputs.PCSX2_RELEASE)
+    # As the check step is disabled, perform repository check here
+    if: github.repository == 'PCSX2/pcsx2'
     name: "Build and publish Flatpak"
     uses: ./.github/workflows/linux_build_flatpak.yml
     with:

.github/workflows/linux_build_flatpak.yml

@@ -54,19 +54,16 @@ jobs:
         uses: actions/checkout@v4
         with:
           set-safe-directory: ${{ env.GITHUB_WORKSPACE }}
+          # 10 here, since the odds of having 10 untagged commits in a row should be slim to none
+          # This is required for the tagging logic in generate-metainfo.sh
+          fetch-depth: 10
+          fetch-tags: true
 
       # Work around container ownership issue
       - name: Set Safe Directory
         shell: bash
         run: git config --global --add safe.directory "*"
 
-      # Hackity hack. When running the workflow on a schedule, we don't have the tag,
-      # it doesn't fetch tags, therefore we don't get a version. So grab them manually.
-      # actions/checkout elides tags, fetch them primarily for releases
-      - name: Fetch tags
-        if: ${{ inputs.fetchTags }}
-        run: git fetch --tags --no-recurse-submodules
-
       - name: Add stable release identifier file
         if: ${{ inputs.stableBuild == true || inputs.stableBuild == 'true' }}
         shell: bash
@@ -125,7 +122,7 @@ jobs:
           cache: true
           restore-cache: true
           cache-key: ${{ inputs.os }} ${{ inputs.platform }} ${{ inputs.compiler }} flatpak ${{ hashFiles('.github/workflows/scripts/linux/flatpak/**/*.json') }}
-  
+
       #- name: Validate build
       #  run: |
       #    flatpak-builder-lint repo repo

.github/workflows/linux_build_qt.yml

@@ -144,6 +144,7 @@ jobs:
             -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
             -DENABLE_SETCAP=OFF \
             -DDISABLE_ADVANCE_SIMD=TRUE \
+            -DUSE_LINKED_FFMPEG=ON \
             -DCMAKE_DISABLE_PRECOMPILE_HEADERS=ON \
             $ADDITIONAL_CMAKE_ARGS
 

.github/workflows/scripts/linux/generate-metainfo.sh

@@ -12,12 +12,24 @@ GIT_DATE=$(git log -1 --pretty=%cd --date=iso8601)
 GIT_VERSION=$(git tag --points-at HEAD)
 GIT_HASH=$(git rev-parse HEAD)
 
-if [[ "${GIT_VERSION}" == "" ]]; then
-	# In the odd event that we run this script before the release gets tagged.
-	GIT_VERSION=$(git describe --tags)
-	if [[ "${GIT_VERSION}" == "" ]]; then
-		GIT_VERSION=$(git rev-parse HEAD)
-	fi
+if [[ -z "${GIT_VERSION}" ]]; then
+    if git branch -r --contains HEAD | grep -q 'origin/master'; then
+        # Our master doesn't have a tagged commit
+        # This happens when the commit is "ci skip" 
+        # abbrev so we have just the latest tag
+        # ie v2.3.420 (Yes, that's the current master at the time of writing)
+        GIT_VERSION=$(git describe --tags --abbrev=0)
+    else
+        # We are probably building a PR
+        # Keep the short SHA in the version
+        # ie v2.3.420-1-g10dc1a2da
+        GIT_VERSION=$(git describe --tags)
+    fi
+
+    if [[ -z "${GIT_VERSION}" ]]; then
+        # Fallback to raw commit hash
+        GIT_VERSION=$(git rev-parse HEAD)
+    fi
 fi
 
 echo "GIT_DATE: ${GIT_DATE}"

CMakeLists.txt

@@ -62,10 +62,12 @@ endif()
 
 # gsrunner
 if(ENABLE_GSRUNNER)
-	if (NOT WIN32)
-		message(WARNING "GSRunner is only supported on Windows and may not build on your system")
+	if (NOT WIN32 AND NOT APPLE)
+		message(WARNING "GSRunner is only supported on Windows and macOS and may not build on your system")
 	endif()
 	add_subdirectory(pcsx2-gsrunner)
+else()
+	add_subdirectory(pcsx2-gsrunner EXCLUDE_FROM_ALL)
 endif()
 
 #-------------------------------------------------------------------------------

bin/docs/ThirdPartyLicenses.html

@@ -24,6 +24,7 @@
 <li><a href="#googletest">GoogleTest</a></li>
 <li><a href="#harfbuzz">HarfBuzz</a></li>
 <li><a href="#jsonformoderncpp">JSON for Modern C++</a></li>
+<li><a href="#kdbindings">KDBindings</a></li>
 <li><a href="#kddockwidgets">KDDockWidgets</a></li>
 <li><a href="#kissfft">kissfft</a></li>
 <li><a href="#libbacktrace">libbacktrace</a></li>
@@ -1910,6 +1911,7 @@ PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
 </pre>
 </div>
 
+<!-- Transitive dependency of KDDockWidgets. -->
 <div id="jsonformoderncpp">
 <h3>JSON for Modern C++ - <a href="https://json.nlohmann.me">https://json.nlohmann.me</a></h3>
 <pre>
@@ -1937,6 +1939,34 @@ SOFTWARE.
 </pre>
 </div>
 
+<!-- Transitive dependency of KDDockWidgets. -->
+<div id="kdbindings">
+<h3>KDBindings - <a href="https://github.com/KDAB/KDBindings">https://github.com/KDAB/KDBindings</a></h3>
+<pre>
+MIT License
+
+Copyright 2021 Klarälvdalens Datakonsult AB, a KDAB Group company &lt;info@kdab.com&gt;
+Copyright 2021 Jeremy Burns
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the &quot;Software&quot;), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
+the Software, and to permit persons to whom the Software is furnished to do so,
+subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED &quot;AS IS&quot;, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
+FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
+COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+</pre>
+</div>
+
 <div id="kddockwidgets">
 <h3>KDDockWidgets - <a href="https://github.com/KDAB/KDDockWidgets">https://github.com/KDAB/KDDockWidgets</a></h3>
 <pre>

bin/resources/fullscreenui/applications-system.svg

@@ -0,0 +1,41 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Generator: SVG Repo Mixer Tools -->
+
+<svg
+   fill="#000000"
+   width="800px"
+   height="800px"
+   viewBox="0 0 32 32"
+   version="1.1"
+   id="svg1"
+   sodipodi:docname="applications-system.svg"
+   inkscape:version="1.4.2 (ebf0e940d0, 2025-05-08)"
+   xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
+   xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
+   xmlns="http://www.w3.org/2000/svg"
+   xmlns:svg="http://www.w3.org/2000/svg">
+  <sodipodi:namedview
+     id="namedview1"
+     pagecolor="#505050"
+     bordercolor="#ffffff"
+     borderopacity="1"
+     inkscape:showpageshadow="0"
+     inkscape:pageopacity="0"
+     inkscape:pagecheckerboard="1"
+     inkscape:deskcolor="#505050"
+     inkscape:zoom="0.97875"
+     inkscape:cx="399.48914"
+     inkscape:cy="400"
+     inkscape:window-width="1920"
+     inkscape:window-height="1010"
+     inkscape:window-x="0"
+     inkscape:window-y="0"
+     inkscape:window-maximized="1"
+     inkscape:current-layer="svg1" />
+  <defs
+     id="defs1" />
+  <path
+     d="M30.015 12.97l-2.567-0.569c-0.2-0.64-0.462-1.252-0.762-1.841l1.389-2.313c0.518-0.829 0.78-2.047 0-2.829l-1.415-1.414c-0.78-0.781-2.098-0.64-2.894-0.088l-2.251 1.434c-0.584-0.303-1.195-0.563-1.829-0.768l-0.576-2.598c-0.172-0.953-1.005-1.984-2.11-1.984h-2c-1.104 0-1.781 1.047-2 2l-0.642 2.567c-0.678 0.216-1.328 0.492-1.948 0.819l-2.308-1.47c-0.795-0.552-2.114-0.692-2.894 0.088l-1.415 1.414c-0.781 0.782-0.519 2 0 2.828l1.461 2.435c-0.274 0.552-0.517 1.123-0.705 1.72l-2.566 0.569c-0.953 0.171-1.984 1.005-1.984 2.109v2c0 1.105 1.047 1.782 2 2l2.598 0.649c0.179 0.551 0.404 1.080 0.658 1.593l-1.462 2.438c-0.518 0.828-0.78 2.047 0 2.828l1.415 1.414c0.78 0.782 2.098 0.64 2.894 0.089l2.313-1.474c0.623 0.329 1.277 0.608 1.96 0.823l0.64 2.559c0.219 0.953 0.896 2 2 2h2c1.105 0 1.938-1.032 2.11-1.985l0.577-2.604c0.628-0.203 1.23-0.459 1.808-0.758l2.256 1.438c0.796 0.552 2.114 0.692 2.895-0.089l1.415-1.414c0.78-0.782 0.518-2 0-2.828l-1.39-2.317c0.279-0.549 0.521-1.12 0.716-1.714l2.599-0.649c0.953-0.219 2-0.895 2-2v-2c0-1.104-1.031-1.938-1.985-2.11zM30.001 16.939c-0.085 0.061-0.245 0.145-0.448 0.192l-3.708 0.926-0.344 1.051c-0.155 0.474-0.356 0.954-0.597 1.428l-0.502 0.986 1.959 3.267c0.125 0.2 0.183 0.379 0.201 0.485l-1.316 1.314c-0.127-0.040-0.271-0.092-0.341-0.14l-3.292-2.099-1.023 0.529c-0.493 0.256-0.999 0.468-1.503 0.631l-1.090 0.352-0.824 3.723c-0.038 0.199-0.145 0.36-0.218 0.417h-1.8c-0.061-0.085-0.145-0.245-0.191-0.448l-0.921-3.681-1.066-0.338c-0.549-0.173-1.097-0.404-1.63-0.684l-1.028-0.543-3.293 2.099c-0.135 0.091-0.279 0.143-0.409 0.143l-1.311-1.276c0.018-0.104 0.072-0.274 0.181-0.449l2.045-3.408-0.487-0.98c-0.227-0.462-0.407-0.895-0.547-1.325l-0.343-1.052-3.671-0.918c-0.231-0.052-0.398-0.139-0.485-0.2v-1.86c0.001 0.001 0.002 0.001 0.005 0.001 0.034 0 0.198-0.117 0.335-0.142l3.772-0.835 0.346-1.103c0.141-0.449 0.333-0.917 0.588-1.43l0.487-0.98-2.024-3.373c-0.125-0.201-0.184-0.38-0.201-0.485l1.315-1.314c0.128 0.041 0.271 0.093 0.34 0.14l3.354 2.138 1.027-0.542c0.527-0.278 1.073-0.507 1.622-0.682l1.063-0.338 0.912-3.649c0.053-0.231 0.138-0.398 0.2-0.485h1.859c-0.014 0.020 0.115 0.195 0.142 0.339l0.84 3.794 1.089 0.352c0.511 0.165 1.023 0.38 1.523 0.639l1.023 0.532 3.224-2.053c0.135-0.092 0.279-0.143 0.409-0.143l1.313 1.276c-0.017 0.104-0.072 0.276-0.181 0.45l-1.98 3.296 0.505 0.988c0.273 0.533 0.48 1.033 0.635 1.529l0.346 1.104 3.697 0.82c0.224 0.041 0.398 0.171 0.434 0.241zM16.013 9.99c-3.321 0-6.023 2.697-6.023 6.010s2.702 6.010 6.023 6.010 6.023-2.697 6.023-6.009c0-3.313-2.702-6.010-6.023-6.010zM16 20c-2.205 0-4-1.794-4-4s1.794-4 4-4c2.206 0 4 1.794 4 4s-1.794 4-4 4z"
+     id="path1"
+     style="fill:#ffffff" />
+</svg>

bin/resources/fullscreenui/back-icon.svg

@@ -0,0 +1,7 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg fill="#ffffff" version="1.1" id="Layer_1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" width="800px" height="800px" viewBox="0 0 72 72" enable-background="new 0 0 72 72" xml:space="preserve">
+
+<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
+<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>

bin/resources/fullscreenui/desktop-mode.svg

@@ -0,0 +1,7 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg fill="#ffffff" width="800px" height="800px" viewBox="0 0 16 16" id="window-16px" xmlns="http://www.w3.org/2000/svg">
+
+<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
+<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>

bin/resources/fullscreenui/drive-cdrom.svg

@@ -0,0 +1,7 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg fill="#ffffff" height="800px" width="800px" version="1.1" id="Layer_1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" viewBox="0 0 512 512" xml:space="preserve">
+
+<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
+<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>

bin/resources/fullscreenui/exit.svg

@@ -0,0 +1,7 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg width="800px" height="800px" viewBox="0 0 24 24" id="SVGRoot" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:cc="http://creativecommons.org/ns#" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd" xmlns:svg="http://www.w3.org/2000/svg" fill="#ffffff">
+
+<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
+<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>

bin/resources/fullscreenui/game-list.svg

@@ -0,0 +1,7 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg height="800px" width="800px" version="1.1" id="_x32_" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" viewBox="0 0 512 512" xml:space="preserve" fill="#ffffff">
+
+<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
+<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>

bin/resources/fullscreenui/media-cdrom.svg

@@ -0,0 +1,7 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg fill="#ffffff" version="1.1" id="Layer_1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" width="800px" height="800px" viewBox="0 0 72 72" enable-background="new 0 0 72 72" xml:space="preserve">
+
+<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
+<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>

bin/resources/fullscreenui/start-bios.svg

@@ -0,0 +1,9 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg fill="#ffffff" width="800px" height="800px" viewBox="-2 -2 24 24" xmlns="http://www.w3.org/2000/svg" preserveAspectRatio="xMinYMin" class="jam jam-microchip">
+
+<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
+<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>
+
+<g id="SVGRepo_iconCarrier">

bin/resources/fullscreenui/start-file.svg

@@ -0,0 +1,7 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg width="800px" height="800px" viewBox="0 0 24 24" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" fill="#000000">
+
+<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
+<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>

bin/resources/game_controller_db.txt

@@ -329,6 +329,7 @@
 030000000d0f0000ee00000000000000,Horipad Mini 4,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a5,start:b9,x:b0,y:b3,platform:Windows,
 030000000d0f0000c100000000000000,Horipad Nintendo Switch Controller,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,misc1:b13,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b0,y:b3,platform:Windows,
 030000000d0f0000f600000000000000,Horipad Nintendo Switch Controller,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b2,y:b3,platform:Windows,
+030000000d0f00000202000000000000,Horipad O Nintendo Switch 2 Controller,a:b1,b:b0,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,misc1:b13,misc2:b14,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b3,y:b2,platform:Windows,
 030000000d0f00006700000000000000,Horipad One,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Windows,
 030000000d0f00009601000000000000,Horipad Steam,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:b8,leftx:a0,lefty:a1,misc2:b2,paddle1:b5,paddle2:b15,paddle3:b18,paddle4:b19,rightshoulder:b7,rightstick:b14,righttrigger:b9,rightx:a3,righty:a4,start:b11,x:b3,y:b4,platform:Windows,
 030000000d0f0000dc00000000000000,Horipad Switch,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b8,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b9,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Windows,
@@ -469,8 +470,10 @@
 030000001008000001e5000000000000,NEXT SNES Controller,a:b2,b:b1,back:b8,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,leftshoulder:b4,rightshoulder:b5,righttrigger:b6,start:b9,x:b3,y:b0,platform:Windows,
 03000000050b00000045000000000000,Nexus,a:b0,b:b1,back:b9,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b7,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b8,righttrigger:a4,rightx:a2,righty:a3,start:b10,x:b2,y:b3,platform:Windows,
 03000000152000000182000000000000,NGDS,a:b2,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a3,righty:a4,start:b9,x:b3,y:b0,platform:Windows,
+030000007e0500006920000000000000,Nintendo Switch 2 Pro Controller,a:b0,b:b1,back:b14,dpdown:b8,dpleft:b10,dpright:b9,dpup:b11,guide:b16,leftshoulder:b12,leftstick:b15,lefttrigger:b13,leftx:a0,lefty:a1~,misc1:b17,misc2:b20,paddle1:b18,paddle2:b19,rightshoulder:b4,rightstick:b7,righttrigger:b5,rightx:a2,righty:a3~,start:b6,x:b2,y:b3,platform:Windows,
 030000007e0500000920000000000000,Nintendo Switch Pro Controller,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,misc1:b13,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b2,y:b3,platform:Windows,
 030000000d0500000308000000000000,Nostromo N45,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b9,leftshoulder:b4,leftstick:b12,lefttrigger:b5,leftx:a0,lefty:a1,rightshoulder:b6,rightstick:b11,righttrigger:b7,rightx:a3,righty:a2,start:b10,x:b2,y:b3,platform:Windows,
+030000007e0500007320000000000000,NSO GameCube Controller,a:b1,b:b3,dpdown:b8,dpleft:b10,dpright:b9,dpup:b11,guide:b16,leftshoulder:b13,lefttrigger:b12,leftx:a0,lefty:a1~,rightshoulder:b5,righttrigger:b4,rightx:a2,righty:a3~,start:b6,x:b0,y:b2,platform:Windows,
 030000007e0500001920000000000000,NSO N64 Controller,+rightx:b8,+righty:b2,-rightx:b3,-righty:b7,a:b1,b:b0,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,lefttrigger:b6,leftx:a0,lefty:a1,misc1:b13,rightshoulder:b5,righttrigger:b10,start:b9,platform:Windows,
 030000007e0500001720000000000000,NSO SNES Controller,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,lefttrigger:b6,rightshoulder:b5,righttrigger:b15,start:b9,x:b2,y:b3,platform:Windows,
 03000000550900001472000000000000,NVIDIA Controller,a:b11,b:b10,back:b13,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b7,leftstick:b5,lefttrigger:a4,leftx:a0,lefty:a1,rightshoulder:b6,rightstick:b4,righttrigger:a5,rightx:a3,righty:a6,start:b3,x:b9,y:b8,platform:Windows,
@@ -1000,13 +1003,16 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000d62000007162000001000000,Moga Pro 2,a:b0,b:b1,back:b9,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b7,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b8,righttrigger:a4,rightx:a2,righty:a3,start:b6,x:b2,y:b3,platform:Mac OS X,
 03000000c62400002a89000000010000,MOGA XP5A Plus,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b21,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Mac OS X,
 03000000c62400002b89000000010000,MOGA XP5A Plus,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Mac OS X,
+03000000853200008906000000010000,Nacon Revolution X Unlimited,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Mac OS X,
 03000000632500007505000000020000,NeoGeo mini PAD Controller,a:b1,b:b0,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,start:b9,x:b2,y:b3,platform:Mac OS X,
 03000000921200004b46000003020000,NES 2-port Adapter,a:b1,b:b0,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,start:b11,platform:Mac OS X,
 030000001008000001e5000006010000,NEXT SNES Controller,a:b2,b:b1,back:b8,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,leftshoulder:b4,rightshoulder:b5,righttrigger:b6,start:b9,x:b3,y:b0,platform:Mac OS X,
+030000007e0500006920000001010000,Nintendo Switch 2 Pro Controller,a:b0,b:b1,back:b14,dpdown:b8,dpleft:b10,dpright:b9,dpup:b11,guide:b16,leftshoulder:b12,leftstick:b15,lefttrigger:b13,leftx:a0,lefty:a1~,misc1:b17,misc2:b20,paddle1:b18,paddle2:b19,rightshoulder:b4,rightstick:b7,righttrigger:b5,rightx:a2,righty:a3~,start:b6,x:b2,y:b3,platform:Mac OS X,
 030000007e0500000920000000000000,Nintendo Switch Pro Controller,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b2,y:b3,platform:Mac OS X,
 030000007e0500000920000001000000,Nintendo Switch Pro Controller,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b2,y:b3,platform:Mac OS X,
 030000007e0500000920000010020000,Nintendo Switch Pro Controller,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b9,leftshoulder:b4,leftstick:b6,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b7,righttrigger:a5,rightx:a3,righty:a4,start:b10,x:b2,y:b3,platform:Mac OS X,
 050000007e05000009200000ff070000,Nintendo Switch Pro Controller,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b9,leftshoulder:b4,leftstick:b6,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b7,righttrigger:a5,rightx:a3,righty:a4,start:b10,x:b2,y:b3,platform:Mac OS X,
+030000007e0500007320000001010000,NSO GameCube Controller,a:b1,b:b3,dpdown:b8,dpleft:b10,dpright:b9,dpup:b11,guide:b16,leftshoulder:b13,lefttrigger:b12,leftx:a0,lefty:a1~,rightshoulder:b5,righttrigger:b4,rightx:a2,righty:a3~,start:b6,x:b0,y:b2,platform:Mac OS X,
 030000007e0500001920000001000000,NSO N64 Controller,+rightx:b8,+righty:b7,-rightx:b3,-righty:b2,a:b1,b:b0,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,lefttrigger:b6,leftx:a0,lefty:a1,misc1:b13,rightshoulder:b5,righttrigger:b10,start:b9,platform:Mac OS X,
 030000007e0500001720000001000000,NSO SNES Controller,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,lefttrigger:b6,rightshoulder:b5,righttrigger:b15,start:b9,x:b2,y:b3,platform:Mac OS X,
 03000000550900001472000025050000,NVIDIA Controller,a:b0,b:b1,back:b17,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b15,leftshoulder:b4,leftstick:b7,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b8,righttrigger:a4,rightx:a2,righty:a5,start:b6,x:b2,y:b3,platform:Mac OS X,
@@ -1481,6 +1487,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 060000007e0500003713000000000000,Nintendo 3DS,a:b0,b:b1,back:b8,dpdown:b11,dpleft:b12,dpright:b13,dpup:b10,leftshoulder:b4,leftx:a0,lefty:a1,rightshoulder:b5,rightx:a2,righty:a3,start:b9,x:b3,y:b2,platform:Linux,
 030000007e0500003703000000016800,Nintendo GameCube Controller,a:b0,b:b2,dpdown:b6,dpleft:b4,dpright:b5,dpup:b7,lefttrigger:a4,leftx:a0,lefty:a1~,rightshoulder:b9,righttrigger:a5,rightx:a2,righty:a3~,start:b8,x:b1,y:b3,platform:Linux,
 03000000790000004618000010010000,Nintendo GameCube Controller Adapter,a:b1,b:b0,dpdown:b14,dpleft:b15,dpright:b13,dpup:b12,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b7,righttrigger:b5,rightx:a5~,righty:a2~,start:b9,x:b2,y:b3,platform:Linux,
+030000007e0500006920000011010000,Nintendo Switch 2 Pro Controller,a:b0,b:b1,back:b14,dpdown:b8,dpleft:b10,dpright:b9,dpup:b11,guide:b16,leftshoulder:b12,leftstick:b15,lefttrigger:b13,leftx:a0,lefty:a1~,misc1:b17,misc2:b20,paddle1:b18,paddle2:b19,rightshoulder:b4,rightstick:b7,righttrigger:b5,rightx:a2,righty:a3~,start:b6,x:b2,y:b3,platform:Linux,
 060000004e696e74656e646f20537700,Nintendo Switch Combined Joy-Cons,a:b0,b:b1,back:b9,dpdown:b15,dpleft:b16,dpright:b17,dpup:b14,guide:b11,leftshoulder:b5,leftstick:b12,lefttrigger:b7,leftx:a0,lefty:a1,misc1:b4,rightshoulder:b6,rightstick:b13,righttrigger:b8,rightx:a2,righty:a3,start:b10,x:b3,y:b2,platform:Linux,
 060000007e0500000620000000000000,Nintendo Switch Combined Joy-Cons,a:b0,b:b1,back:b9,dpdown:b15,dpleft:b16,dpright:b17,dpup:b14,guide:b11,leftshoulder:b5,leftstick:b12,lefttrigger:b7,leftx:a0,lefty:a1,misc1:b4,rightshoulder:b6,rightstick:b13,righttrigger:b8,rightx:a2,righty:a3,start:b10,x:b3,y:b2,platform:Linux,
 060000007e0500000820000000000000,Nintendo Switch Combined Joy-Cons,a:b0,b:b1,back:b9,dpdown:b15,dpleft:b16,dpright:b17,dpup:b14,guide:b11,leftshoulder:b5,leftstick:b12,lefttrigger:b7,leftx:a0,lefty:a1,misc1:b4,rightshoulder:b6,rightstick:b13,righttrigger:b8,rightx:a2,righty:a3,start:b10,x:b3,y:b2,platform:Linux,
@@ -1495,6 +1502,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 050000007e0500003003000001000000,Nintendo Wii U Pro Controller,a:b0,b:b1,back:b8,dpdown:b14,dpleft:b15,dpright:b16,dpup:b13,guide:b10,leftshoulder:b4,leftstick:b11,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b12,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b3,y:b2,platform:Linux,
 050000005a1d00000218000003000000,Nokia GC 5000,a:b9,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 030000000d0500000308000010010000,Nostromo n45 Dual Analog,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b9,leftshoulder:b4,leftstick:b12,lefttrigger:b5,leftx:a0,lefty:a1,rightshoulder:b6,rightstick:b11,righttrigger:b7,rightx:a3,righty:a2,start:b10,x:b2,y:b3,platform:Linux,
+030000007e0500007320000011010000,NSO GameCube Controller,a:b1,b:b3,dpdown:b8,dpleft:b10,dpright:b9,dpup:b11,guide:b16,leftshoulder:b13,lefttrigger:b12,leftx:a0,lefty:a1~,rightshoulder:b5,righttrigger:b4,rightx:a2,righty:a3~,start:b6,x:b0,y:b2,platform:Linux,
 030000007e0500001920000011810000,NSO N64 Controller,+rightx:b2,+righty:b3,-rightx:b4,-righty:b10,a:b0,b:b1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,lefttrigger:b8,leftx:a0,lefty:a1,misc1:b5,rightshoulder:b7,righttrigger:b9,start:b11,platform:Linux,
 050000007e0500001920000001000000,NSO N64 Controller,+rightx:b8,+righty:b7,-rightx:b3,-righty:b2,a:b1,b:b0,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,lefttrigger:b6,leftx:a0,lefty:a1,misc1:b13,rightshoulder:b5,righttrigger:b10,start:b9,platform:Linux,
 050000007e0500001920000001800000,NSO N64 Controller,+rightx:b2,+righty:b3,-rightx:b4,-righty:b10,a:b0,b:b1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,lefttrigger:b8,leftx:a0,lefty:a1,misc1:b5,rightshoulder:b7,righttrigger:b9,start:b11,platform:Linux,
@@ -1607,6 +1615,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 030000009b2800003200000001010000,Raphnet GC and N64 Adapter,a:b0,b:b7,dpdown:b11,dpleft:b12,dpright:b13,dpup:b10,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b2,righttrigger:b5,rightx:a3,righty:a4,start:b3,x:b1,y:b8,platform:Linux,
 030000009b2800006000000001010000,Raphnet GC and N64 Adapter,a:b0,b:b7,dpdown:b11,dpleft:b12,dpright:b13,dpup:b10,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b2,righttrigger:b5,rightx:a3,righty:a4,start:b3,x:b1,y:b8,platform:Linux,
 030000009b2800006100000001010000,Raphnet N64 Adapter,+rightx:b9,+righty:b7,-rightx:b8,-righty:b6,a:b0,b:b1,dpdown:b11,dpleft:b12,dpright:b13,dpup:b10,leftshoulder:b4,lefttrigger:b2,leftx:a0,lefty:a1,rightshoulder:b5,start:b3,platform:Linux,
+030000009b2800006400000001010000,Raphnet N64 Adapter,+rightx:b9,+righty:b7,-rightx:b8,-righty:b6,a:b0,b:b1,dpdown:b11,dpleft:b12,dpright:b13,dpup:b10,leftshoulder:b4,lefttrigger:b2,leftx:a0,lefty:a1,rightshoulder:b5,start:b3,platform:Linux,
 030000009b2800008000000020020000,Raphnet Wii Classic Adapter,a:b1,b:b4,back:b2,dpdown:b13,dpleft:b14,dpright:b15,dpup:b12,leftshoulder:b6,rightshoulder:b7,start:b3,x:b0,y:b5,platform:Linux,
 030000009b2800008000000001010000,Raphnet Wii Classic Adapter V3,a:b1,b:b4,back:b2,dpdown:b13,dpleft:b14,dpright:b15,dpup:b12,leftshoulder:b6,rightshoulder:b7,start:b3,x:b0,y:b5,platform:Linux,
 03000000f8270000bf0b000011010000,Razer Kishi,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,

bin/resources/shaders/dx11/convert.fx

@@ -110,48 +110,24 @@ uint ps_convert_rgba8_16bits(PS_INPUT input) : SV_Target0
 	return ((i.x & 0x00F8u) >> 3) | ((i.y & 0x00F8u) << 2) | ((i.z & 0x00f8u) << 7) | ((i.w & 0x80u) << 8);
 }
 
-PS_OUTPUT ps_datm1(PS_INPUT input)
+void ps_datm1(PS_INPUT input)
 {
-	PS_OUTPUT output;
-	
 	clip(sample_c(input.t).a - 127.5f / 255); // >= 0x80 pass
-	
-	output.c = 0;
-
-	return output;
 }
 
-PS_OUTPUT ps_datm0(PS_INPUT input)
+void ps_datm0(PS_INPUT input)
 {
-	PS_OUTPUT output;
-	
 	clip(127.5f / 255 - sample_c(input.t).a); // < 0x80 pass (== 0x80 should not pass)
-	
-	output.c = 0;
-
-	return output;
 }
 
-PS_OUTPUT ps_datm1_rta_correction(PS_INPUT input)
+void ps_datm1_rta_correction(PS_INPUT input)
 {
-	PS_OUTPUT output;
-
 	clip(sample_c(input.t).a - 254.5f / 255); // >= 0x80 pass
-
-	output.c = 0;
-
-	return output;
 }
 
-PS_OUTPUT ps_datm0_rta_correction(PS_INPUT input)
+void ps_datm0_rta_correction(PS_INPUT input)
 {
-	PS_OUTPUT output;
-
 	clip(254.5f / 255 - sample_c(input.t).a); // < 0x80 pass (== 0x80 should not pass)
-
-	output.c = 0;
-
-	return output;
 }
 
 PS_OUTPUT ps_rta_correction(PS_INPUT input)
@@ -313,6 +289,127 @@ float ps_convert_rgb5a1_float16_biln(PS_INPUT input) : SV_Depth
 	SAMPLE_RGBA_DEPTH_BILN(rgb5a1_to_depth16);
 }
 
+PS_OUTPUT ps_convert_rgb5a1_8i(PS_INPUT input)
+{
+	PS_OUTPUT output;
+
+	// Convert a RGB5A1 texture into a 8 bits packed texture
+	// Input column: 16x2 RGB5A1 pixels
+	// 0: 16 RGBA
+	// 1: 16 RGBA
+	// Output column: 16x4 Index pixels
+	// 0: 16 R5G2
+	// 1: 16 R5G2
+	// 2: 16 G2B5A1
+	// 3: 16 G2B5A1
+	uint2 pos = uint2(input.p.xy);
+
+	// Collapse separate R G B A areas into their base pixel
+	uint2 column = (pos & ~uint2(0u, 3u)) / uint2(1,2);
+	uint2 subcolumn = (pos & uint2(0u, 1u));
+	column.x -= (column.x / 128) * 64;
+	column.y += (column.y / 32) * 32;
+	
+	uint PSM = uint(DOFFSET);
+	
+	// Deal with swizzling differences
+	if ((PSM & 0x8) != 0) // PSMCT16S
+	{
+		if ((pos.x & 32) != 0)
+		{
+			column.y += 32; // 4 columns high times 4 to get bottom 4 blocks
+			column.x &= ~32;
+		}
+		
+		if ((pos.x & 64) != 0)
+		{
+			column.x -= 32;
+		}
+		
+		if (((pos.x & 16) != 0) != ((pos.y & 16) != 0))
+		{
+			column.x ^= 16; 
+			column.y ^= 8;
+		}
+		
+		if ((PSM & 0x30) != 0) // PSMZ16S - Untested but hopefully ok if anything uses it.
+		{
+			column.x ^= 32;
+			column.y ^= 16;
+		}
+	}
+	else // PSMCT16
+	{
+		if ((pos.y & 32) != 0)
+		{
+			column.y -= 16;
+			column.x += 32;
+		}
+		
+		if ((pos.x & 96) != 0)
+		{
+			uint multi = (pos.x & 96) / 32;
+			column.y += 16 * multi; // 4 columns high times 4 to get bottom 4 blocks
+			column.x -= (pos.x & 96);
+		}
+		
+		if (((pos.x & 16) != 0) != ((pos.y & 16) != 0))
+		{
+			column.x ^= 16; 
+			column.y ^= 8;
+		}
+		
+		if ((PSM & 0x30) != 0) // PSMZ16 - Untested but hopefully ok if anything uses it.
+		{
+			column.x ^= 32;
+			column.y ^= 32;
+		}
+	}
+	
+	uint2 coord = column | subcolumn;
+
+	// Compensate for potentially differing page pitch.
+	uint SBW = uint(EMODA);
+	uint DBW = uint(EMODC);
+	uint2 block_xy = coord / uint2(64,64);
+	uint block_num = (block_xy.y * (DBW / 128)) + block_xy.x;
+	uint2 block_offset = uint2((block_num % (SBW / 64)) * 64, (block_num / (SBW / 64)) * 64);
+	coord = (coord % uint2(64, 64)) + block_offset;
+
+	// Apply offset to cols 1 and 2
+	uint is_col23 = pos.y & 4u;
+	uint is_col13 = pos.y & 2u;
+	uint is_col12 = is_col23 ^ (is_col13 << 1);
+	coord.x ^= is_col12; // If cols 1 or 2, flip bit 3 of x
+
+	float ScaleFactor = BGColor.x;
+	if (floor(ScaleFactor) != ScaleFactor)
+		coord = uint2(float2(coord) * ScaleFactor);
+	else
+		coord *= uint(ScaleFactor);
+
+	float4 pixel = Texture.Load(int3(int2(coord), 0));
+	uint4 denorm_c = (uint4)(pixel * 255.5f);
+	if ((pos.y & 2u) == 0u)
+	{
+		uint red = (denorm_c.r >> 3) & 0x1Fu;
+		uint green = (denorm_c.g >> 3) & 0x1Fu;
+		float sel0 = (float)(((green << 5) | red) & 0xFF) / 255.0f;
+		
+		output.c = (float4)(sel0);
+	}
+	else
+	{
+		uint green = (denorm_c.g >> 3) & 0x1Fu;
+		uint blue = (denorm_c.b >> 3) & 0x1Fu;
+		uint alpha = denorm_c.a & 0x80u;
+		float sel0 = (float)((alpha | (blue << 2) | (green >> 3)) & 0xFF) / 255.0f;
+
+		output.c = (float4)(sel0);
+	}
+	return output;
+}
+
 PS_OUTPUT ps_convert_rgba_8i(PS_INPUT input)
 {
 	PS_OUTPUT output;

bin/resources/shaders/opengl/convert.glsl

@@ -237,9 +237,131 @@ void ps_convert_rgb5a1_float16_biln()
 }
 #endif
 
+#ifdef ps_convert_rgb5a1_8i
+uniform uint SBW;
+uniform uint DBW;
+uniform uint PSM;
+uniform float ScaleFactor;
+
+void ps_convert_rgb5a1_8i()
+{
+	// Convert a RGB5A1 texture into a 8 bits packed texture
+	// Input column: 16x2 RGB5A1 pixels
+	// 0: 16 RGBA
+	// 1: 16 RGBA
+	// Output column: 16x4 Index pixels
+	// 0: 16 R5G2
+	// 1: 16 R5G2
+	// 2: 16 G2B5A1
+	// 3: 16 G2B5A1
+	
+	uvec2 pos = uvec2(gl_FragCoord.xy);
+
+	// Collapse separate R G B A areas into their base pixel
+	uvec2 column = (pos & ~uvec2(0u, 3u)) / uvec2(1,2);
+	uvec2 subcolumn = (pos & uvec2(0u, 1u));
+	column.x -= (column.x / 128) * 64;
+	column.y += (column.y / 32) * 32;
+	
+	// Deal with swizzling differences
+	if ((PSM & 0x8) != 0) // PSMCT16S
+	{
+		if ((pos.x & 32) != 0)
+		{
+			column.y += 32; // 4 columns high times 4 to get bottom 4 blocks
+			column.x &= ~32;
+		}
+		
+		if ((pos.x & 64) != 0)
+		{
+			column.x -= 32;
+		}
+		
+		if (((pos.x & 16) != 0) != ((pos.y & 16) != 0))
+		{
+			column.x ^= 16; 
+			column.y ^= 8;
+		}
+		
+		if ((PSM & 0x30) != 0) // PSMZ16S - Untested but hopefully ok if anything uses it.
+		{
+			column.x ^= 32;
+			column.y ^= 16;
+		}
+	}
+	else // PSMCT16
+	{
+		if ((pos.y & 32) != 0)
+		{
+			column.y -= 16;
+			column.x += 32;
+		}
+		
+		if ((pos.x & 96) != 0)
+		{
+			uint multi = (pos.x & 96) / 32;
+			column.y += 16 * multi; // 4 columns high times 4 to get bottom 4 blocks
+			column.x -= (pos.x & 96);
+		}
+		
+		if (((pos.x & 16) != 0) != ((pos.y & 16) != 0))
+		{
+			column.x ^= 16; 
+			column.y ^= 8;
+		}
+		
+		if ((PSM & 0x30) != 0) // PSMZ16 - Untested but hopefully ok if anything uses it.
+		{
+			column.x ^= 32;
+			column.y ^= 32;
+		}
+	}
+	uvec2 coord = column | subcolumn;
+	
+	// Compensate for potentially differing page pitch.
+	uvec2 block_xy = coord / uvec2(64u, 64u);
+	uint block_num = (block_xy.y * (DBW / 128u)) + block_xy.x;
+	uvec2 block_offset = uvec2((block_num % (SBW / 64u)) * 64u, (block_num / (SBW / 64u)) * 64u);
+	coord = (coord % uvec2(64u, 64u)) + block_offset;
+
+	// Apply offset to cols 1 and 2
+	uint is_col23 = pos.y & 4u;
+	uint is_col13 = pos.y & 2u;
+	uint is_col12 = is_col23 ^ (is_col13 << 1);
+	coord.x ^= is_col12; // If cols 1 or 2, flip bit 3 of x
+
+	if (floor(ScaleFactor) != ScaleFactor)
+		coord = uvec2(vec2(coord) * ScaleFactor);
+	else
+		coord *= uvec2(ScaleFactor);
+
+	vec4 pixel = texelFetch(TextureSampler, ivec2(coord), 0);
+	
+	uvec4 denorm_c = uvec4(pixel * 255.5f);
+	if ((pos.y & 2u) == 0u)
+	{
+		uint red = (denorm_c.r >> 3) & 0x1Fu;
+		uint green = (denorm_c.g >> 3) & 0x1Fu;
+		float sel0 = float(((green << 5) | red) & 0xFF) / 255.0f;
+		
+		SV_Target0 = vec4(sel0);
+	}
+	else
+	{
+		uint green = (denorm_c.g >> 3) & 0x1Fu;
+		uint blue = (denorm_c.b >> 3) & 0x1Fu;
+		uint alpha = denorm_c.a & 0x80u;
+		float sel0 = float((alpha | (blue << 2) | (green >> 3)) & 0xFF) / 255.0f;
+
+		SV_Target0 = vec4(sel0);
+	}
+}
+#endif
+
 #ifdef ps_convert_rgba_8i
 uniform uint SBW;
 uniform uint DBW;
+uniform uint PSM;
 uniform float ScaleFactor;
 
 void ps_convert_rgba_8i()

bin/resources/shaders/vulkan/convert.glsl

@@ -104,7 +104,6 @@ void ps_datm1()
 {
 	if(sample_c(v_tex).a < (127.5f / 255.0f)) // >= 0x80 pass
 		discard;
-
 }
 #endif
 
@@ -307,12 +306,138 @@ void ps_convert_rgb5a1_float16_biln()
 }
 #endif
 
+#ifdef ps_convert_rgb5a1_8i
+layout(push_constant) uniform cb10
+{
+	uint SBW;
+	uint DBW;
+	uint PSM;
+	float cb_pad1;
+	float ScaleFactor;
+	vec3 cb_pad2;
+};
+
+void ps_convert_rgb5a1_8i()
+{
+	// Convert a RGB5A1 texture into a 8 bits packed texture
+	// Input column: 16x2 RGB5A1 pixels
+	// 0: 16 RGBA
+	// 1: 16 RGBA
+	// Output column: 16x4 Index pixels
+	// 0: 16 R5G2
+	// 1: 16 R5G2
+	// 2: 16 G2B5A1
+	// 3: 16 G2B5A1
+	
+	uvec2 pos = uvec2(gl_FragCoord.xy);
+
+	// Collapse separate R G B A areas into their base pixel
+	uvec2 column = (pos & ~uvec2(0u, 3u)) / uvec2(1,2);
+	uvec2 subcolumn = (pos & uvec2(0u, 1u));
+	column.x -= (column.x / 128) * 64;
+	column.y += (column.y / 32) * 32;
+	
+	// Deal with swizzling differences
+	if ((PSM & 0x8) != 0) // PSMCT16S
+	{
+		if ((pos.x & 32) != 0)
+		{
+			column.y += 32; // 4 columns high times 4 to get bottom 4 blocks
+			column.x &= ~32;
+		}
+		
+		if ((pos.x & 64) != 0)
+		{
+			column.x -= 32;
+		}
+		
+		if (((pos.x & 16) != 0) != ((pos.y & 16) != 0))
+		{
+			column.x ^= 16; 
+			column.y ^= 8;
+		}
+		
+		if ((PSM & 0x30) != 0) // PSMZ16S - Untested but hopefully ok if anything uses it.
+		{
+			column.x ^= 32;
+			column.y ^= 16;
+		}
+	}
+	else // PSMCT16
+	{
+		if ((pos.y & 32) != 0)
+		{
+			column.y -= 16;
+			column.x += 32;
+		}
+		
+		if ((pos.x & 96) != 0)
+		{
+			uint multi = (pos.x & 96) / 32;
+			column.y += 16 * multi; // 4 columns high times 4 to get bottom 4 blocks
+			column.x -= (pos.x & 96);
+		}
+		
+		if (((pos.x & 16) != 0) != ((pos.y & 16) != 0))
+		{
+			column.x ^= 16; 
+			column.y ^= 8;
+		}
+		
+		if ((PSM & 0x30) != 0) // PSMZ16 - Untested but hopefully ok if anything uses it.
+		{
+			column.x ^= 32;
+			column.y ^= 32;
+		}
+	}
+	uvec2 coord = column | subcolumn;
+	
+	// Compensate for potentially differing page pitch.
+	uvec2 block_xy = coord / uvec2(64u, 64u);
+	uint block_num = (block_xy.y * (DBW / 128u)) + block_xy.x;
+	uvec2 block_offset = uvec2((block_num % (SBW / 64u)) * 64u, (block_num / (SBW / 64u)) * 64u);
+	coord = (coord % uvec2(64u, 64u)) + block_offset;
+
+	// Apply offset to cols 1 and 2
+	uint is_col23 = pos.y & 4u;
+	uint is_col13 = pos.y & 2u;
+	uint is_col12 = is_col23 ^ (is_col13 << 1);
+	coord.x ^= is_col12; // If cols 1 or 2, flip bit 3 of x
+
+	if (floor(ScaleFactor) != ScaleFactor)
+		coord = uvec2(vec2(coord) * ScaleFactor);
+	else
+		coord *= uvec2(ScaleFactor);
+
+	vec4 pixel = texelFetch(samp0, ivec2(coord), 0);
+	uvec4 denorm_c = uvec4(pixel * 255.5f);
+	if ((pos.y & 2u) == 0u)
+	{
+		uint red = (denorm_c.r >> 3) & 0x1Fu;
+		uint green = (denorm_c.g >> 3) & 0x1Fu;
+		float sel0 = float(((green << 5) | red) & 0xFF) / 255.0f;
+		
+		o_col0 = vec4(sel0);
+	}
+	else
+	{
+		uint green = (denorm_c.g >> 3) & 0x1Fu;
+		uint blue = (denorm_c.b >> 3) & 0x1Fu;
+		uint alpha = denorm_c.a & 0x80u;
+		float sel0 = float((alpha | (blue << 2) | (green >> 3)) & 0xFF) / 255.0f;
+
+		o_col0 = vec4(sel0);
+	}
+}
+#endif
+
 #ifdef ps_convert_rgba_8i
 layout(push_constant) uniform cb10
 {
 	uint SBW;
 	uint DBW;
-	uvec2 cb_pad1;
+	uint PSM;
+	float cb_pad1;
 	float ScaleFactor;
 	vec3 cb_pad2;
 };

cmake/BuildParameters.cmake

@@ -7,7 +7,7 @@ include(GNUInstallDirs)
 # Misc option
 #-------------------------------------------------------------------------------
 option(ENABLE_TESTS "Enables building the unit tests" ON)
-option(ENABLE_GSRUNNER "Enables building the GSRunner" OFF)
+option(ENABLE_GSRUNNER "Enables building the GSRunner by default.  It can still be built with `make pcsx2-gsrunner` otherwise." OFF)
 option(LTO_PCSX2_CORE "Enable LTO/IPO/LTCG on the subset of pcsx2 that benefits most from it but not anything else")
 option(USE_VTUNE "Plug VTUNE to profile GS JIT.")
 option(PACKAGE_MODE "Use this option to ease packaging of PCSX2 (developer/distribution option)")

common/CocoaTools.h

@@ -31,6 +31,19 @@ namespace CocoaTools
 	bool DelayedLaunch(std::string_view file);
 	/// Open a Finder window to the given URL
 	bool ShowInFinder(std::string_view file);
+	/// Get the path to the resources directory of the current application
+	std::optional<std::string> GetResourcePath();
+
+	/// Create a window
+	void* CreateWindow(std::string_view title, uint32_t width, uint32_t height);
+	/// Destroy a window
+	void DestroyWindow(void* window);
+	/// Make a WindowInfo from the given window
+	void GetWindowInfoFromWindow(WindowInfo* wi, void* window);
+	/// Run cocoa event loop
+	void RunCocoaEventLoop(bool wait_forever = false);
+	/// Posts an event to the main telling `RunCocoaEventLoop(true)` to exit
+	void StopMainThreadEventLoop();
 }
 
 #endif // __APPLE__

common/CocoaTools.mm

@@ -224,5 +224,103 @@ bool CocoaTools::DelayedLaunch(std::string_view file)
 bool CocoaTools::ShowInFinder(std::string_view file)
 {
 	return [[NSWorkspace sharedWorkspace] selectFile:NSStringFromStringView(file)
-	                        inFileViewerRootedAtPath:nil];
+	                        inFileViewerRootedAtPath:@""];
 }
+
+std::optional<std::string> CocoaTools::GetResourcePath()
+{ @autoreleasepool {
+	if (NSBundle* bundle = [NSBundle mainBundle])
+	{
+		NSString* rsrc = [bundle resourcePath];
+		NSString* root = [bundle bundlePath];
+		if ([rsrc isEqualToString:root])
+			rsrc = [rsrc stringByAppendingString:@"/resources"];
+		return [rsrc UTF8String];
+	}
+	return std::nullopt;
+}}
+
+// MARK: - GSRunner
+
+void* CocoaTools::CreateWindow(std::string_view title, u32 width, u32 height)
+{
+	if (!NSApp)
+	{
+		[NSApplication sharedApplication];
+		[NSApp setActivationPolicy:NSApplicationActivationPolicyRegular];
+		[NSApp finishLaunching];
+	}
+	constexpr NSWindowStyleMask style = NSWindowStyleMaskTitled | NSWindowStyleMaskClosable | NSWindowStyleMaskMiniaturizable | NSWindowStyleMaskResizable;
+	NSScreen* mainScreen = [NSScreen mainScreen];
+	// Center the window on the screen, because why not
+	NSRect screenFrame = [mainScreen frame];
+	NSRect viewFrame = screenFrame;
+	viewFrame.size = NSMakeSize(width, height);
+	viewFrame.origin.x += (screenFrame.size.width - viewFrame.size.width) / 2;
+	viewFrame.origin.y += (screenFrame.size.height - viewFrame.size.height) / 2;
+	NSWindow* window = [[NSWindow alloc]
+		initWithContentRect:viewFrame
+		          styleMask:style
+		            backing:NSBackingStoreBuffered
+		              defer:NO];
+	[window setTitle:NSStringFromStringView(title)];
+	[window makeKeyAndOrderFront:window];
+	return (__bridge_retained void*)window;
+}
+
+void CocoaTools::DestroyWindow(void* window)
+{
+	(void)(__bridge_transfer NSWindow*)window;
+}
+
+void CocoaTools::GetWindowInfoFromWindow(WindowInfo* wi, void* cf_window)
+{
+	if (cf_window)
+	{
+		NSWindow* window = (__bridge NSWindow*)cf_window;
+		float scale = [window backingScaleFactor];
+		NSView* view = [window contentView];
+		NSRect dims = [view frame];
+		wi->type = WindowInfo::Type::MacOS;
+		wi->window_handle = (__bridge void*)view;
+		wi->surface_width = dims.size.width * scale;
+		wi->surface_height = dims.size.height * scale;
+		wi->surface_scale = scale;
+	}
+	else
+	{
+		wi->type = WindowInfo::Type::Surfaceless;
+	}
+}
+
+static constexpr short STOP_EVENT_LOOP = 0x100;
+
+void CocoaTools::RunCocoaEventLoop(bool forever)
+{
+	NSDate* end = forever ? [NSDate distantFuture] : [NSDate distantPast];
+	[NSApplication sharedApplication]; // Ensure NSApp is initialized
+	while (true)
+	{ @autoreleasepool {
+		NSEvent* ev = [NSApp nextEventMatchingMask:NSEventMaskAny
+		                                 untilDate:end
+		                                    inMode:NSDefaultRunLoopMode
+		                                   dequeue:YES];
+		if (!ev || ([ev type] == NSEventTypeApplicationDefined && [ev subtype] == STOP_EVENT_LOOP))
+			break;
+		[NSApp sendEvent:ev];
+	}}
+}
+
+void CocoaTools::StopMainThreadEventLoop()
+{ @autoreleasepool {
+	NSEvent* ev = [NSEvent otherEventWithType:NSEventTypeApplicationDefined
+	                                 location:{}
+	                            modifierFlags:0
+	                                timestamp:0
+	                             windowNumber:0
+	                                  context:nil
+	                                  subtype:STOP_EVENT_LOOP
+	                                    data1:0
+	                                    data2:0];
+	[NSApp postEvent:ev atStart:NO];
+}}

pcsx2-gsrunner/Main.cpp

@@ -16,6 +16,7 @@
 #include "fmt/format.h"
 
 #include "common/Assertions.h"
+#include "common/CocoaTools.h"
 #include "common/Console.h"
 #include "common/CrashHandler.h"
 #include "common/FileSystem.h"
@@ -56,7 +57,8 @@ namespace GSRunner
 	static bool CreatePlatformWindow();
 	static void DestroyPlatformWindow();
 	static std::optional<WindowInfo> GetPlatformWindowInfo();
-	static void PumpPlatformMessages();
+	static void PumpPlatformMessages(bool forever = false);
+	static void StopPlatformMessagePump();
 } // namespace GSRunner
 
 static constexpr u32 WINDOW_WIDTH = 640;
@@ -417,6 +419,16 @@ void Host::OnCreateMemoryCardOpenRequested()
 	// noop
 }
 
+bool Host::InBatchMode()
+{
+	return false;
+}
+
+bool Host::InNoGUIMode()
+{
+	return false;
+}
+
 bool Host::ShouldPreferHostFileSelector()
 {
 	return false;
@@ -532,7 +544,7 @@ bool GSRunner::ParseCommandLineArgs(int argc, char* argv[], VMBootParameters& pa
 			{
 				std::string str(argv[++i]);
 
-				s_settings_interface.SetBoolValue("EmuCore/GS", "dump", true);
+				s_settings_interface.SetBoolValue("EmuCore/GS", "DumpGSData", true);
 
 				if (str.find("rt") != std::string::npos)
 					s_settings_interface.SetBoolValue("EmuCore/GS", "SaveRT", true);
@@ -764,7 +776,7 @@ bool GSRunner::ParseCommandLineArgs(int argc, char* argv[], VMBootParameters& pa
 		return false;
 	}
 
-	if (s_settings_interface.GetBoolValue("EmuCore/GS", "dump") && !dumpdir.empty())
+	if (s_settings_interface.GetBoolValue("EmuCore/GS", "DumpGSData") && !dumpdir.empty())
 	{
 		if (s_settings_interface.GetStringValue("EmuCore/GS", "HWDumpDirectory").empty())
 			s_settings_interface.SetStringValue("EmuCore/GS", "HWDumpDirectory", dumpdir.c_str());
@@ -809,6 +821,22 @@ void GSRunner::DumpStats()
 #define main real_main
 #endif
 
+static void CPUThreadMain(VMBootParameters* params) {
+	if (VMManager::Initialize(*params))
+	{
+		// run until end
+		GSDumpReplayer::SetLoopCount(s_loop_count);
+		VMManager::SetState(VMState::Running);
+		while (VMManager::GetState() == VMState::Running)
+			VMManager::Execute();
+		VMManager::Shutdown(false);
+		GSRunner::DumpStats();
+	}
+
+	VMManager::Internal::CPUThreadShutdown();
+	GSRunner::StopPlatformMessagePump();
+}
+
 int main(int argc, char* argv[])
 {
 	CrashHandler::Install();
@@ -837,16 +865,9 @@ int main(int argc, char* argv[])
 	VMManager::ApplySettings();
 	GSDumpReplayer::SetIsDumpRunner(true);
 
-	if (VMManager::Initialize(params))
-	{
-		// run until end
-		GSDumpReplayer::SetLoopCount(s_loop_count);
-		VMManager::SetState(VMState::Running);
-		while (VMManager::GetState() == VMState::Running)
-			VMManager::Execute();
-		VMManager::Shutdown(false);
-		GSRunner::DumpStats();
-	}
+	std::thread cputhread(CPUThreadMain, &params);
+	GSRunner::PumpPlatformMessages(/*forever=*/true);
+	cputhread.join();
 
 	VMManager::Internal::CPUThreadShutdown();
 	GSRunner::DestroyPlatformWindow();
@@ -859,9 +880,6 @@ void Host::PumpMessagesOnCPUThread()
 	// update GS thread copy of frame number
 	MTGS::RunOnGSThread([frame_number = GSDumpReplayer::GetFrameNumber()]() { s_dump_frame_number = frame_number; });
 	MTGS::RunOnGSThread([loop_number = GSDumpReplayer::GetLoopCount()]() { s_loop_number = loop_number; });
-
-	// process any window messages (but we shouldn't really have any)
-	GSRunner::PumpPlatformMessages();
 }
 
 s32 Host::Internal::GetTranslatedStringImpl(
@@ -975,16 +993,32 @@ std::optional<WindowInfo> GSRunner::GetPlatformWindowInfo()
 	return wi;
 }
 
-void GSRunner::PumpPlatformMessages()
+static constexpr int SHUTDOWN_MSG = WM_APP + 0x100;
+static DWORD MainThreadID;
+
+void GSRunner::PumpPlatformMessages(bool forever)
 {
 	MSG msg;
-	while (PeekMessageW(&msg, NULL, 0, 0, PM_REMOVE))
+	while (true)
 	{
-		TranslateMessage(&msg);
-		DispatchMessageW(&msg);
+		while (PeekMessageW(&msg, NULL, 0, 0, PM_REMOVE))
+		{
+			if (msg.message == SHUTDOWN_MSG)
+				return;
+			TranslateMessage(&msg);
+			DispatchMessageW(&msg);
+		}
+		if (!forever)
+			return;
+		WaitMessage();
 	}
 }
 
+void GSRunner::StopPlatformMessagePump()
+{
+	PostThreadMessageW(MainThreadID, SHUTDOWN_MSG, 0, 0);
+}
+
 LRESULT CALLBACK WndProc(HWND hwnd, UINT msg, WPARAM wParam, LPARAM lParam)
 {
 	return DefWindowProcW(hwnd, msg, wParam, lParam);
@@ -1003,7 +1037,51 @@ int wmain(int argc, wchar_t** argv)
 		u8_argptrs.push_back(u8_args[i].data());
 	u8_argptrs.push_back(nullptr);
 
+	MainThreadID = GetCurrentThreadId();
+
 	return real_main(argc, u8_argptrs.data());
 }
 
-#endif // _WIN32
+#elif defined(__APPLE__)
+
+static void* s_window;
+static WindowInfo s_wi;
+
+bool GSRunner::CreatePlatformWindow()
+{
+	pxAssertRel(!s_window, "Tried to create window when there already was one!");
+	s_window = CocoaTools::CreateWindow("PCSX2 GS Runner", WINDOW_WIDTH, WINDOW_HEIGHT);
+	CocoaTools::GetWindowInfoFromWindow(&s_wi, s_window);
+	PumpPlatformMessages();
+	return s_window;
+}
+
+void GSRunner::DestroyPlatformWindow()
+{
+	if (s_window) {
+		CocoaTools::DestroyWindow(s_window);
+		s_window = nullptr;
+	}
+}
+
+std::optional<WindowInfo> GSRunner::GetPlatformWindowInfo()
+{
+	WindowInfo wi;
+	if (s_window)
+		wi = s_wi;
+	else
+		wi.type = WindowInfo::Type::Surfaceless;
+	return wi;
+}
+
+void GSRunner::PumpPlatformMessages(bool forever)
+{
+	CocoaTools::RunCocoaEventLoop(forever);
+}
+
+void GSRunner::StopPlatformMessagePump()
+{
+	CocoaTools::StopMainThreadEventLoop();
+}
+
+#endif // _WIN32 / __APPLE__

pcsx2-qt/Debugger/Docking/DockManager.cpp

@@ -142,12 +142,12 @@ void DockManager::switchToLayout(DockLayout::Index layout_index, bool blink_tab)
 			layout.thaw();
 
 			int tab_index = static_cast<int>(layout_index);
-			if (m_menu_bar && tab_index >= 0)
+			if (tab_index >= 0 && m_menu_bar)
 				m_menu_bar->onCurrentLayoutChanged(layout_index);
 		}
 	}
 
-	if (blink_tab)
+	if (blink_tab && m_menu_bar)
 		m_menu_bar->startBlink(m_current_layout);
 }
 
@@ -512,7 +512,8 @@ void DockManager::newLayoutClicked()
 {
 	// The plus button has just been made the current tab, so set it back to the
 	// one corresponding to the current layout again.
-	m_menu_bar->onCurrentLayoutChanged(m_current_layout);
+	if (m_menu_bar)
+		m_menu_bar->onCurrentLayoutChanged(m_current_layout);
 
 	auto name_validator = [this](const QString& name) {
 		return !hasNameConflict(name, DockLayout::INVALID_INDEX);

pcsx2-qt/Debugger/Memory/MemorySearchView.cpp

@@ -311,7 +311,7 @@ void searchWorker(DebugInterface* cpu, std::vector<SearchResult>& searchResults,
 			const auto readValue = readValueAtAddress<T>(cpu, addr);
 
 			const bool doesMatch = handleSearchComparison(searchComparison, addr, &searchResult, searchValue, readValue);
-			if (!doesMatch)
+			if (doesMatch)
 				searchResult = MemorySearchView::SearchResult(addr, QVariant::fromValue(readValue), searchType);
 
 			return !doesMatch;

pcsx2-qt/GameList/GameListWidget.h

@@ -92,7 +92,7 @@ public Q_SLOTS:
 	void refreshGridCovers();
 
 protected:
-	void resizeEvent(QResizeEvent* event);
+	void resizeEvent(QResizeEvent* event) override;
 	bool event(QEvent* event) override;
 
 private:

pcsx2-qt/LogWindow.cpp

@@ -43,7 +43,7 @@ void LogWindow::updateSettings()
 {
 	std::unique_lock lock(s_log_mutex);
 
-	const bool new_enabled = Host::GetBaseBoolSettingValue("Logging", "EnableLogWindow", false) && !QtHost::InNoGUIMode();
+	const bool new_enabled = Host::GetBaseBoolSettingValue("Logging", "EnableLogWindow", false) && !Host::InNoGUIMode();
 	const bool attach_to_main = Host::GetBaseBoolSettingValue("Logging", "AttachLogWindowToMainWindow", true);
 	const bool curr_enabled = Log::IsHostOutputEnabled();
 

pcsx2-qt/MainWindow.cpp

@@ -1130,7 +1130,7 @@ bool MainWindow::shouldHideMainWindow() const
 	// NOTE: We can't use isRenderingToMain() here, because this happens post-fullscreen-switch.
 	return (Host::GetBoolSettingValue("UI", "HideMainWindowWhenRunning", false) && !g_emu_thread->shouldRenderToMain()) ||
 		   (g_emu_thread->shouldRenderToMain() && (isRenderingFullscreen() || m_is_temporarily_windowed)) ||
-		   QtHost::InNoGUIMode();
+		   Host::InNoGUIMode();
 }
 
 bool MainWindow::shouldMouseLock() const
@@ -1306,7 +1306,7 @@ bool MainWindow::requestShutdown(bool allow_confirm, bool allow_save_to_state, b
 	// reshow the main window during display updates, because otherwise fullscreen transitions and renderer switches
 	// would briefly show and then hide the main window. So instead, we do it on shutdown, here. Except if we're in
 	// batch mode, when we're going to exit anyway.
-	if (!isRenderingToMain() && isHidden() && !QtHost::InBatchMode() && !g_emu_thread->isRunningFullscreenUI())
+	if (!isRenderingToMain() && isHidden() && !Host::InBatchMode() && !g_emu_thread->isRunningFullscreenUI())
 		updateWindowState(true);
 
 	// Clear the VM valid state early. That way we can't do anything in the UI if we take a while to shut down.
@@ -2060,7 +2060,7 @@ void MainWindow::onVMStopped()
 	updateInputRecordingActions(false);
 
 	// If we're closing or in batch mode, quit the whole application now.
-	if (m_is_closing || QtHost::InBatchMode())
+	if (m_is_closing || Host::InBatchMode())
 	{
 		quit();
 		return;

pcsx2-qt/QtHost.cpp

@@ -602,7 +602,7 @@ void Host::CheckForSettingsChanges(const Pcsx2Config& old_config)
 
 bool EmuThread::shouldRenderToMain() const
 {
-	return !Host::GetBoolSettingValue("UI", "RenderToSeparateWindow", false) && !QtHost::InNoGUIMode();
+	return !Host::GetBoolSettingValue("UI", "RenderToSeparateWindow", false) && !Host::InNoGUIMode();
 }
 
 void EmuThread::toggleSoftwareRendering()
@@ -1268,7 +1268,7 @@ void Host::RequestVMShutdown(bool allow_confirm, bool allow_save_state, bool def
 
 		// This will probably call shutdownVM() again, but by the time it runs, we'll have already shut down
 		// and it'll be a noop.
-		if (QtHost::InBatchMode())
+		if (Host::InBatchMode())
 			QMetaObject::invokeMethod(g_main_window, "requestExit", Qt::QueuedConnection, Q_ARG(bool, false));
 	}
 }
@@ -1437,12 +1437,12 @@ void Host::CommitBaseSettingChanges()
 	}
 }
 
-bool QtHost::InBatchMode()
+bool Host::InBatchMode()
 {
 	return s_batch_mode;
 }
 
-bool QtHost::InNoGUIMode()
+bool Host::InNoGUIMode()
 {
 	return s_nogui_mode;
 }

pcsx2-qt/QtHost.h

@@ -245,12 +245,6 @@ namespace QtHost
 	/// Sets the icon theme, based on the current style (light/dark).
 	void SetIconThemeFromStyle();
 
-	/// Sets batch mode (exit after game shutdown).
-	bool InBatchMode();
-
-	/// Sets NoGUI mode (implys batch mode, does not display main window, exits on shutdown).
-	bool InNoGUIMode();
-
 	/// Returns true if the calling thread is the UI thread.
 	bool IsOnUIThread();
 

pcsx2/CMakeLists.txt

@@ -1237,9 +1237,16 @@ fixup_file_properties(PCSX2)
 force_include_last(PCSX2_FLAGS "/(usr|local)/include/?$")
 
 if (APPLE)
+	find_library(APPKIT_LIBRARY AppKit)
+	find_library(IOKIT_LIBRARY IOKit)
 	find_library(METAL_LIBRARY Metal)
 	find_library(QUARTZCORE_LIBRARY QuartzCore)
-	target_link_libraries(PCSX2_FLAGS INTERFACE ${METAL_LIBRARY} ${QUARTZCORE_LIBRARY})
+	target_link_libraries(PCSX2_FLAGS INTERFACE
+		${APPKIT_LIBRARY}
+		${IOKIT_LIBRARY}
+		${METAL_LIBRARY}
+		${QUARTZCORE_LIBRARY}
+	)
 endif()
 
 set_property(GLOBAL PROPERTY PCSX2_SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR})

pcsx2/DebugTools/DebugInterface.cpp

@@ -44,7 +44,7 @@ bool DebugInterface::m_pause_on_entry = false;
 
 bool DebugInterface::isAlive()
 {
-	return VMManager::HasValidVM() && g_FrameCount > 0;
+	return VMManager::HasValidVM();
 }
 
 bool DebugInterface::isCpuPaused()

pcsx2/GS/GSLocalMemory.cpp

@@ -446,6 +446,13 @@ std::vector<GSVector2i>* GSLocalMemory::GetPage2TileMap(const GIFRegTEX0& TEX0)
 	return p2t;
 }
 
+u32 GSLocalMemory::IsPageAlignedMasked(u32 psm, const GSVector4i& rc)
+{
+	const psm_t& psm_s = m_psm[psm];
+	const GSVector4i pgmsk = GSVector4i(psm_s.pgs).xyxy() - GSVector4i(1);
+	return ((rc & pgmsk) == GSVector4i::zero()).mask();
+}
+
 bool GSLocalMemory::IsPageAligned(u32 psm, const GSVector4i& rc)
 {
 	const psm_t& psm_s = m_psm[psm];

pcsx2/GS/GSLocalMemory.h

@@ -524,6 +524,7 @@ public:
 	GSPixelOffset4* GetPixelOffset4(const GIFRegFRAME& FRAME, const GIFRegZBUF& ZBUF);
 	std::vector<GSVector2i>* GetPage2TileMap(const GIFRegTEX0& TEX0);
 	static bool HasOverlap(u32 src_bp, u32 src_bw, u32 src_psm, GSVector4i src_rect, u32 dst_bp, u32 dst_bw, u32 dst_psm, GSVector4i dst_rect);
+	static u32 IsPageAlignedMasked(u32 psm, const GSVector4i& rc);
 	static bool IsPageAligned(u32 psm, const GSVector4i& rc);
 	static u32 GetStartBlockAddress(u32 bp, u32 bw, u32 psm, GSVector4i rect);
 	static u32 GetEndBlockAddress(u32 bp, u32 bw, u32 psm, GSVector4i rect);

pcsx2/GS/GSState.cpp

@@ -575,7 +575,7 @@ void GSState::GIFPackedRegHandlerSTQ(const GIFPackedReg* RESTRICT r)
 
 	// Vexx (character shadow)
 	// q = 0 (st also 0 on the first 16 vertices), setting it to 1.0f to avoid div by zero later
-	q = q.blend8(GSVector4i::cast(GSVector4::m_one), q == GSVector4i::zero());
+	q = q.blend8(GSVector4i::cast(GSVector4(FLT_MIN)), q == GSVector4i::zero());
 
 	// Suikoden 4
 	// creates some nan for Q. Let's avoid undefined behavior (See GIFRegHandlerRGBAQ)
@@ -671,7 +671,7 @@ void GSState::GIFPackedRegHandlerSTQRGBAXYZF2(const GIFPackedReg* RESTRICT r, u3
 		GSVector4i q = GSVector4i::loadl(&r[0].U64[1]);
 		const GSVector4i rgba = (GSVector4i::load<false>(&r[1]) & GSVector4i::x000000ff()).ps32().pu16();
 
-		q = q.blend8(GSVector4i::cast(GSVector4::m_one), q == GSVector4i::zero()); // see GIFPackedRegHandlerSTQ
+		q = q.blend8(GSVector4i::cast(GSVector4(FLT_MIN)), q == GSVector4i::zero()); // see GIFPackedRegHandlerSTQ
 
 		m_v.m[0] = st.upl64(rgba.upl32(q)); // TODO: only store the last one
 
@@ -705,7 +705,7 @@ void GSState::GIFPackedRegHandlerSTQRGBAXYZ2(const GIFPackedReg* RESTRICT r, u32
 		GSVector4i q = GSVector4i::loadl(&r[0].U64[1]);
 		const GSVector4i rgba = (GSVector4i::load<false>(&r[1]) & GSVector4i::x000000ff()).ps32().pu16();
 
-		q = q.blend8(GSVector4i::cast(GSVector4::m_one), q == GSVector4i::zero()); // see GIFPackedRegHandlerSTQ
+		q = q.blend8(GSVector4i::cast(GSVector4(FLT_MIN)), q == GSVector4i::zero()); // see GIFPackedRegHandlerSTQ
 
 		m_v.m[0] = st.upl64(rgba.upl32(q)); // TODO: only store the last one
 
@@ -4237,7 +4237,10 @@ void GSState::CalcAlphaMinMax(const int tex_alpha_min, const int tex_alpha_max)
 
 	if (IsCoverageAlpha())
 	{
-		min = 128;
+		// HW renderer doesn't currently support AA, so its min is 128.
+		// If we add AA support to the HW renderer, this will need to be changed.
+		// (Will probably only be supported with ROV/FBFetch so we would want to check for that.)
+		min = GSIsHardwareRenderer() ? 128 : 0;
 		max = 128;
 	}
 	else

pcsx2/GS/Renderers/Common/GSDevice.cpp

@@ -68,6 +68,7 @@ const char* shaderName(ShaderConvert value)
 		case ShaderConvert::DEPTH_COPY:             return "ps_depth_copy";
 		case ShaderConvert::DOWNSAMPLE_COPY:        return "ps_downsample_copy";
 		case ShaderConvert::RGBA_TO_8I:             return "ps_convert_rgba_8i";
+		case ShaderConvert::RGB5A1_TO_8I:           return "ps_convert_rgb5a1_8i";
 		case ShaderConvert::CLUT_4:                 return "ps_convert_clut_4";
 		case ShaderConvert::CLUT_8:                 return "ps_convert_clut_8";
 		case ShaderConvert::YUV:                    return "ps_yuv";

pcsx2/GS/Renderers/Common/GSDevice.h

@@ -44,6 +44,7 @@ enum class ShaderConvert
 	DEPTH_COPY,
 	DOWNSAMPLE_COPY,
 	RGBA_TO_8I,
+	RGB5A1_TO_8I,
 	CLUT_4,
 	CLUT_8,
 	YUV,

pcsx2/GS/Renderers/Common/GSRenderer.cpp

@@ -617,72 +617,71 @@ void GSRenderer::VSync(u32 field, bool registers_written, bool idle_frame)
 			EndPresentFrame();
 
 		PerformanceMetrics::Update(registers_written, fb_sprite_frame, skip_frame);
-		return;
 	}
-
-	if (!idle_frame)
-		g_gs_device->AgePool();
-
-
-	g_perfmon.EndFrame(idle_frame);
-
-	if ((g_perfmon.GetFrame() & 0x1f) == 0)
-		g_perfmon.Update();
-
-	// Little bit ugly, but we can't do CAS inside the render pass.
-	GSVector4i src_rect;
-	GSVector4 src_uv, draw_rect;
-	GSTexture* current = g_gs_device->GetCurrent();
-	if (current && !blank_frame)
+	else
 	{
-		src_rect = CalculateDrawSrcRect(current, m_real_size);
-		src_uv = GSVector4(src_rect) / GSVector4(current->GetSize()).xyxy();
-		draw_rect = CalculateDrawDstRect(g_gs_device->GetWindowWidth(), g_gs_device->GetWindowHeight(),
-			src_rect, current->GetSize(), s_display_alignment, g_gs_device->UsesLowerLeftOrigin(),
-			GetVideoMode() == GSVideoMode::SDTV_480P);
-		s_last_draw_rect = draw_rect;
+		if (!idle_frame)
+			g_gs_device->AgePool();
 
-		if (GSConfig.CASMode != GSCASMode::Disabled)
-		{
-			static bool cas_log_once = false;
-			if (g_gs_device->Features().cas_sharpening)
-			{
-				// sharpen only if the IR is higher than the display resolution
-				const bool sharpen_only = (GSConfig.CASMode == GSCASMode::SharpenOnly ||
-										   (current->GetWidth() > g_gs_device->GetWindowWidth() &&
-											   current->GetHeight() > g_gs_device->GetWindowHeight()));
-				g_gs_device->CAS(current, src_rect, src_uv, draw_rect, sharpen_only);
-			}
-			else if (!cas_log_once)
-			{
-				Host::AddIconOSDMessage("CASUnsupported", ICON_FA_EXCLAMATION_TRIANGLE,
-					TRANSLATE_SV("GS",
-						"CAS is not available, your graphics driver does not support the required functionality."),
-					10.0f);
-				cas_log_once = true;
-			}
-		}
-	}
+		g_perfmon.EndFrame(idle_frame);
 
-	if (BeginPresentFrame(false))
-	{
+		if ((g_perfmon.GetFrame() & 0x1f) == 0)
+			g_perfmon.Update();
+
+		// Little bit ugly, but we can't do CAS inside the render pass.
+		GSVector4i src_rect;
+		GSVector4 src_uv, draw_rect;
+		GSTexture* current = g_gs_device->GetCurrent();
 		if (current && !blank_frame)
 		{
-			const u64 current_time = Common::Timer::GetCurrentValue();
-			const float shader_time = static_cast<float>(Common::Timer::ConvertValueToSeconds(current_time - m_shader_time_start));
+			src_rect = CalculateDrawSrcRect(current, m_real_size);
+			src_uv = GSVector4(src_rect) / GSVector4(current->GetSize()).xyxy();
+			draw_rect = CalculateDrawDstRect(g_gs_device->GetWindowWidth(), g_gs_device->GetWindowHeight(),
+				src_rect, current->GetSize(), s_display_alignment, g_gs_device->UsesLowerLeftOrigin(),
+				GetVideoMode() == GSVideoMode::SDTV_480P);
+			s_last_draw_rect = draw_rect;
 
-			g_gs_device->PresentRect(current, src_uv, nullptr, draw_rect,
-				s_tv_shader_indices[GSConfig.TVShader], shader_time, GSConfig.LinearPresent != GSPostBilinearMode::Off);
+			if (GSConfig.CASMode != GSCASMode::Disabled)
+			{
+				static bool cas_log_once = false;
+				if (g_gs_device->Features().cas_sharpening)
+				{
+					// sharpen only if the IR is higher than the display resolution
+					const bool sharpen_only = (GSConfig.CASMode == GSCASMode::SharpenOnly ||
+					                           (current->GetWidth() > g_gs_device->GetWindowWidth() &&
+					                            current->GetHeight() > g_gs_device->GetWindowHeight()));
+					g_gs_device->CAS(current, src_rect, src_uv, draw_rect, sharpen_only);
+				}
+				else if (!cas_log_once)
+				{
+					Host::AddIconOSDMessage("CASUnsupported", ICON_FA_EXCLAMATION_TRIANGLE,
+						TRANSLATE_SV("GS", "CAS is not available, your graphics driver does not support the required functionality."),
+						10.0f);
+					cas_log_once = true;
+				}
+			}
 		}
 
-		EndPresentFrame();
+		if (BeginPresentFrame(false))
+		{
+			if (current && !blank_frame)
+			{
+				const u64 current_time = Common::Timer::GetCurrentValue();
+				const float shader_time = static_cast<float>(Common::Timer::ConvertValueToSeconds(current_time - m_shader_time_start));
 
-		if (GSConfig.OsdShowGPU)
-			PerformanceMetrics::OnGPUPresent(g_gs_device->GetAndResetAccumulatedGPUTime());
+				g_gs_device->PresentRect(current, src_uv, nullptr, draw_rect,
+					s_tv_shader_indices[GSConfig.TVShader], shader_time, GSConfig.LinearPresent != GSPostBilinearMode::Off);
+			}
+
+			EndPresentFrame();
+
+			if (GSConfig.OsdShowGPU)
+				PerformanceMetrics::OnGPUPresent(g_gs_device->GetAndResetAccumulatedGPUTime());
+		}
+
+		PerformanceMetrics::Update(registers_written, fb_sprite_frame, false);
 	}
 
-	PerformanceMetrics::Update(registers_written, fb_sprite_frame, false);
-
 	// snapshot
 	if (!m_snapshot.empty())
 	{

pcsx2/GS/Renderers/DX11/GSDevice11.cpp

@@ -551,9 +551,18 @@ void GSDevice11::Destroy()
 	m_rs.reset();
 
 	if (m_state.rt_view)
+	{
 		m_state.rt_view->Release();
+		m_state.rt_view = nullptr;
+	}
+	m_state.cached_rt_view = nullptr;
+
 	if (m_state.dsv)
+	{
 		m_state.dsv->Release();
+		m_state.dsv = nullptr;
+	}
+	m_state.cached_dsv = nullptr;
 
 	m_shader_cache.Close();
 
@@ -950,11 +959,13 @@ GSDevice::PresentResult GSDevice11::BeginPresent(bool frame_skip)
 		m_state.rt_view->Release();
 	m_state.rt_view = m_swap_chain_rtv.get();
 	m_state.rt_view->AddRef();
+	m_state.cached_rt_view = nullptr;
 	if (m_state.dsv)
 	{
 		m_state.dsv->Release();
 		m_state.dsv = nullptr;
 	}
+	m_state.cached_dsv = nullptr;
 
 	g_perfmon.Put(GSPerfMon::RenderPasses, 1);
 
@@ -1105,14 +1116,14 @@ float GSDevice11::GetAndResetAccumulatedGPUTime()
 void GSDevice11::DrawPrimitive()
 {
 	g_perfmon.Put(GSPerfMon::DrawCalls, 1);
-	PSUpdateShaderState();
+	PSUpdateShaderState(true, true);
 	m_ctx->Draw(m_vertex.count, m_vertex.start);
 }
 
 void GSDevice11::DrawIndexedPrimitive()
 {
 	g_perfmon.Put(GSPerfMon::DrawCalls, 1);
-	PSUpdateShaderState();
+	PSUpdateShaderState(true, true);
 	m_ctx->DrawIndexed(m_index.count, m_index.start, m_vertex.start);
 }
 
@@ -1120,7 +1131,7 @@ void GSDevice11::DrawIndexedPrimitive(int offset, int count)
 {
 	pxAssert(offset + count <= (int)m_index.count);
 	g_perfmon.Put(GSPerfMon::DrawCalls, 1);
-	PSUpdateShaderState();
+	PSUpdateShaderState(true, true);
 	m_ctx->DrawIndexed(count, m_index.start + offset, m_vertex.start);
 }
 
@@ -1280,6 +1291,9 @@ void GSDevice11::StretchRect(GSTexture* sTex, const GSVector4& sRect, GSTexture*
 	GSVector2i ds;
 	if (dTex)
 	{
+		// ps unbind conflicting srvs
+		PSUnbindConflictingSRVs(dTex);
+
 		ds = dTex->GetSize();
 		if (draw_in_depth)
 			OMSetRenderTargets(nullptr, dTex);
@@ -1292,6 +1306,7 @@ void GSDevice11::StretchRect(GSTexture* sTex, const GSVector4& sRect, GSTexture*
 	}
 
 	// om
+
 	if (draw_in_depth)
 		OMSetDepthStencilState(m_convert.dss_write.get(), 0);
 	else
@@ -1329,7 +1344,7 @@ void GSDevice11::StretchRect(GSTexture* sTex, const GSVector4& sRect, GSTexture*
 	PSSetSamplerState(linear ? m_convert.ln.get() : m_convert.pt.get());
 	PSSetShader(ps, ps_cb);
 
-	//
+	// draw
 
 	DrawPrimitive();
 }
@@ -1341,6 +1356,9 @@ void GSDevice11::PresentRect(GSTexture* sTex, const GSVector4& sRect, GSTexture*
 	GSVector2i ds;
 	if (dTex)
 	{
+		// ps unbind conflicting srvs
+		PSUnbindConflictingSRVs(dTex);
+
 		ds = dTex->GetSize();
 		OMSetRenderTargets(dTex, nullptr);
 	}
@@ -1356,6 +1374,7 @@ void GSDevice11::PresentRect(GSTexture* sTex, const GSVector4& sRect, GSTexture*
 	m_ctx->UpdateSubresource(m_present.ps_cb.get(), 0, nullptr, &cb, 0, 0);
 
 	// om
+
 	OMSetDepthStencilState(m_convert.dss.get(), 0);
 	OMSetBlendState(m_convert.bs[D3D11_COLOR_WRITE_ENABLE_ALL].get(), 0);
 
@@ -1388,7 +1407,7 @@ void GSDevice11::PresentRect(GSTexture* sTex, const GSVector4& sRect, GSTexture*
 	PSSetSamplerState(linear ? m_convert.ln.get() : m_convert.pt.get());
 	PSSetShader(m_present.ps[static_cast<u32>(shader)].get(), m_present.ps_cb.get());
 
-	//
+	// draw
 
 	DrawPrimitive();
 }
@@ -1417,14 +1436,14 @@ void GSDevice11::ConvertToIndexedTexture(GSTexture* sTex, float sScale, u32 offs
 	{
 		float scale;
 		float pad1[3];
-		u32 SBW, DBW, pad3;
+		u32 SBW, DBW, SPSM;
 	};
 
-	const Uniforms cb = {sScale, {}, SBW, DBW};
+	const Uniforms cb = {sScale, {}, SBW, DBW, SPSM};
 	m_ctx->UpdateSubresource(m_merge.cb.get(), 0, nullptr, &cb, 0, 0);
 
 	const GSVector4 dRect(0, 0, dTex->GetWidth(), dTex->GetHeight());
-	const ShaderConvert shader = ShaderConvert::RGBA_TO_8I;
+	const ShaderConvert shader = ((SPSM & 0xE) == 0) ? ShaderConvert::RGBA_TO_8I : ShaderConvert::RGB5A1_TO_8I;
 	StretchRect(sTex, GSVector4::zero(), dTex, dRect, m_convert.ps[static_cast<int>(shader)].get(), m_merge.cb.get(), nullptr, false);
 }
 
@@ -1454,6 +1473,7 @@ void GSDevice11::DrawMultiStretchRects(const MultiStretchRect* rects, u32 num_re
 
 	VSSetShader(m_convert.vs.get(), nullptr);
 	PSSetShader(m_convert.ps[static_cast<int>(shader)].get(), nullptr);
+	PSUnbindConflictingSRVs(dTex);
 
 	OMSetDepthStencilState(dTex->IsRenderTarget() ? m_convert.dss.get() : m_convert.dss_write.get(), 0);
 	OMSetRenderTargets(dTex->IsRenderTarget() ? dTex : nullptr, dTex->IsDepthStencil() ? dTex : nullptr);
@@ -2110,7 +2130,7 @@ void GSDevice11::RenderImGui()
 
 			// Since we don't have the GSTexture...
 			m_state.ps_sr_views[0] = reinterpret_cast<ID3D11ShaderResourceView*>(pcmd->GetTexID());
-			PSUpdateShaderState();
+			PSUpdateShaderState(true, true);
 
 			m_ctx->DrawIndexed(pcmd->ElemCount, m_index.start + pcmd->IdxOffset, vertex_offset + pcmd->VtxOffset);
 		}
@@ -2130,6 +2150,10 @@ void GSDevice11::SetupDATE(GSTexture* rt, GSTexture* ds, const GSVertexPT1* vert
 
 	m_ctx->ClearDepthStencilView(*static_cast<GSTexture11*>(ds), D3D11_CLEAR_STENCIL, 0.0f, 0);
 
+	// ps unbind conflicting srvs
+
+	PSUnbindConflictingSRVs(ds);
+
 	// om
 
 	OMSetDepthStencilState(m_date.dss.get(), 1);
@@ -2147,11 +2171,12 @@ void GSDevice11::SetupDATE(GSTexture* rt, GSTexture* ds, const GSVertexPT1* vert
 	VSSetShader(m_convert.vs.get(), nullptr);
 
 	// ps
+
 	PSSetShaderResource(0, rt);
 	PSSetSamplerState(m_convert.pt.get());
 	PSSetShader(m_convert.ps[SetDATMShader(datm)].get(), nullptr);
 
-	//
+	// draw
 
 	DrawPrimitive();
 }
@@ -2357,15 +2382,68 @@ void GSDevice11::PSSetShader(ID3D11PixelShader* ps, ID3D11Buffer* ps_cb)
 	}
 }
 
-void GSDevice11::PSUpdateShaderState()
+void GSDevice11::PSUpdateShaderState(const bool sr_update, const bool ss_update)
 {
-	m_ctx->PSSetShaderResources(0, m_state.ps_sr_views.size(), m_state.ps_sr_views.data());
-	m_ctx->PSSetSamplers(0, m_state.ps_ss.size(), m_state.ps_ss.data());
+	// Shader resource caching requires srv/rtv hazards to be resolved, ensure PSUnbindConflictingSRVs handle.
+	if (sr_update)
+	{
+		bool sr_changed = false;
+		for (size_t i = 0; i < m_state.ps_sr_views.size(); ++i)
+		{
+			if (m_state.ps_cached_sr_views[i] != m_state.ps_sr_views[i])
+			{
+				sr_changed = true;
+				break;
+			}
+		}
+
+		if (sr_changed)
+		{
+			m_state.ps_cached_sr_views = m_state.ps_sr_views;
+			m_ctx->PSSetShaderResources(0, m_state.ps_sr_views.size(), m_state.ps_sr_views.data());
+		}
+	}
+
+	if (ss_update)
+	{
+		bool ss_changed = false;
+		for (size_t i = 0; i < m_state.ps_ss.size(); ++i)
+		{
+			if (m_state.ps_cached_ss[i] != m_state.ps_ss[i])
+			{
+				ss_changed = true;
+				break;
+			}
+		}
+
+		if (ss_changed)
+		{
+			m_state.ps_cached_ss = m_state.ps_ss;
+			m_ctx->PSSetSamplers(0, m_state.ps_ss.size(), m_state.ps_ss.data());
+		}
+	}
+}
+
+void GSDevice11::PSUnbindConflictingSRVs(GSTexture* tex1, GSTexture* tex2)
+{
+	// Make sure no SRVs are bound using the same texture before binding it to a RTV.
+	bool changed = false;
+	for (size_t i = 0; i < m_state.ps_sr_views.size(); i++)
+	{
+		if ((tex1 && m_state.ps_sr_views[i] == *(GSTexture11*)tex1) || (tex2 && m_state.ps_sr_views[i] == *(GSTexture11*)tex2))
+		{
+			m_state.ps_sr_views[i] = nullptr;
+			changed = true;
+		}
+	}
+
+	if (changed)
+		PSUpdateShaderState(true, false);
 }
 
 void GSDevice11::OMSetDepthStencilState(ID3D11DepthStencilState* dss, u8 sref)
 {
-	if (m_state.dss != dss || m_state.sref != sref)
+	if (m_state.dss != dss || (dss && m_state.sref != sref))
 	{
 		m_state.dss = dss;
 		m_state.sref = sref;
@@ -2376,7 +2454,7 @@ void GSDevice11::OMSetDepthStencilState(ID3D11DepthStencilState* dss, u8 sref)
 
 void GSDevice11::OMSetBlendState(ID3D11BlendState* bs, u8 bf)
 {
-	if (m_state.bs != bs || m_state.bf != bf)
+	if (m_state.bs != bs || (bs && m_state.bf != bf))
 	{
 		m_state.bs = bs;
 		m_state.bf = bf;
@@ -2413,6 +2491,7 @@ void GSDevice11::OMSetRenderTargets(GSTexture* rt, GSTexture* ds, const GSVector
 		if (rtv)
 			rtv->AddRef();
 		m_state.rt_view = rtv;
+		m_state.cached_rt_view = rt;
 	}
 	if (m_state.dsv != dsv)
 	{
@@ -2421,6 +2500,7 @@ void GSDevice11::OMSetRenderTargets(GSTexture* rt, GSTexture* ds, const GSVector
 		if (dsv)
 			dsv->AddRef();
 		m_state.dsv = dsv;
+		m_state.cached_dsv = ds;
 	}
 	if (changed)
 		m_ctx->OMSetRenderTargets(1, &rtv, dsv);
@@ -2523,18 +2603,18 @@ void GSDevice11::RenderHW(GSHWDrawConfig& config)
 		}
 	}
 
-	GSTexture* primid_tex = nullptr;
+	GSTexture* primid_texture = nullptr;
 	if (config.destination_alpha == GSHWDrawConfig::DestinationAlphaMode::PrimIDTracking)
 	{
-		primid_tex = CreateRenderTarget(rtsize.x, rtsize.y, GSTexture::Format::PrimID, false);
-		if (!primid_tex)
+		primid_texture = CreateRenderTarget(rtsize.x, rtsize.y, GSTexture::Format::PrimID, false);
+		if (!primid_texture)
 		{
-			Console.WriteLn("D3D11: Failed to allocate DATE image, aborting draw.");
+			Console.Warning("D3D11: Failed to allocate DATE image, aborting draw.");
 			return;
 		}
 
 		StretchRect(colclip_rt ? colclip_rt : config.rt, GSVector4(config.drawarea) / GSVector4(rtsize).xyxy(),
-			primid_tex, GSVector4(config.drawarea), m_date.primid_init_ps[static_cast<u8>(config.datm)].get(), nullptr, false);
+			primid_texture, GSVector4(config.drawarea), m_date.primid_init_ps[static_cast<u8>(config.datm)].get(), nullptr, false);
 	}
 	else if (config.destination_alpha == GSHWDrawConfig::DestinationAlphaMode::Stencil ||
 			 config.destination_alpha == GSHWDrawConfig::DestinationAlphaMode::StencilOne)
@@ -2596,6 +2676,9 @@ void GSDevice11::RenderHW(GSHWDrawConfig& config)
 	}
 	IASetPrimitiveTopology(topology);
 
+	// Should be called before changing local srv state.
+	PSUnbindConflictingSRVs(colclip_rt ? colclip_rt : config.rt, config.ds);
+
 	if (config.tex)
 	{
 		CommitClear(config.tex);
@@ -2645,24 +2728,44 @@ void GSDevice11::RenderHW(GSHWDrawConfig& config)
 	SetupVS(config.vs, &config.cb_vs);
 	SetupPS(config.ps, &config.cb_ps, config.sampler);
 
-	if (config.destination_alpha == GSHWDrawConfig::DestinationAlphaMode::PrimIDTracking)
+	if (primid_texture)
 	{
 		OMDepthStencilSelector dss = config.depth;
 		dss.zwe = 0;
 		const OMBlendSelector blend(GSHWDrawConfig::ColorMaskSelector(1),
 			GSHWDrawConfig::BlendState(true, CONST_ONE, CONST_ONE, 3 /* MIN */, CONST_ONE, CONST_ZERO, false, 0));
 		SetupOM(dss, blend, 0);
-		OMSetRenderTargets(primid_tex, config.ds, &config.scissor);
+		OMSetRenderTargets(primid_texture, config.ds, &config.scissor);
 		DrawIndexedPrimitive();
 
 		config.ps.date = 3;
 		config.alpha_second_pass.ps.date = 3;
 		SetupPS(config.ps, nullptr, config.sampler);
-		PSSetShaderResource(3, primid_tex);
 	}
 
+	// Avoid changing framebuffer just to switch from rt+depth to rt and vice versa.
+	GSTexture* draw_rt = colclip_rt ? colclip_rt : config.rt;
+	GSTexture* draw_ds = config.ds;
+	// Make sure no tex is bound as both rtv and srv at the same time.
+	// All conflicts should've been taken care of by PSUnbindConflictingSRVs.
+	// It is fine to do the optimiation when on slot 0 tex is fb, tex is ds, and slot 2 sw blend as they are copies bound to srv.
+	if (!draw_rt && draw_ds && m_state.rt_view && m_state.cached_rt_view && m_state.rt_view == *(GSTexture11*)m_state.cached_rt_view &&
+		m_state.cached_dsv == draw_ds && config.tex != m_state.cached_rt_view && m_state.cached_rt_view->GetSize() == draw_ds->GetSize())
+	{
+		draw_rt = m_state.cached_rt_view;
+	}
+	else if (!draw_ds && draw_rt && m_state.dsv && m_state.cached_dsv && m_state.dsv == *(GSTexture11*)m_state.cached_dsv &&
+		m_state.cached_rt_view == draw_rt && config.tex != m_state.cached_dsv && m_state.cached_dsv->GetSize() == draw_rt->GetSize())
+	{
+		draw_ds = m_state.cached_dsv;
+	}
+
+	OMSetRenderTargets(draw_rt, draw_ds, &config.scissor);
+
+	if (primid_texture)
+		PSSetShaderResource(3, primid_texture);
+
 	SetupOM(config.depth, OMBlendSelector(config.colormask, config.blend), config.blend.constant);
-	OMSetRenderTargets(colclip_rt ? colclip_rt : config.rt, config.ds, &config.scissor);
 	DrawIndexedPrimitive();
 
 	if (config.blend_multi_pass.enable)
@@ -2697,8 +2800,8 @@ void GSDevice11::RenderHW(GSHWDrawConfig& config)
 	if (draw_ds_clone)
 		Recycle(draw_ds_clone);
 
-	if (primid_tex)
-		Recycle(primid_tex);
+	if (primid_texture)
+		Recycle(primid_texture);
 
 	if (colclip_rt)
 	{

pcsx2/GS/Renderers/DX11/GSDevice11.h

@@ -145,9 +145,11 @@ private:
 		ID3D11VertexShader* vs;
 		ID3D11Buffer* vs_cb;
 		std::array<ID3D11ShaderResourceView*, MAX_TEXTURES> ps_sr_views;
+		std::array<ID3D11ShaderResourceView*, MAX_TEXTURES> ps_cached_sr_views;
 		ID3D11PixelShader* ps;
 		ID3D11Buffer* ps_cb;
 		std::array<ID3D11SamplerState*, MAX_SAMPLERS> ps_ss;
+		std::array<ID3D11SamplerState*, MAX_SAMPLERS> ps_cached_ss;
 		GSVector2i viewport;
 		GSVector4i scissor;
 		u32 vb_stride;
@@ -157,6 +159,8 @@ private:
 		u8 bf;
 		ID3D11RenderTargetView* rt_view;
 		ID3D11DepthStencilView* dsv;
+		GSTexture* cached_rt_view;
+		GSTexture* cached_dsv;
 	} m_state;
 
 	std::array<std::array<wil::com_ptr_nothrow<ID3D11Query>, 3>, NUM_TIMESTAMP_QUERIES> m_timestamp_queries = {};
@@ -324,7 +328,8 @@ public:
 
 	void PSSetShaderResource(int i, GSTexture* sr);
 	void PSSetShader(ID3D11PixelShader* ps, ID3D11Buffer* ps_cb);
-	void PSUpdateShaderState();
+	void PSUpdateShaderState(const bool sr_update, const bool ss_update);
+	void PSUnbindConflictingSRVs(GSTexture* tex1 = nullptr, GSTexture* tex2 = nullptr);
 	void PSSetSamplerState(ID3D11SamplerState* ss0);
 
 	void OMSetDepthStencilState(ID3D11DepthStencilState* dss, u8 sref);

pcsx2/GS/Renderers/DX12/GSDevice12.cpp

@@ -1486,15 +1486,15 @@ void GSDevice12::ConvertToIndexedTexture(
 	{
 		float scale;
 		float pad1[3];
-		u32 SBW, DBW, pad2;
+		u32 SBW, DBW, SPSM;
 	};
 
-	const Uniforms cb = {sScale, {}, SBW, DBW};
+	const Uniforms cb = {sScale, {}, SBW, DBW, SPSM};
 	SetUtilityRootSignature();
 	SetUtilityPushConstants(&cb, sizeof(cb));
 
 	const GSVector4 dRect(0, 0, dTex->GetWidth(), dTex->GetHeight());
-	const ShaderConvert shader = ShaderConvert::RGBA_TO_8I;
+	const ShaderConvert shader = ((SPSM & 0xE) == 0) ? ShaderConvert::RGBA_TO_8I : ShaderConvert::RGB5A1_TO_8I;
 	DoStretchRect(static_cast<GSTexture12*>(sTex), GSVector4::zero(), static_cast<GSTexture12*>(dTex), dRect,
 		m_convert[static_cast<int>(shader)].get(), false, true);
 }
@@ -3903,7 +3903,7 @@ void GSDevice12::RenderHW(GSHWDrawConfig& config)
 		config.rt = backup_rt;
 		if (!date_image)
 		{
-			Console.WriteLn("D3D12: Failed to allocate DATE image, aborting draw.");
+			Console.Warning("D3D12: Failed to allocate DATE image, aborting draw.");
 			return;
 		}
 	}

pcsx2/GS/Renderers/HW/GSHwHack.cpp

@@ -184,40 +184,6 @@ bool GSHwHack::GSC_NamcoGames(GSRendererHW& r, int& skip)
 {
 	if (skip == 0)
 	{
-		if (r.IsPossibleChannelShuffle() && !(RTBP0 & 31))
-		{
-			GSVertex* v = &r.m_vertex.buff[0];
-
-			// Make sure we're detecting the right effect.
-			if (((v[1].XYZ.X - v[0].XYZ.X) >> 4) != 8 || ((v[1].XYZ.Y - v[0].XYZ.Y) >> 4) != 14)
-				return false;
-
-			GSTextureCache::Target* rt = g_texture_cache->LookupTarget(GIFRegTEX0::Create(RTBP0, RFBW, RFPSM),
-				GSVector2i(1, 1), r.GetTextureScaleFactor(), GSTextureCache::RenderTarget);
-			if (!rt)
-				return false;
-
-			GL_INS("GSC_NamcoGames(): HLE channel shuffle");
-
-			// have to set up the palette ourselves too, since GSC executes before it does
-			r.m_mem.m_clut.Read32(RTEX0, r.m_draw_env->TEXA);
-			std::shared_ptr<GSTextureCache::Palette> palette =
-				g_texture_cache->LookupPaletteObject(r.m_mem.m_clut, GSLocalMemory::m_psm[RTEX0.PSM].pal, true);
-			if (!palette)
-				return false;
-
-			GSHWDrawConfig& conf = r.BeginHLEHardwareDraw(
-				rt->GetTexture(), nullptr, rt->GetScale(), rt->GetTexture(), rt->GetScale(), rt->GetUnscaledRect());
-			conf.pal = palette->GetPaletteGSTexture();
-			conf.ps.channel = ChannelFetch_RGB;
-			conf.colormask.wa = false;
-			r.EndHLEHardwareDraw(false);
-
-			// 12 pages: 2 calls by channel, 3 channels, 1 blit
-			skip = 12 * (3 + 3 + 1);
-			return true;
-		}
-
 		if (!s_nativeres && r.PRIM->PRIM == GS_SPRITE && RTME && RTEX0.TFX == 1 && RFPSM == RTPSM && RTPSM == PSMCT32 && RFBMSK == 0xFF000000 && r.m_index.tail > 2)
 		{
 			GSVertex* v = &r.m_vertex.buff[0];
@@ -431,31 +397,6 @@ bool GSHwHack::GSC_TalesOfLegendia(GSRendererHW& r, int& skip)
 	return true;
 }
 
-bool GSHwHack::GSC_Kunoichi(GSRendererHW& r, int& skip)
-{
-	if (skip == 0)
-	{
-		if (!RTME && (RFBP == 0x0 || RFBP == 0x00700 || RFBP == 0x00800) && RFPSM == PSMCT32 && RFBMSK == 0x00FFFFFF)
-		{
-			// Removes depth effects(shadows) not rendered correctly on all renders.
-			skip = 3;
-		}
-		if (RTME && (RFBP == 0x0700 || RFBP == 0) && RTBP0 == 0x0e00 && RTPSM == 0 && RFBMSK == 0)
-		{
-			skip = 1; // Removes black screen (not needed anymore maybe)?
-		}
-	}
-	else
-	{
-		if (RTME && (RFBP == 0x0e00) && RFPSM == PSMCT32 && RFBMSK == 0xFF000000)
-		{
-			skip = 0;
-		}
-	}
-
-	return true;
-}
-
 bool GSHwHack::GSC_ZettaiZetsumeiToshi2(GSRendererHW& r, int& skip)
 {
 	if (skip == 0)
@@ -505,27 +446,6 @@ bool GSHwHack::GSC_ZettaiZetsumeiToshi2(GSRendererHW& r, int& skip)
 	return true;
 }
 
-bool GSHwHack::GSC_SakuraWarsSoLongMyLove(GSRendererHW& r, int& skip)
-{
-	if (skip == 0)
-	{
-		if (RTME == 0 && RFBP != RTBP0 && RTBP0 && RFBMSK == 0x00FFFFFF)
-		{
-			skip = 3; // Remove darkness
-		}
-		else if (RTME == 0 && RFBP == RTBP0 && (RTBP0 == 0x1200 || RTBP0 == 0x1180 || RTBP0 == 0) && RFBMSK == 0x00FFFFFF)
-		{
-			skip = 3; // Remove darkness
-		}
-		else if (RTME && (RFBP == 0 || RFBP == 0x1180) && RFPSM == PSMCT32 && RTBP0 == 0x3F3F && RTPSM == PSMT8)
-		{
-			skip = 1; // Floodlight
-		}
-	}
-
-	return true;
-}
-
 bool GSHwHack::GSC_UltramanFightingEvolution(GSRendererHW& r, int& skip)
 {
 	if (skip == 0)
@@ -902,23 +822,6 @@ bool GSHwHack::GSC_MetalGearSolid3(GSRendererHW& r, int& skip)
 	return true;
 }
 
-bool GSHwHack::GSC_HitmanBloodMoney(GSRendererHW& r, int& skip)
-{
-	// The game does a stupid thing where it backs up the last 2 pages of the framebuffer with shuffles, uploads a CT32 texture to it
-	// then copies the RGB back (keeping the new alpha only). It's pretty gross, I have no idea why they didn't just upload a new alpha.
-	// This is a real pain to emulate with the current state of things, so let's just clear the dirty area from the upload and pretend it wasn't there.
-	
-	// Catch the first draw of the copy back.
-	if (RFBP > 0 && RTPSM == PSMT8H && RFPSM == PSMCT32)
-	{
-		GSTextureCache::Target* target = g_texture_cache->FindOverlappingTarget(RFBP, RFBP + 1);
-		if (target)
-			target->m_dirty.clear();
-	}
-
-	return false;
-}
-
 bool GSHwHack::OI_PointListPalette(GSRendererHW& r, GSTexture* rt, GSTexture* ds, GSTextureCache::Source* t)
 {
 	const u32 n_vertices = r.m_vertex.next;
@@ -1393,12 +1296,10 @@ bool GSHwHack::MV_Ico(GSRendererHW& r)
 #define CRC_F(name) { #name, &GSHwHack::name }
 
 const GSHwHack::Entry<GSRendererHW::GSC_Ptr> GSHwHack::s_get_skip_count_functions[] = {
-	CRC_F(GSC_Kunoichi),
 	CRC_F(GSC_Manhunt2),
 	CRC_F(GSC_MidnightClub3),
 	CRC_F(GSC_SacredBlaze),
 	CRC_F(GSC_GuitarHero),
-	CRC_F(GSC_SakuraWarsSoLongMyLove),
 	CRC_F(GSC_SFEX3),
 	CRC_F(GSC_DTGames),
 	CRC_F(GSC_TalesOfLegendia),
@@ -1410,7 +1311,6 @@ const GSHwHack::Entry<GSRendererHW::GSC_Ptr> GSHwHack::s_get_skip_count_function
 	CRC_F(GSC_NFSUndercover),
 	CRC_F(GSC_PolyphonyDigitalGames),
 	CRC_F(GSC_MetalGearSolid3),
-	CRC_F(GSC_HitmanBloodMoney),
 	CRC_F(GSC_Battlefield2),
 
 	// Channel Effect

pcsx2/GS/Renderers/HW/GSHwHack.h

@@ -16,9 +16,7 @@ public:
 	static bool GSC_BlackAndBurnoutSky(GSRendererHW& r, int& skip);
 	static bool GSC_MidnightClub3(GSRendererHW& r, int& skip);
 	static bool GSC_TalesOfLegendia(GSRendererHW& r, int& skip);
-	static bool GSC_Kunoichi(GSRendererHW& r, int& skip);
 	static bool GSC_ZettaiZetsumeiToshi2(GSRendererHW& r, int& skip);
-	static bool GSC_SakuraWarsSoLongMyLove(GSRendererHW& r, int& skip);
 	static bool GSC_UltramanFightingEvolution(GSRendererHW& r, int& skip);
 	static bool GSC_TalesofSymphonia(GSRendererHW& r, int& skip);
 	static bool GSC_UrbanReign(GSRendererHW& r, int& skip);
@@ -28,7 +26,6 @@ public:
 	static bool GSC_Battlefield2(GSRendererHW& r, int& skip);
 	static bool GSC_PolyphonyDigitalGames(GSRendererHW& r, int& skip);
 	static bool GSC_MetalGearSolid3(GSRendererHW& r, int& skip);
-	static bool GSC_HitmanBloodMoney(GSRendererHW& r, int& skip);
 
 	static bool OI_PointListPalette(GSRendererHW& r, GSTexture* rt, GSTexture* ds, GSTextureCache::Source* t);
 	static bool OI_DBZBTGames(GSRendererHW& r, GSTexture* rt, GSTexture* ds, GSTextureCache::Source* t);

pcsx2/GS/Renderers/HW/GSRendererHW.cpp

@@ -2230,6 +2230,7 @@ void GSRendererHW::Draw()
 	// We mess with this state as an optimization, so take a copy and use that instead.
 	const GSDrawingContext* context = m_context;
 	m_cached_ctx.TEX0 = context->TEX0;
+	m_cached_ctx.TEXA = m_draw_env->TEXA;
 	m_cached_ctx.CLAMP = context->CLAMP;
 	m_cached_ctx.TEST = context->TEST;
 	m_cached_ctx.FRAME = context->FRAME;
@@ -2359,11 +2360,10 @@ void GSRendererHW::Draw()
 	const u32 fm_mask = GSLocalMemory::m_psm[m_cached_ctx.FRAME.PSM].fmsk;
 
 	// Note required to compute TryAlphaTest below. So do it now.
-	const GSDrawingEnvironment& env = *m_draw_env;
 	const GSLocalMemory::psm_t& tex_psm = GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM];
 	if (PRIM->TME && tex_psm.pal > 0)
 	{
-		m_mem.m_clut.Read32(m_cached_ctx.TEX0, env.TEXA);
+		m_mem.m_clut.Read32(m_cached_ctx.TEX0, m_cached_ctx.TEXA);
 		if (m_mem.m_clut.GetGPUTexture())
 		{
 			CalcAlphaMinMax(0, 255);
@@ -2672,6 +2672,48 @@ void GSRendererHW::Draw()
 				GSVector2i(1, 1));
 			height_invalid = false;
 		}
+		const u32 vert_index = (m_vt.m_primclass == GS_TRIANGLE_CLASS) ? 2 : 1;
+		u32 const_color = m_vertex.buff[m_index.buff[vert_index]].RGBAQ.U32[0];
+		u32 fb_mask = m_cached_ctx.FRAME.FBMSK;
+
+		// If we could just check the colour, it would be great, but Echo Night decided it's going to set the alpha and green to 128, for some reason, and actually be 32bit, so it ruined my day.
+		GSTextureCache::Target* rt_tgt = g_texture_cache->GetExactTarget(m_cached_ctx.FRAME.Block(), m_cached_ctx.FRAME.FBW, GSTextureCache::RenderTarget, m_cached_ctx.FRAME.Block() + 1);
+		const bool clear_16bit_likely = !(context->FRAME.PSM & 0x2) && ((rt_tgt && (rt_tgt->m_TEX0.PSM & 2)) || (!rt_tgt && ((static_cast<int>(context->FRAME.FBW) * 64) <= (PCRTCDisplays.GetResolution().x >> 1) || m_r.height() <= (PCRTCDisplays.GetResolution().y >> 1))));
+
+		rt_tgt = nullptr;
+
+		if (clear_16bit_likely && ((const_color != 0 && (const_color >> 16) == (const_color & 0xFFFF) && ((const_color >> 8) & 0xFF) != (const_color & 0xFF)) ||
+												(fb_mask != 0 && (fb_mask >> 16) == (fb_mask & 0xFFFF) && ((fb_mask >> 8) & 0xFF) != (fb_mask & 0xFF))))
+		{
+			const int get_next_ctx = m_env.PRIM.CTXT;
+			const GSDrawingContext& next_ctx = m_env.CTXT[get_next_ctx];
+
+			GL_CACHE("Clear 16bit with 32bit %d", s_n);
+
+			// May have already been resized through the split draw checks.
+			if (!(m_cached_ctx.FRAME.PSM & 2))
+			{
+				if (next_ctx.FRAME.FBW == (m_cached_ctx.FRAME.FBW * 2))
+				{
+					m_cached_ctx.FRAME.FBW *= 2;
+					m_r.z *= 2;
+				}
+				else
+				{
+					m_r.w *= 2;
+				}
+			}
+
+			// Convert colour and masks to 16bit and set a custom TEXA for this draw.
+			const_color = ((const_color & 0x1F) << 3) | ((const_color & 0x3E0) << 6) | ((const_color & 0x7C00) << 9) | ((const_color & 0x8000) << 16);
+			m_cached_ctx.FRAME.FBMSK = ((fb_mask & 0x1F) << 3) | ((fb_mask & 0x3E0) << 6) | ((fb_mask & 0x7C00) << 9) | ((fb_mask & 0x8000) << 16);
+			m_cached_ctx.TEXA.AEM = 0;
+			m_cached_ctx.TEXA.TA0 = 0;
+			m_cached_ctx.TEXA.TA1 = 128;
+			m_cached_ctx.FRAME.PSM = (m_cached_ctx.FRAME.PSM & 2) ? m_cached_ctx.FRAME.PSM : PSMCT16;
+			m_vertex.buff[m_index.buff[1]].RGBAQ.U32[0] = const_color;
+			ReplaceVerticesWithSprite(m_r, GSVector2i(m_r.width(), m_r.height()));
+		}
 
 		// Be careful of being 1 pixel from filled.
 		const bool page_aligned = (m_r.w % pgs.y) == (pgs.y - 1) || (m_r.w % pgs.y) == 0;
@@ -2862,7 +2904,9 @@ void GSRendererHW::Draw()
 
 		GIFRegTEX0 FRAME_TEX0;
 		bool shuffle_target = false;
-		if (!no_rt && GSLocalMemory::m_psm[m_cached_ctx.FRAME.PSM].bpp == 16 && GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].bpp >= 16 &&
+		const u32 page_alignment = GSLocalMemory::IsPageAlignedMasked(m_cached_ctx.TEX0.PSM, m_r);
+		const bool page_aligned = (page_alignment & 0xF0F0) != 0; // Make sure Y is page aligned.
+		if (!no_rt && page_aligned && m_cached_ctx.ZBUF.ZMSK && GSLocalMemory::m_psm[m_cached_ctx.FRAME.PSM].bpp == 16 && GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].bpp >= 16 &&
 			(m_vt.m_primclass == GS_SPRITE_CLASS || (m_vt.m_primclass == GS_TRIANGLE_CLASS && (m_index.tail % 6) == 0 && TrianglesAreQuads(true) && m_index.tail > 6)))
 		{
 			// Tail check is to make sure we have enough strips to go all the way across the page, or if it's using a region clamp could be used to draw strips.
@@ -2934,7 +2978,7 @@ void GSRendererHW::Draw()
 		}
 
 		possible_shuffle = !no_rt && (((shuffle_target /*&& GSLocalMemory::m_psm[m_cached_ctx.FRAME.PSM].bpp == 16*/) /*|| (m_cached_ctx.FRAME.Block() == m_cached_ctx.TEX0.TBP0 && ((m_cached_ctx.TEX0.PSM & 0x6) || m_cached_ctx.FRAME.PSM != m_cached_ctx.TEX0.PSM))*/) || IsPossibleChannelShuffle());
-		const bool need_aem_color = GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].trbpp <= 24 && GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].pal == 0 && ((NeedsBlending() && m_context->ALPHA.C == 0) || IsDiscardingDstAlpha()) && m_draw_env->TEXA.AEM;
+		const bool need_aem_color = GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].trbpp <= 24 && GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].pal == 0 && ((NeedsBlending() && m_context->ALPHA.C == 0) || IsDiscardingDstAlpha()) && m_cached_ctx.TEXA.AEM;
 		const u32 color_mask = (m_vt.m_max.c > GSVector4i::zero()).mask();
 		const bool texture_function_color = m_cached_ctx.TEX0.TFX == TFX_DECAL || (color_mask & 0xFFF) || (m_cached_ctx.TEX0.TFX > TFX_DECAL && (color_mask & 0xF000));
 		const bool texture_function_alpha = m_cached_ctx.TEX0.TFX != TFX_MODULATE || (color_mask & 0xF000);
@@ -2951,8 +2995,8 @@ void GSRendererHW::Draw()
 		}
 		else
 		{
-			src = tex_psm.depth ? g_texture_cache->LookupDepthSource(true, TEX0, env.TEXA, MIP_CLAMP, tmm.coverage, possible_shuffle, m_vt.IsLinear(), m_cached_ctx.FRAME, req_color, req_alpha) :
-								  g_texture_cache->LookupSource(true, TEX0, env.TEXA, MIP_CLAMP, tmm.coverage, (GSConfig.HWMipmap || GSConfig.TriFilter == TriFiltering::Forced) ? &hash_lod_range : nullptr,
+			src = tex_psm.depth ? g_texture_cache->LookupDepthSource(true, TEX0, m_cached_ctx.TEXA, MIP_CLAMP, tmm.coverage, possible_shuffle, m_vt.IsLinear(), m_cached_ctx.FRAME, req_color, req_alpha) :
+								  g_texture_cache->LookupSource(true, TEX0, m_cached_ctx.TEXA, MIP_CLAMP, tmm.coverage, (GSConfig.HWMipmap || GSConfig.TriFilter == TriFiltering::Forced) ? &hash_lod_range : nullptr,
 					possible_shuffle, m_vt.IsLinear(), m_cached_ctx.FRAME, req_color, req_alpha);
 
 			if (!src) [[unlikely]]
@@ -3043,19 +3087,19 @@ void GSRendererHW::Draw()
 			if (scale_draw == 1)
 			{
 				target_scale = 1.0f;
-				m_downscale_source = src->m_from_target->GetScale() > 1.0f;
+				m_downscale_source = src->m_from_target ? src->m_from_target->GetScale() > 1.0f : false;
 			}
 			else
-				m_downscale_source = GSConfig.UserHacks_NativeScaling != GSNativeScaling::Aggressive ? false : src->m_from_target->GetScale() > 1.0f; // Bad for GTA + Full Spectrum Warrior, good for Sacred Blaze + Parappa.
+				m_downscale_source = (GSConfig.UserHacks_NativeScaling != GSNativeScaling::Aggressive || !src->m_from_target) ? false : src->m_from_target->GetScale() > 1.0f; // Bad for GTA + Full Spectrum Warrior, good for Sacred Blaze + Parappa.
 		}
 		else
 		{
 			// if it's directly copying keep the scale - Ratchet and clank hits this, stops edge garbage happening.
 			// Keep it to small targets of 256 or lower.
-			if (scale_draw == -1 && src && src->m_from_target && src->m_from_target->m_downscaled && ((static_cast<int>(m_cached_ctx.FRAME.FBW * 64) <= (PCRTCDisplays.GetResolution().x >> 1) &&
+			if (scale_draw == -1 && src && (!src->m_from_target || (src->m_from_target && src->m_from_target->m_downscaled)) && ((static_cast<int>(m_cached_ctx.FRAME.FBW * 64) <= (PCRTCDisplays.GetResolution().x >> 1) &&
 				(GSVector4i(m_vt.m_min.p).xyxy() == GSVector4i(m_vt.m_min.t).xyxy()).alltrue() && (GSVector4i(m_vt.m_max.p).xyxy() == GSVector4i(m_vt.m_max.t).xyxy()).alltrue()) || possible_shuffle))
 			{
-				target_scale = src->m_from_target->GetScale();
+				target_scale = src->m_from_target ? src->m_from_target->GetScale() : 1.0f;
 				scale_draw = 1;
 				scaled_copy = true;
 			}
@@ -3071,7 +3115,7 @@ void GSRendererHW::Draw()
 	// This upscaling hack is for games which construct P8 textures by drawing a bunch of small sprites in C32,
 	// then reinterpreting it as P8. We need to keep the off-screen intermediate textures at native resolution,
 	// but not propagate that through to the normal render targets. Test Case: Crash Wrath of Cortex.
-	if (no_ds && src && !m_channel_shuffle && src->m_from_target && (GSConfig.UserHacks_NativePaletteDraw || (src->m_from_target->m_downscaled && scale_draw <= 1)) &&
+	if (no_ds && src && !m_channel_shuffle && src->m_from_target && (GSConfig.UserHacks_NativePaletteDraw || (src->m_target_direct  && src->m_from_target->m_downscaled && scale_draw <= 1)) &&
 		src->m_scale == 1.0f && (src->m_TEX0.PSM == PSMT8 || src->m_TEX0.TBP0 == m_cached_ctx.FRAME.Block()))
 	{
 		GL_CACHE("HW: Using native resolution for target based on texture source");
@@ -3164,6 +3208,144 @@ void GSRendererHW::Draw()
 				}
 			}
 		}
+
+		if (no_rt && ds && ds->m_TEX0.TBP0 != m_cached_ctx.ZBUF.Block())
+		{
+			const GSLocalMemory::psm_t& zbuf_psm = GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM];
+			int vertical_offset = ((static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32) / std::max(static_cast<int>(ds->m_TEX0.TBW), 1)) * zbuf_psm.pgs.y; // I know I could just not shift it..
+			int texture_offset = 0;
+			int horizontal_offset = ((static_cast<int>((m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0)) / 32) % static_cast<int>(std::max(ds->m_TEX0.TBW, 1U))) * zbuf_psm.pgs.x;
+			// Used to reduce the offset made later in channel shuffles
+			m_target_offset = std::abs(static_cast<int>((m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0)) >> 5);
+
+			if (vertical_offset < 0)
+			{
+				ds->m_TEX0.TBP0 = m_cached_ctx.ZBUF.Block();
+				GSVector2i new_size = ds->m_unscaled_size;
+				// Make sure to use the original format for the offset.
+				const int new_offset = std::abs((vertical_offset / zbuf_psm.pgs.y) * GSLocalMemory::m_psm[ds->m_TEX0.PSM].pgs.y);
+				texture_offset = new_offset;
+
+				new_size.y += new_offset;
+
+				const GSVector4i new_drect = GSVector4i(0, new_offset * ds->m_scale, new_size.x * ds->m_scale, new_size.y * ds->m_scale);
+				ds->ResizeTexture(new_size.x, new_size.y, true, true, new_drect);
+
+				if (src && src->m_from_target && src->m_from_target == ds && src->m_target_direct)
+				{
+					src->m_texture = ds->m_texture;
+
+					// If we've moved it and the source is expecting to be inside this target, we need to update the region to point to it.
+					int max_region_y = src->m_region.GetMaxY() + new_offset;
+					if (max_region_y == new_offset)
+						max_region_y = new_size.y;
+
+					src->m_region.SetY(src->m_region.GetMinY() + new_offset, max_region_y);
+				}
+
+				ds->m_valid.y += new_offset;
+				ds->m_valid.w += new_offset;
+				ds->m_drawn_since_read.y += new_offset;
+				ds->m_drawn_since_read.w += new_offset;
+
+				g_texture_cache->CombineAlignedInsideTargets(ds, src);
+
+				if (ds->m_dirty.size())
+				{
+					for (int i = 0; i < static_cast<int>(ds->m_dirty.size()); i++)
+					{
+						ds->m_dirty[i].r.y += new_offset;
+						ds->m_dirty[i].r.w += new_offset;
+					}
+				}
+
+				t_size.y += std::abs(vertical_offset);
+				vertical_offset = 0;
+			}
+
+			if (horizontal_offset < 0)
+			{
+				// Thankfully this doesn't really happen, but catwoman moves the framebuffer backwards 1 page with a channel shuffle, which is really messy and not easy to deal with.
+				// Hopefully the quick channel shuffle will just guess this and run with it.
+				ds->m_TEX0.TBP0 += horizontal_offset;
+				horizontal_offset = 0;
+			}
+
+			if (vertical_offset || horizontal_offset)
+			{
+				GSVertex* v = &m_vertex.buff[0];
+
+				for (u32 i = 0; i < m_vertex.tail; i++)
+				{
+					v[i].XYZ.X += horizontal_offset << 4;
+					v[i].XYZ.Y += vertical_offset << 4;
+				}
+
+				if (texture_offset && src && src->m_from_target && src->m_target_direct && src->m_from_target == ds)
+				{
+					GSVector4i src_region = src->GetRegionRect();
+
+					if (src_region.rempty())
+					{
+						src_region = GSVector4i::loadh(ds->m_unscaled_size);
+						src_region.y += texture_offset;
+					}
+					else
+					{
+						src_region.y += texture_offset;
+						src_region.w += texture_offset;
+					}
+					src->m_region.SetX(src_region.x, src_region.z);
+					src->m_region.SetY(src_region.y, src_region.w);
+				}
+
+				m_context->scissor.in.x += horizontal_offset;
+				m_context->scissor.in.z += horizontal_offset;
+				m_context->scissor.in.y += vertical_offset;
+				m_context->scissor.in.w += vertical_offset;
+				m_r.y += vertical_offset;
+				m_r.w += vertical_offset;
+				m_r.x += horizontal_offset;
+				m_r.z += horizontal_offset;
+				m_in_target_draw = ds->m_TEX0.TBP0 != m_cached_ctx.ZBUF.Block();
+				m_vt.m_min.p.x += horizontal_offset;
+				m_vt.m_max.p.x += horizontal_offset;
+				m_vt.m_min.p.y += vertical_offset;
+				m_vt.m_max.p.y += vertical_offset;
+
+				t_size.y = ds->m_unscaled_size.y - vertical_offset;
+				t_size.x = ds->m_unscaled_size.x - horizontal_offset;
+			}
+
+			// Don't resize if the BPP don't match.
+			GSVector2i new_size = GetValidSize(src, possible_shuffle);
+			if (new_size.x > ds->m_unscaled_size.x || new_size.y > ds->m_unscaled_size.y)
+			{
+				const u32 new_width = std::max(new_size.x, ds->m_unscaled_size.x);
+				const u32 new_height = std::max(new_size.y, ds->m_unscaled_size.y);
+
+				//DevCon.Warning("HW: Resizing texture %d x %d draw %d", ds->m_unscaled_size.x, new_height, s_n);
+				ds->ResizeTexture(new_width, new_height);
+			}
+			else if ((IsPageCopy() || is_possible_mem_clear) && m_r.width() <= zbuf_psm.pgs.x && m_r.height() <= zbuf_psm.pgs.y)
+			{
+				const int get_next_ctx = m_env.PRIM.CTXT;
+				const GSDrawingContext& next_ctx = m_env.CTXT[get_next_ctx];
+				GSVector4i update_valid = GSVector4i::loadh(GSVector2i(horizontal_offset + GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x, GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y + vertical_offset));
+				ds->UpdateValidity(update_valid, true);
+				if (is_possible_mem_clear)
+				{
+					if ((horizontal_offset + GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x) >= static_cast<int>(ds->m_TEX0.TBW * 64) && next_ctx.ZBUF.Block() == (m_cached_ctx.ZBUF.Block() + 0x20))
+					{
+						update_valid.x = 0;
+						update_valid.z = GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x;
+						update_valid.y += GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y;
+						update_valid.w += GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y;
+						ds->UpdateValidity(update_valid, true);
+					}
+				}
+			}
+		}
 	}
 
 	if (!no_rt)
@@ -3265,7 +3447,7 @@ void GSRendererHW::Draw()
 				return;
 			}
 
-			rt = g_texture_cache->CreateTarget(FRAME_TEX0, t_size, GetValidSize(src, possible_shuffle), (GSConfig.UserHacks_NativeScaling != GSNativeScaling::Off && scale_draw < 0 && is_possible_mem_clear != ClearType::NormalClear) ? src->m_from_target->GetScale() : target_scale, 
+			rt = g_texture_cache->CreateTarget(FRAME_TEX0, t_size, GetValidSize(src, possible_shuffle), (GSConfig.UserHacks_NativeScaling != GSNativeScaling::Off && scale_draw < 0 && is_possible_mem_clear != ClearType::NormalClear) ? ((src && src->m_from_target) ? src->m_from_target->GetScale() : (ds ? ds->m_scale : 1.0f)) : target_scale, 
 												GSTextureCache::RenderTarget, true, fm, false, force_preload, preserve_rt_color || possible_shuffle, lookup_rect, src);
 
 			if (!rt) [[unlikely]]
@@ -3463,17 +3645,18 @@ void GSRendererHW::Draw()
 		if (ds && rt && (m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) != (m_cached_ctx.FRAME.Block() - rt->m_TEX0.TBP0))
 		{
 			m_using_temp_z = true;
-			const int page_offset = (static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32);
-			const int z_vertical_offset = (page_offset / std::max(ds->m_TEX0.TBW, 1U)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y;
-			const int z_horizontal_offset = (page_offset % std::max(ds->m_TEX0.TBW, 1U)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x;
+			const int page_offset = static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32;
+			const int rt_page_offset = static_cast<int>(m_cached_ctx.FRAME.Block() - rt->m_TEX0.TBP0) / 32;
+			const int z_vertical_offset = (page_offset / std::max(static_cast<int>(ds->m_TEX0.TBW), 1)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y;
+			const int z_horizontal_offset = (page_offset % std::max(static_cast<int>(ds->m_TEX0.TBW), 1)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x;
 			if (g_texture_cache->GetTemporaryZ() != nullptr)
 			{
 				GSTextureCache::TempZAddress z_address_info = g_texture_cache->GetTemporaryZInfo();
 
-				const int old_z_vertical_offset = (page_offset / std::max(ds->m_TEX0.TBW, 1U)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y;
-				const int old_z_horizontal_offset = (page_offset % std::max(ds->m_TEX0.TBW, 1U)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x;
+				const int old_z_vertical_offset = (z_address_info.offset / std::max(static_cast<int>(ds->m_TEX0.TBW), 1)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y;
+				const int old_z_horizontal_offset = (z_address_info.offset % std::max(static_cast<int>(ds->m_TEX0.TBW), 1)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x;
 
-				if (ds->m_TEX0.TBP0 != z_address_info.ZBP || z_address_info.offset != page_offset)
+				if (ds->m_TEX0.TBP0 != z_address_info.ZBP || z_address_info.offset != page_offset || z_address_info.rt_offset != rt_page_offset)
 					g_texture_cache->InvalidateTemporaryZ();
 				else if (!m_r.rintersect(z_address_info.rect_since).rempty() && m_cached_ctx.TEST.ZTST > ZTST_ALWAYS)
 				{
@@ -3493,10 +3676,11 @@ void GSRendererHW::Draw()
 
 			if (g_texture_cache->GetTemporaryZ() == nullptr)
 			{
+				ds->Update(); // We need to update any dirty bits of Z before the copy
+
 				m_temp_z_full_copy = false;
 				const int get_next_ctx = m_env.PRIM.CTXT;
 				const GSDrawingContext& next_ctx = m_env.CTXT[get_next_ctx];
-				const int rt_page_offset = (static_cast<int>(m_cached_ctx.FRAME.Block() - rt->m_TEX0.TBP0) / 32);
 				const int vertical_page_offset = (rt_page_offset / std::max(static_cast<int>(rt->m_TEX0.TBW), 1));
 				const int vertical_offset = vertical_page_offset * frame_psm.pgs.y;
 				const int horizontal_offset = (rt_page_offset - (vertical_page_offset * std::max(static_cast<int>(rt->m_TEX0.TBW), 1))) * frame_psm.pgs.x; 
@@ -3504,36 +3688,19 @@ void GSRendererHW::Draw()
 				const u32 horizontal_size = std::max(rt->m_unscaled_size.x, ds->m_unscaled_size.x);
 				const u32 vertical_size = std::max(rt->m_unscaled_size.y, ds->m_unscaled_size.y);
 
-				GSVector4i dRect = GSVector4i(horizontal_offset * ds->m_scale, vertical_offset * ds->m_scale, ds->m_unscaled_size.x * ds->m_scale, ds->m_unscaled_size.y * ds->m_scale);
+				GSVector4i dRect = GSVector4i(horizontal_offset * ds->m_scale, vertical_offset * ds->m_scale, (horizontal_offset + (ds->m_unscaled_size.x - z_horizontal_offset)) * ds->m_scale, (vertical_offset + (ds->m_unscaled_size.y - z_vertical_offset)) * ds->m_scale);
 
-				
-				const int new_height = std::max(static_cast<int>(vertical_size * ds->m_scale), dRect.w);
-				const int new_width = std::max(static_cast<int>(horizontal_size * ds->m_scale), dRect.z);
-				const int height_diff = new_height - (ds->m_unscaled_size.y * ds->m_scale);
-				const int width_diff = new_width - (ds->m_unscaled_size.x * ds->m_scale);
+				// Size here should match whichever is biggest, since that's probably what's going to happen with it further down.
+				const int new_height = std::min(2048, std::max(t_size.y, static_cast<int>(vertical_size))) * ds->m_scale;
+				const int new_width = std::min(2048, std::max(t_size.x, static_cast<int>(horizontal_size))) * ds->m_scale;
 
 				if (GSTexture* tex = g_gs_device->CreateDepthStencil(new_width, new_height, GSTexture::Format::DepthStencil, true))
 				{
-					GSVector4 sRect = GSVector4(static_cast<float>(z_horizontal_offset) / static_cast<float>(ds->m_unscaled_size.x), static_cast<float>(z_vertical_offset) / static_cast<float>(ds->m_unscaled_size.y), 1.0f - (static_cast<float>(horizontal_offset - z_horizontal_offset) / static_cast<float>(ds->m_unscaled_size.x)), 1.0f - (static_cast<float>(vertical_offset - z_vertical_offset) / static_cast<float>(ds->m_unscaled_size.y)));
+					GSVector4 sRect = GSVector4(static_cast<float>(z_horizontal_offset) / static_cast<float>(ds->m_unscaled_size.x), static_cast<float>(z_vertical_offset) / static_cast<float>(ds->m_unscaled_size.y), 1.0f , 1.0f);
 
 					const bool restricted_copy = !(((next_ctx.ZBUF.ZBP == m_context->ZBUF.ZBP && next_ctx.FRAME.FBP == m_context->FRAME.FBP)) && !(IsPossibleChannelShuffle() && !IsPageCopy()));
 
-					if (!restricted_copy)
-					{
-						if (height_diff)
-						{
-							const int adjust = std::min(height_diff, (vertical_offset - z_vertical_offset));
-							sRect.w += static_cast<float>(adjust) / static_cast<float>(ds->m_unscaled_size.y);
-							dRect.w += adjust;
-						}
-						if (width_diff)
-						{
-							const int adjust = std::min(width_diff, (horizontal_offset - z_horizontal_offset));
-							sRect.z += static_cast<float>(adjust) / static_cast<float>(ds->m_unscaled_size.x);
-							dRect.z += adjust;
-						}
-					}
-					else
+					if (restricted_copy)
 					{
 						// m_r already has horizontal_offset (rt offset) applied)
 						dRect = GSVector4i(m_r.x * ds->m_scale, m_r.y * ds->m_scale, ((1 + m_r.z) * ds->m_scale), ((1 + m_r.w) * ds->m_scale));
@@ -3546,10 +3713,13 @@ void GSRendererHW::Draw()
 
 					GL_CACHE("HW: RT in RT Z copy on draw %d z_vert_offset %d", s_n, page_offset);
 
-					g_gs_device->StretchRect(ds->m_texture, sRect, tex, GSVector4(dRect), ShaderConvert::DEPTH_COPY, false);
-					g_perfmon.Put(GSPerfMon::TextureCopies, 1);
+					if (m_cached_ctx.TEST.ZTST > ZTST_ALWAYS || !dRect.rintersect(GSVector4i(GSVector4(m_r) * ds->m_scale)).eq(dRect))
+					{
+						g_gs_device->StretchRect(ds->m_texture, sRect, tex, GSVector4(dRect), ShaderConvert::DEPTH_COPY, false);
+						g_perfmon.Put(GSPerfMon::TextureCopies, 1);
+					}
 					g_texture_cache->SetTemporaryZ(tex);
-					g_texture_cache->SetTemporaryZInfo(ds->m_TEX0.TBP0, page_offset);
+					g_texture_cache->SetTemporaryZInfo(ds->m_TEX0.TBP0, page_offset, rt_page_offset);
 					t_size.y = std::max(static_cast<int>(new_height / ds->m_scale), t_size.y);
 				}
 				else
@@ -3899,7 +4069,8 @@ void GSRendererHW::Draw()
 		bool valid_width_change = false;
 		if (rt && ((!is_possible_mem_clear || blending_cd) || rt->m_TEX0.PSM != FRAME_TEX0.PSM) && !m_in_target_draw)
 		{
-			valid_width_change = rt->m_TEX0.TBW != FRAME_TEX0.TBW;
+			const u32 frame_mask = (m_cached_ctx.FRAME.FBMSK & frame_psm.fmsk);
+			valid_width_change = rt->m_TEX0.TBW != FRAME_TEX0.TBW && (frame_mask != (frame_psm.fmsk & 0x00FFFFFF) || rt->m_valid_rgb == false);
 			if (valid_width_change && !m_cached_ctx.ZBUF.ZMSK && (m_cached_ctx.FRAME.FBMSK & 0xFF000000))
 			{
 				// Alpha could be a font, and since the width is changing it's no longer valid.
@@ -4038,6 +4209,10 @@ void GSRendererHW::Draw()
 
 		const int new_w = std::min(2048, std::max(new_size.x, std::max(rt ? rt->m_unscaled_size.x : 0, ds ? ds_size.x : 0)));
 		const int new_h = std::min(2048, std::max(new_size.y, std::max(rt ? rt->m_unscaled_size.y : 0, ds ? ds_size.y : 0)));
+
+		const bool full_cover_clear = is_possible_mem_clear && GSLocalMemory::IsPageAligned(m_cached_ctx.FRAME.PSM, m_r) && m_r.x == 0 && m_r.y == 0 && !preserve_rt_rgb &&
+									  !IsPageCopy() && m_r.width() == (m_cached_ctx.FRAME.FBW * 64);
+
 		if (rt)
 		{
 			const u32 old_end_block = rt->m_end_block;
@@ -4084,9 +4259,9 @@ void GSRendererHW::Draw()
 					g_texture_cache->AddDirtyRectTarget(rt, GSVector4i(rt->m_valid.x, rt->m_valid.w, rt->m_valid.z, new_h), rt->m_TEX0.PSM, rt->m_TEX0.TBW, mask, false);
 					g_texture_cache->GetTargetSize(rt->m_TEX0.TBP0, rt->m_TEX0.TBW, rt->m_TEX0.PSM, 0, new_h);
 				}
-
-				rt->ResizeValidity(rt->m_valid.rintersect(rt->GetUnscaledRect()));
-				rt->ResizeDrawn(rt->m_drawn_since_read.rintersect(rt->GetUnscaledRect()));
+				const bool rt_cover = full_cover_clear && (m_r.height() + frame_psm.pgs.y) >= rt->m_valid.height();
+				rt->ResizeValidity(rt_cover ? update_rect : rt->m_valid.rintersect(rt->GetUnscaledRect()));
+				rt->ResizeDrawn(rt_cover ? update_rect : rt->m_drawn_since_read.rintersect(rt->GetUnscaledRect()));
 			}
 
 			const bool rt_update = can_update_size || (m_texture_shuffle && (src && rt && src->m_from_target != rt));
@@ -4094,7 +4269,7 @@ void GSRendererHW::Draw()
 			// If it's updating from a texture shuffle, limit the size to the source size.
 			if (rt_update && !can_update_size)
 			{
-				if(src->m_from_target)
+				if (src->m_from_target)
 					update_rect = update_rect.rintersect(src->m_from_target->m_valid);
 
 				update_rect = update_rect.rintersect(GSVector4i::loadh(GSVector2i(new_w, new_h)));
@@ -4164,16 +4339,18 @@ void GSRendererHW::Draw()
 						DevCon.Warning("HW: Temporary depth buffer creation failed.");
 				}
 			}
+			const bool z_masked = m_cached_ctx.ZBUF.ZMSK;
+
 			if (!m_texture_shuffle && !m_channel_shuffle)
 			{
-				ds->ResizeValidity(ds->GetUnscaledRect());
-				ds->ResizeDrawn(ds->GetUnscaledRect());
+				const bool z_cover = full_cover_clear && (m_r.height() + GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y) >= ds->m_valid.height();
+				ds->ResizeValidity(z_cover ? m_r : ds->GetUnscaledRect());
+				ds->ResizeDrawn(z_cover ? m_r : ds->GetUnscaledRect());
 			}
 
 			// Limit to 2x the vertical height of the resolution (for double buffering)
 			// Dark cloud writes to 424 when the buffer is only 416 high, but masks the Z.
 			// Updating the valid causes the Z to overlap the framebuffer, which is obviously incorrect.
-			const bool z_masked = m_cached_ctx.ZBUF.ZMSK;
 			const bool z_update = can_update_size && !z_masked;
 
 			if (rt && m_using_temp_z)
@@ -4181,7 +4358,10 @@ void GSRendererHW::Draw()
 				const GSLocalMemory::psm_t& z_psm = GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM];
 				const int vertical_offset = ((static_cast<int>(m_cached_ctx.FRAME.Block() - rt->m_TEX0.TBP0) / 32) / std::max(static_cast<int>(rt->m_TEX0.TBW), 1)) * frame_psm.pgs.y;
 				const int z_vertical_offset = ((static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32) / std::max(static_cast<int>(ds->m_TEX0.TBW), 1)) * z_psm.pgs.y;
-				const GSVector4i ds_rect = m_r - GSVector4i(vertical_offset - z_vertical_offset);
+				const int z_horizontal_offset = ((static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32) % std::max(rt->m_TEX0.TBW, 1U)) * z_psm.pgs.x;
+				const int horizontal_offset = ((static_cast<int>(m_cached_ctx.FRAME.Block() - rt->m_TEX0.TBP0) / 32) % std::max(static_cast<int>(rt->m_TEX0.TBW), 1)) * frame_psm.pgs.x;
+				
+				const GSVector4i ds_rect = m_r - GSVector4i(horizontal_offset - z_horizontal_offset, vertical_offset - z_vertical_offset).xyxy();
 				ds->UpdateValidity(ds_rect, z_update && (can_update_size || (ds_rect.w <= (resolution.y * 2) && !m_texture_shuffle)));
 				ds->UpdateDrawn(ds_rect, z_update && (can_update_size || (ds_rect.w <= (resolution.y * 2) && !m_texture_shuffle)));
 			}
@@ -4426,14 +4606,14 @@ void GSRendererHW::Draw()
 		{
 			const int get_next_ctx = m_env.PRIM.CTXT;
 			const GSDrawingContext& next_ctx = m_env.CTXT[get_next_ctx];
-			const int z_vertical_offset = ((static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32) / std::max(rt->m_TEX0.TBW, 1U)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y;
-			const int z_horizontal_offset = ((static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32) % std::max(rt->m_TEX0.TBW, 1U)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x;
+			const int z_vertical_offset = ((static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32) / std::max(static_cast<int>(rt->m_TEX0.TBW), 1)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.y;
+			const int z_horizontal_offset = ((static_cast<int>(m_cached_ctx.ZBUF.Block() - ds->m_TEX0.TBP0) / 32) % std::max(static_cast<int>(rt->m_TEX0.TBW), 1)) * GSLocalMemory::m_psm[m_cached_ctx.ZBUF.PSM].pgs.x;
 			const int vertical_offset = ((static_cast<int>(m_cached_ctx.FRAME.Block() - rt->m_TEX0.TBP0) / 32) / std::max(static_cast<int>(rt->m_TEX0.TBW), 1)) * frame_psm.pgs.y;
 			const int horizontal_offset = ((static_cast<int>(m_cached_ctx.FRAME.Block() - rt->m_TEX0.TBP0) / 32) % std::max(static_cast<int>(rt->m_TEX0.TBW), 1)) * frame_psm.pgs.x;
 
 			if (was_written)
 			{
-				const GSVector4i ds_real_rect = real_rect - GSVector4i(vertical_offset - z_vertical_offset);
+				const GSVector4i ds_real_rect = real_rect - GSVector4i(horizontal_offset - z_horizontal_offset, vertical_offset - z_vertical_offset).xyxy();
 				ds->UpdateValidity(ds_real_rect, !z_masked && (can_update_size || (ds_real_rect.w <= (resolution.y * 2) && !m_texture_shuffle)));
 			}
 
@@ -6197,7 +6377,7 @@ __ri void GSRendererHW::EmulateTextureSampler(const GSTextureCache::Target* rt,
 	// 2/ Only keep aem when it is useful (avoid useless shader permutation)
 	if (m_conf.ps.shuffle)
 	{
-		const GIFRegTEXA& TEXA = m_draw_env->TEXA;
+		const GIFRegTEXA& TEXA = m_cached_ctx.TEXA;
 
 		// Force a 32 bits access (normally shuffle is done on 16 bits)
 		// m_ps_sel.tex_fmt = 0; // removed as an optimization
@@ -6251,7 +6431,7 @@ __ri void GSRendererHW::EmulateTextureSampler(const GSTextureCache::Target* rt,
 	}
 	else if (tex->m_target)
 	{
-		const GIFRegTEXA& TEXA = m_draw_env->TEXA;
+		const GIFRegTEXA& TEXA = m_cached_ctx.TEXA;
 
 		// Use an old target. AEM and index aren't resolved it must be done
 		// on the GPU
@@ -7088,13 +7268,13 @@ __ri void GSRendererHW::DrawPrims(GSTextureCache::Target* rt, GSTextureCache::Ta
 			const GSVector4i shuffle_rect = GSVector4i(m_vt.m_min.p.x, m_vt.m_min.p.y, m_vt.m_max.p.x, m_vt.m_max.p.y);
 			if (!rt->m_valid.rintersect(shuffle_rect).eq(rt->m_valid) || (m_cached_ctx.FRAME.FBMSK & 0xFFFC0000))
 			{
-				rt_new_alpha_max = std::max(static_cast<int>((std::max(m_draw_env->TEXA.TA1, m_draw_env->TEXA.TA0) & 0x80) + 127), rt_new_alpha_max) | fba_value;
-				rt_new_alpha_min = std::min(static_cast<int>(std::min(m_draw_env->TEXA.TA1, m_draw_env->TEXA.TA0) & 0x80), rt_new_alpha_min);
+				rt_new_alpha_max = std::max(static_cast<int>((std::max(m_cached_ctx.TEXA.TA1, m_cached_ctx.TEXA.TA0) & 0x80) + 127), rt_new_alpha_max) | fba_value;
+				rt_new_alpha_min = std::min(static_cast<int>(std::min(m_cached_ctx.TEXA.TA1, m_cached_ctx.TEXA.TA0) & 0x80), rt_new_alpha_min);
 			}
 			else
 			{
-				rt_new_alpha_max = (std::max(m_draw_env->TEXA.TA1, m_draw_env->TEXA.TA0) & 0x80) + 127 | fba_value;
-				rt_new_alpha_min = (std::min(m_draw_env->TEXA.TA1, m_draw_env->TEXA.TA0) & 0x80) | fba_value;
+				rt_new_alpha_max = (std::max(m_cached_ctx.TEXA.TA1, m_cached_ctx.TEXA.TA0) & 0x80) + 127 | fba_value;
+				rt_new_alpha_min = (std::min(m_cached_ctx.TEXA.TA1, m_cached_ctx.TEXA.TA0) & 0x80) | fba_value;
 			}
 			rt->m_alpha_range = true;
 		}
@@ -8009,7 +8189,7 @@ bool GSRendererHW::CanUseSwPrimRender(bool no_rt, bool no_ds, bool draw_sprite_t
 					return false;
 			}
 
-			const bool need_aem_color = GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].trbpp <= 24 && GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].pal == 0 && ((NeedsBlending() && m_context->ALPHA.C == 0) || IsDiscardingDstAlpha()) && m_draw_env->TEXA.AEM;
+			const bool need_aem_color = GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].trbpp <= 24 && GSLocalMemory::m_psm[m_cached_ctx.TEX0.PSM].pal == 0 && ((NeedsBlending() && m_context->ALPHA.C == 0) || IsDiscardingDstAlpha()) && m_cached_ctx.TEXA.AEM;
 			const u32 color_mask = (m_vt.m_max.c > GSVector4i::zero()).mask();
 			const bool texture_function_color = m_cached_ctx.TEX0.TFX == TFX_DECAL || (color_mask & 0xFFF) || (m_cached_ctx.TEX0.TFX > TFX_DECAL && (color_mask & 0xF000));
 			const bool texture_function_alpha = m_cached_ctx.TEX0.TFX != TFX_MODULATE || (color_mask & 0xF000);
@@ -8038,7 +8218,7 @@ bool GSRendererHW::CanUseSwPrimRender(bool no_rt, bool no_ds, bool draw_sprite_t
 
 					if (start_bp < dirty_end_bp && end_bp > dirty_start_bp)
 					{
-						if (dirty_start_bp > start_bp || dirty_end_bp < end_bp)
+						if (dirty_start_bp <= start_bp && dirty_end_bp >= end_bp)
 						{
 							return true;
 						}
@@ -8189,8 +8369,8 @@ bool GSRendererHW::DetectDoubleHalfClear(bool& no_rt, bool& no_ds)
 			return false;
 	}
 
-	// Z and color must be constant and the same
-	if (m_vt.m_eq.rgba != 0xFFFF || !m_vt.m_eq.z)
+	// Z and color must be constant and the same and both are enabled.
+	if (m_vt.m_eq.rgba != 0xFFFF || !m_vt.m_eq.z || (no_ds != no_rt))
 		return false;
 
 	const u32 write_color = GetConstantDirectWriteMemClearColor();
@@ -8269,7 +8449,7 @@ bool GSRendererHW::DetectDoubleHalfClear(bool& no_rt, bool& no_ds)
 	// bang up next to each other, or a double half clear. The two are really difficult to differentiate.
 	// Have to check both contexts, because God of War 2 likes to do this in-between setting TRXDIR, which
 	// causes a flush, and we don't have the next context backed up index set.
-	bool horizontal = false;
+	bool horizontal = std::abs(static_cast<int>(m_cached_ctx.FRAME.FBP) - static_cast<int>(m_cached_ctx.ZBUF.ZBP)) == (m_cached_ctx.FRAME.FBW >> 1);
 	const bool possible_next_clear = !m_env.PRIM.TME && !(m_env.SCANMSK.MSK & 2) && !m_env.CTXT[next_ctx].TEST.ATE && !m_env.CTXT[next_ctx].TEST.DATE &&
 		(!m_env.CTXT[next_ctx].TEST.ZTE || m_env.CTXT[next_ctx].TEST.ZTST == ZTST_ALWAYS);
 
@@ -8352,6 +8532,14 @@ bool GSRendererHW::DetectDoubleHalfClear(bool& no_rt, bool& no_ds)
 	else
 	{
 		const int height = m_r.height();
+
+		// We don't want to double half clear already full sized targets, making them double the size, this could be very bad.
+		// This gets triggered by Monster Lab which clears the Z and FRAME in one go, butted up against each other.
+		// It's highly unlikely that it will actually require a > 600 high framebuffer, but check with the display height first.
+		const int display_height = PCRTCDisplays.GetResolution().y;
+		if ((display_height != 0 && height >= (display_height - 1)) || height > 300)
+			return false;
+
 		m_r.w = ((half - base) / m_cached_ctx.FRAME.FBW) * frame_psm.pgs.y;
 		m_r.w += m_r.y + height;
 	}
@@ -8834,6 +9022,9 @@ bool GSRendererHW::TextureCoversWithoutGapsNotEqual()
 
 int GSRendererHW::IsScalingDraw(GSTextureCache::Source* src, bool no_gaps)
 {
+	if (GSConfig.UserHacks_NativeScaling == GSNativeScaling::Off)
+		return 0;
+
 	const GSVector2i draw_size = GSVector2i(m_vt.m_max.p.x - m_vt.m_min.p.x, m_vt.m_max.p.y - m_vt.m_min.p.y);
 	GSVector2i tex_size = GSVector2i(m_vt.m_max.t.x - m_vt.m_min.t.x, m_vt.m_max.t.y - m_vt.m_min.t.y);
 
@@ -8844,10 +9035,13 @@ int GSRendererHW::IsScalingDraw(GSTextureCache::Source* src, bool no_gaps)
 
 	// Try to catch cases of stupid draws like Manhunt and Syphon Filter where they sample a single pixel.
 	// Also make sure it's grabbing most of the texture.
-	if (tex_size.x == 0 || tex_size.y == 0 || draw_size.x == 0 || draw_size.y == 0 || !is_target_src)
+	if (tex_size.x == 0 || tex_size.y == 0 || draw_size.x == 0 || draw_size.y == 0)
 		return 0;
 
-	if (is_target_src && src->m_from_target->m_downscaled && std::abs(draw_size.x - tex_size.x) <= 1 && std::abs(draw_size.y - tex_size.y) <= 1)
+	const bool no_resize = (std::abs(draw_size.x - tex_size.x) <= 1 && std::abs(draw_size.y - tex_size.y) <= 1);
+	const bool can_maintain = no_resize || (!is_target_src && m_index.tail == 2);
+
+	if (!src || ((!is_target_src || src->m_from_target->m_downscaled) && can_maintain))
 		return -1;
 
 	const GSDrawingContext& next_ctx = m_env.CTXT[m_env.PRIM.CTXT];
@@ -8860,8 +9054,9 @@ int GSRendererHW::IsScalingDraw(GSTextureCache::Source* src, bool no_gaps)
 	const bool is_downscale = m_cached_ctx.TEX0.TBW >= m_cached_ctx.FRAME.FBW && draw_size.x <= (tex_size.x * 0.75f) && draw_size.y <= (tex_size.y * 0.75f);
 	// Check we're getting most of the texture and not just stenciling a part of it.
 	// Only allow non-bilineared downscales if it's most of the target (misdetections of shadows in Naruto, Transformers etc), otherwise it's fine.
-	const GSVector2i tex_size_half = GSVector2i((src->GetRegion().HasX() ? src->GetRegionSize().x : src->m_from_target->m_valid.width()) / 2, (src->GetRegion().HasY() ? src->GetRegionSize().y : src->m_from_target->m_valid.height()) / 2);
-	const bool possible_downscale = m_context->TEX1.MMAG == 1 || src->m_from_target->m_downscaled || tex_size.x >= tex_size_half.x || tex_size.y >= tex_size_half.y;
+	const GSVector4i src_valid = src->m_from_target ? src->m_from_target->m_valid : src->m_valid_rect;
+	const GSVector2i tex_size_half = GSVector2i((src->GetRegion().HasX() ? src->GetRegionSize().x : src_valid.width()) / 2, (src->GetRegion().HasY() ? src->GetRegionSize().y : src_valid.height()) / 2);
+	const bool possible_downscale = m_context->TEX1.MMIN == 1 || !src->m_from_target || src->m_from_target->m_downscaled || tex_size.x >= tex_size_half.x || tex_size.y >= tex_size_half.y;
 
 	if (is_downscale && (draw_size.x >= PCRTCDisplays.GetResolution().x || !possible_downscale))
 		return 0;
@@ -8879,6 +9074,39 @@ int GSRendererHW::IsScalingDraw(GSTextureCache::Source* src, bool no_gaps)
 		return is_upscale ? 2 : 1;
 	}
 
+	// Last ditched check if it's doing a lot of small draws exactly the same which could be recursive lighting bloom.
+	if (m_vt.m_primclass == GS_SPRITE_CLASS && m_index.tail > 2 && !no_gaps_or_single_sprite && m_context->TEX1.MMAG == 1 && !m_context->ALPHA.IsOpaque())
+	{
+		GSVertex* v = &m_vertex.buff[0];
+		float tw = 1 << src->m_TEX0.TW;
+		float th = 1 << src->m_TEX0.TH;
+
+		const int first_u = (PRIM->FST) ? (v[1].U - v[0].U) >> 4 : std::floor(static_cast<int>(tw * v[1].ST.S) - static_cast<int>(tw * v[0].ST.S));
+		const int first_v = (PRIM->FST) ? (v[1].V - v[0].V) >> 4 : std::floor(static_cast<int>(th * v[1].ST.T) - static_cast<int>(th * v[0].ST.T));
+		const int first_x = (v[1].XYZ.X - v[0].XYZ.X) >> 4;
+		const int first_y = (v[1].XYZ.Y - v[0].XYZ.Y) >> 4;
+
+		if (first_x > first_u && first_y > first_v && !no_resize && std::abs(draw_size.x - first_x) <= 4 && std::abs(draw_size.y - first_y) <= 4)
+		{
+			for (u32 i = 2; i < m_index.tail; i += 2)
+			{
+				const int next_u = (PRIM->FST) ? (v[i + 1].U - v[i].U) >> 4 : std::floor(static_cast<int>(tw * v[i + 1].ST.S) - static_cast<int>(tw * v[i].ST.S));
+				const int next_v = (PRIM->FST) ? (v[i + 1].V - v[i].V) >> 4 : std::floor(static_cast<int>(th * v[i + 1].ST.T) - static_cast<int>(th * v[i].ST.T));
+				const int next_x = (v[i + 1].XYZ.X - v[i].XYZ.X) >> 4;
+				const int next_y = (v[i + 1].XYZ.Y - v[i].XYZ.Y) >> 4;
+
+				if (std::abs(draw_size.x - next_x) > 4 || std::abs(draw_size.y - next_y) > 4)
+					break;
+
+				if (next_u != first_u || next_v != first_v || next_x != first_x || next_y != first_y)
+					break;
+
+				if (i + 2 >= m_index.tail)
+					return 2;
+			}
+		}
+	}
+
 	return 0;
 }
 

pcsx2/GS/Renderers/HW/GSRendererHW.h

@@ -140,6 +140,7 @@ private:
 	struct HWCachedCtx
 	{
 		GIFRegTEX0 TEX0;
+		GIFRegTEXA TEXA;
 		GIFRegCLAMP CLAMP;
 		GIFRegTEST TEST;
 		GIFRegFRAME FRAME;

pcsx2/GS/Renderers/HW/GSTextureCache.cpp

@@ -228,19 +228,22 @@ bool GSTextureCache::CanTranslate(u32 bp, u32 bw, u32 spsm, GSVector4i r, u32 db
 {
 	const GSVector2i src_page_size = GSLocalMemory::m_psm[spsm].pgs;
 	const GSVector2i dst_page_size = GSLocalMemory::m_psm[dpsm].pgs;
-	const bool bp_page_aligned_bp = ((bp & ~((1 << 5) - 1)) == bp) || bp == dbp;
+	const u32 src_bw = std::max(1U, bw);
+	const u32 dst_bw = std::max(1U, dbw);
 	const bool block_layout_match = GSLocalMemory::m_psm[spsm].bpp == GSLocalMemory::m_psm[dpsm].bpp;
+	const bool bp_page_aligned_bp = ((bp & ~((1 << 5) - 1)) == bp) || bp == dbp || (block_layout_match && src_bw == dst_bw);
 	const GSVector4i page_mask(GSVector4i((src_page_size.x - 1), (src_page_size.y - 1)).xyxy());
 	const GSVector4i masked_rect(r & ~page_mask);
-	const int src_pixel_width = static_cast<int>(bw * 64);
+	const int src_pixel_width = src_bw * 64;
+	const int dst_pixel_width = dst_bw * 64;
 	// We can do this if:
 	// The page width matches.
 	// The rect width is less than the width of the destination texture and the height is less than or equal to 1 page high.
 	// The rect width and height is equal to the page size and it covers the width of the incoming bw, so lines are sequential.
 	const bool page_aligned_rect = masked_rect.xyxy().eq(r.xyxy());
-	const bool width_match = ((bw * 64) / src_page_size.x) == ((dbw * 64) / dst_page_size.x);
+	const bool width_match = (src_pixel_width / src_page_size.x) == (dst_pixel_width / dst_page_size.x);
 	const bool sequential_pages = page_aligned_rect && r.x == 0 && r.z == src_pixel_width;
-	const bool single_row = (((bw * 64) / src_page_size.x) <= ((dbw * 64) / dst_page_size.x)) && r.width() <= src_pixel_width && r.height() <= src_page_size.y;
+	const bool single_row = ((src_pixel_width / src_page_size.x) <= (dst_pixel_width / dst_page_size.x)) && r.width() <= src_pixel_width && r.height() <= src_page_size.y;
 	const bool single_page_aligned = page_aligned_rect && r.z <= src_page_size.x && r.w <= src_page_size.y;
 	if (block_layout_match)
 	{
@@ -276,7 +279,7 @@ GSVector4i GSTextureCache::TranslateAlignedRectByPage(u32 tbp, u32 tebp, u32 tbw
 	int page_offset = (static_cast<int>(sbp) - static_cast<int>(tbp)) >> 5;
 	int block_offset = (static_cast<int>(sbp) - static_cast<int>(tbp)) & 0x1F;
 
-	if (!(s_psm.bpp == t_psm.bpp))
+	if (!(s_psm.bpp == t_psm.bpp) || block_offset)
 	{
 		if (block_offset)
 			in_rect = in_rect.ralign<Align_Outside>(s_psm.bs);
@@ -299,6 +302,33 @@ GSVector4i GSTextureCache::TranslateAlignedRectByPage(u32 tbp, u32 tebp, u32 tbw
 			DevCon.Warning("Error translating rect");
 			return GSVector4i::zero();
 		}
+
+		const bool block_matched_format = s_psm.bpp == t_psm.bpp && (clamped_sbw == clamped_tbw || clamped_sbw == 1);
+		// If there is block offset left over, try to adjust to that.
+		if (block_matched_format)
+		{
+			GSVector4i b2a_offset = GSVector4i::zero();
+			const GSVector4i target_rect = GSVector4i(0, 0, src_bw, 2048);
+			bool offset_found = false;
+
+			for (b2a_offset.x = target_rect.x; b2a_offset.x < target_rect.z; b2a_offset.x += s_psm.bs.x)
+			{
+				for (b2a_offset.y = target_rect.y; b2a_offset.y < target_rect.w; b2a_offset.y += s_psm.bs.y)
+				{
+					const u32 a_candidate_bp = s_psm.info.bn(b2a_offset.x, b2a_offset.y, tbp, sbw);
+					if (sbp == a_candidate_bp)
+					{
+						offset_found = true;
+						break;
+					}
+				}
+				if (offset_found)
+					break;
+			}
+
+			if (offset_found)
+				in_rect = (in_rect + b2a_offset.xyxy()).max_i32(GSVector4i(0));
+		}
 	}
 
 	GSVector4i new_rect = GSVector4i::zero();
@@ -397,13 +427,15 @@ GSVector4i GSTextureCache::TranslateAlignedRectByPage(u32 tbp, u32 tebp, u32 tbw
 			}
 		}
 	}
-	else // Widths match
+	else if (!block_offset) // Widths match
 	{
 		const int horizontal_dst_page_offset = page_offset % clamped_tbw;
 		const int vertical_dst_page_offset = page_offset / clamped_tbw;
 		GSVector4i offset_rect(horizontal_dst_page_offset * t_psm.pgs.x, vertical_dst_page_offset * t_psm.pgs.y);
 		new_rect = in_rect + offset_rect.xyxy();
 	}
+	else
+		new_rect = in_rect;
 
 	if (new_rect.z > dst_bw)
 	{
@@ -602,6 +634,8 @@ void GSTextureCache::DirtyRectByPage(u32 sbp, u32 spsm, u32 sbw, Target* t, GSVe
 	{
 		RGBAMask rgba;
 		rgba._u32 = GSUtil::GetChannelMask(spsm);
+		// FIXME: This could be a problem if used when the valid area is smaller than dirty area and it needs the data during expansion on a later draw.
+		// This happens on Kamen Rider - Seigi no Keifu if the invalidatevideomem function was to use this function for depth clearing.
 		AddDirtyRectTarget(t, t->m_valid, t->m_TEX0.PSM, t->m_TEX0.TBW, rgba);
 		return;
 	}
@@ -618,7 +652,7 @@ void GSTextureCache::DirtyRectByPage(u32 sbp, u32 spsm, u32 sbw, Target* t, GSVe
 	int block_offset = static_cast<int>(sbp) - static_cast<int>(target_bp);
 	// Different format needs to be page aligned, unless the block layout matches, then we can block align
 	// Might be able to translate the original rect.
-	if (!(src_info->bpp == dst_info->bpp))
+	if (!(src_info->bpp == dst_info->bpp) || block_offset)
 	{
 		const int src_bpp = src_info->bpp;
 		const bool column_align = !block_offset && src_r.z <= src_info->cs.x && src_r.w <= src_info->cs.y && src_info->depth == dst_info->depth;
@@ -741,7 +775,7 @@ void GSTextureCache::DirtyRectByPage(u32 sbp, u32 spsm, u32 sbw, Target* t, GSVe
 			const int yblocks = in_rect.height() / src_info->bs.y;
 			// if !(block_offset & 0x7) is false, this is technically incorrect, but FFX hates it and starts making a mess, so it's better this way without adding complexity.
 			// TODO maybe: Add per block invalidation? ugh, would have to keep that to small blocks. 2 blocks in the case of FFX.
-			if ((xblocks <= 4 && yblocks <= 2) || req_depth_offset)
+			if ((xblocks <= (src_info->pgs.x / src_info->bs.x) && yblocks <= (src_info->pgs.y / src_info->bs.y)) || req_depth_offset)
 			{
 				GSVector4i b2a_offset = GSVector4i::zero();
 				const GSVector4i target_rect = GSVector4i(0, 0, src_width, 2048);
@@ -1165,8 +1199,8 @@ GSTextureCache::Source* GSTextureCache::LookupDepthSource(const bool is_depth, c
 		if (inside_target)
 		{
 			// Need to set it up as a region target.
-			src->m_region.SetX(block_boundary_rect.x, src->m_from_target->m_valid.z);
-			src->m_region.SetY(block_boundary_rect.y, src->m_from_target->m_valid.w);
+			src->m_region.SetX(block_boundary_rect.x, std::max(src->m_from_target->m_valid.z, block_boundary_rect.z));
+			src->m_region.SetY(block_boundary_rect.y, std::max(src->m_from_target->m_valid.w, block_boundary_rect.w));
 		}
 
 		if (GSRendererHW::GetInstance()->IsTBPFrameOrZ(dst->m_TEX0.TBP0))
@@ -1279,11 +1313,6 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 			// Make sure it is page aligned, otherwise things get messy with the pixel order (Tomb Raider Legend).
 			if (t->m_used)
 			{
-				// If the BP is block offset it's unlikely to be a target, but a target can be made by a HW move, so we need to check for a match.
-				// Good for Baldurs Gate Dark Alliance (HW Move), bad for Tomb Raider Legends (just offset).
-				if (((bp & (BLOCKS_PER_PAGE - 1)) != (t->m_TEX0.TBP0 & (BLOCKS_PER_PAGE - 1))) && (bp & (BLOCKS_PER_PAGE - 1)))
-					continue;
-
 				//const bool overlaps = t->Inside(bp, bw, psm, block_boundary_rect);
 				const bool overlaps = t->Overlaps(bp, bw, psm, block_boundary_rect);
 				// Try to make sure the target has available what we need, be careful of self referencing frames with font in the alpha.
@@ -1293,6 +1322,12 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 				if (!overlaps || (found_t && (GSState::s_n - dst->m_last_draw) < (GSState::s_n - t->m_last_draw)))
 					continue;
 
+				// If the BP is offset in to a page and the format does not match, trying to match up the correct position is very difficult since we don't swizzle.
+				// Tomb Raider Legends does a block level BP in PSMT8 over a C16 target, which is just a nightmare to get right.
+				// Baldurs Gate used to have this too, but now we can translate HW moves inside targets when the format matches.
+				if (((bp & (BLOCKS_PER_PAGE - 1)) != (t->m_TEX0.TBP0 & (BLOCKS_PER_PAGE - 1))) && (bp & (BLOCKS_PER_PAGE - 1)))
+					continue;
+
 				const bool width_match = (std::max(64U, bw * 64U) >> GSLocalMemory::m_psm[psm].info.pageShiftX()) ==
 										 (std::max(64U, t->m_TEX0.TBW * 64U) >> GSLocalMemory::m_psm[t->m_TEX0.PSM].info.pageShiftX());
 
@@ -1596,14 +1631,12 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 					{
 						// It is a complex to convert the code in shader. As a reference, let's do it on the CPU,
 						// it will be slow but can work even with upscaling, also fine tune it so it's not enabled when not needed.
-						if (psm == PSMT4 || (GSConfig.UserHacks_CPUFBConversion && psm == PSMT8 && (!possible_shuffle || GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp != 32)) ||
-							(psm == PSMT8H && GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp == 16))
+						if (psm == PSMT4 || (psm == PSMT8H && GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp == 16))
 						{
-							// Forces 4-bit and 8-bit frame buffer conversion to be done on the CPU instead of the GPU, but performance will be slower.
-							// There is no dedicated shader to handle 4-bit conversion (Beyond Good and Evil and Stuntman).
+							// Enable readbacks on PSMT4 as we don't have a dedicated shader (Beyond Good and Evil and Stuntman).
+							// Enable readbacks on PSMT8H 16bit as we don't have a dedicated shader (History Channel - Battle for the Pacific, Sea World - Shamu's Big Adventure). 
 							// Note: Stuntman no longer hits the PSMT4 code path.
-							// Direct3D10/11 and OpenGL support 8-bit fb conversion but don't render some corner cases properly (Harry Potter games).
-							// The hack can fix glitches in some games.
+							// Note2: Harry Potter is now properly handled with shader conversion and no need to enable frame buffer conversion.
 							if (!t->m_drawn_since_read.rempty())
 							{
 								t->UnscaleRTAlpha();
@@ -1617,8 +1650,7 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 						{
 							const bool outside_target = !t->Overlaps(bp, bw, psm, r);
 
-							// We don't have a shader for this.
-							if (!possible_shuffle && TEX0.PSM == PSMT8 && ((GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp != 32) || outside_target))
+							if (!possible_shuffle && TEX0.PSM == PSMT8 && outside_target)
 							{
 								continue;
 							}
@@ -1647,7 +1679,7 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 				// Make sure the texture actually is INSIDE the RT, it's possibly not valid if it isn't.
 				// Also check BP >= TBP, create source isn't equpped to expand it backwards and all data comes from the target. (GH3)
 				else if (GSConfig.UserHacks_TextureInsideRt >= GSTextureInRtMode::InsideTargets &&
-						 (GSLocalMemory::m_psm[color_psm].bpp >= 16 || (/*possible_shuffle &&*/ GSLocalMemory::m_psm[color_psm].bpp == 8 && GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp == 32)) && // Channel shuffles or non indexed lookups.
+						 (GSLocalMemory::m_psm[color_psm].bpp >= 16 || (/*possible_shuffle &&*/ GSLocalMemory::m_psm[color_psm].bpp == 8 && GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp >= 16)) && // Channel shuffles or non indexed lookups.
 					t->m_age <= 1 && (!found_t || t->m_last_draw > dst->m_last_draw) /*&& CanTranslate(bp, bw, psm, block_boundary_rect, t->m_TEX0.TBP0, t->m_TEX0.PSM, t->m_TEX0.TBW)*/)
 				{
 					u32 rt_tbw = std::max(1U, t->m_TEX0.TBW);
@@ -1666,7 +1698,7 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 					// Keep note that 2 bw is basically 1 normal page, as bw is in 64 pixels, and 8bit pages are 128 pixels wide, aka 2 bw.
 					// Also check for 4HH/HL and 8H which use the alpha channel, if the page order is wrong this can cause problems as well (Jak X font).
 					else if (!possible_shuffle && GSLocalMemory::m_psm[psm].trbpp <= 8 && (GSUtil::GetChannelMask(t->m_TEX0.PSM) != 0xF || 
-							(GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp == 32 && (!(block_boundary_rect.w <= GSLocalMemory::m_psm[psm].pgs.y &&
+							((GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp != 16 || GSLocalMemory::m_psm[psm].bpp < 16) && (!(block_boundary_rect.w <= GSLocalMemory::m_psm[psm].pgs.y &&
 							((GSLocalMemory::m_psm[psm].bpp == 32) ? bw : ((bw + 1) / 2)) <= t->m_TEX0.TBW) &&
 							!(((GSLocalMemory::m_psm[psm].bpp == 32) ? bw : ((bw + 1) / 2)) == rt_tbw)))))
 					{
@@ -1760,7 +1792,7 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 
 							//rect = rect.rintersect(t->m_valid);
 
-							if (rect.rintersect(t->m_valid).rempty())
+							if (rect.rintersect(t->m_valid - GSVector4i(0, 1).xyxy()).rempty())
 								continue;
 
 							if (!t->m_dirty.empty())
@@ -1773,10 +1805,6 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 							if (!t->HasValidBitsForFormat(psm, req_color, req_alpha, rt_tbw == TEX0.TBW) && !(possible_shuffle && GSLocalMemory::m_psm[psm].bpp == 16 && GSLocalMemory::m_psm[t->m_TEX0.PSM].bpp == 32))
 								continue;
 
-							// Be careful of shuffles where it can shuffle the width of the target, even though it may not have all been drawn to.
-							if (!possible_shuffle && frame.Block() != TEX0.TBP0 && !t->Inside(bp, bw, psm, block_boundary_rect))
-								continue;
-
 							x_offset = rect.x;
 							y_offset = rect.y;
 							dst = t;
@@ -2007,7 +2035,6 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 			GL_CACHE("TC: src miss (0x%x, 0x%x, %s)", TEX0.TBP0, psm_s.pal > 0 ? TEX0.CBP : 0, psm_str(TEX0.PSM));
 		}
 #endif
-
 		// This is for the condition where the target doesn't exist on a shuffle and it needs to load from memory.
 		// The Godfather clears the depth buffer with a normal clear, so our depth target gets deleted, then because it finds no target
 		// it assumes it really is 16bits, causing the texture to be full of garbage, and our shuffle handling becomes a mess.
@@ -2019,11 +2046,29 @@ GSTextureCache::Source* GSTextureCache::LookupSource(const bool is_color, const
 			{
 				rect.y /= 2;
 				rect.w /= 2;
+
+				if (region.HasY())
+				{
+					const u32 min_y = region.GetMinY() / 2;
+					const u32 max_y = region.GetMaxY() / 2;
+
+					region.ClearY();
+					region.SetY(min_y, max_y);
+				}
 			}
 			else
 			{
 				rect.x /= 2;
 				rect.z /= 2;
+
+				if (region.HasX())
+				{
+					const u32 min_x = region.GetMinX() / 2;
+					const u32 max_x = region.GetMaxX() / 2;
+
+					region.ClearX();
+					region.SetX(min_x, max_x);
+				}
 			}
 			if (TEX0.TBP0 == frame.Block())
 			{
@@ -2337,6 +2382,15 @@ GSTextureCache::Target* GSTextureCache::LookupTarget(GIFRegTEX0 TEX0, const GSVe
 			// Probably pointing to half way through the target
 			else if (!min_rect.rempty() && GSConfig.UserHacks_TextureInsideRt >= GSTextureInRtMode::InsideTargets)
 			{
+				// Some games misuse the scissor so it ends up valid 1 pixel over, which causes hell for us. So check if it still overlaps without the extra pixel.
+				const GSVector4i adjusted_valid = GSVector4i(t->m_valid.x, t->m_valid.y, std::min(t->m_valid.z, static_cast<int>(t->m_TEX0.TBW) * 64), t->m_valid.w - 1);
+				const u32 adjusted_endblock = GSLocalMemory::GetEndBlockAddress(t->m_TEX0.TBP0, t->m_TEX0.TBW, t->m_TEX0.PSM, adjusted_valid);
+				if (adjusted_endblock <= bp)
+				{
+					i++;
+					continue;
+				}
+
 				const u32 widthpage_offset = (std::abs(static_cast<int>(bp - t->m_TEX0.TBP0)) >> 5) % std::max(t->m_TEX0.TBW, 1U);
 				const bool is_aligned_ok = widthpage_offset == 0 || ((min_rect.width() <= static_cast<int>((t->m_TEX0.TBW - widthpage_offset) * 64) && (t->m_TEX0.TBW == TEX0.TBW || TEX0.TBW == 1)) && bp >= t->m_TEX0.TBP0);
 				const bool no_target_or_newer = (!dst || ((GSState::s_n - dst->m_last_draw) < (GSState::s_n - t->m_last_draw)));
@@ -2344,10 +2398,11 @@ GSTextureCache::Target* GSTextureCache::LookupTarget(GIFRegTEX0 TEX0, const GSVe
 				const bool ds_offset = !ds || offset != 0;
 				const bool is_double_buffer = TEX0.TBP0 == ((((t->m_end_block + 1) - t->m_TEX0.TBP0) / 2) + t->m_TEX0.TBP0);
 				const bool source_match = src && src->m_TEX0.TBP0 == bp && src->m_TEX0.TBW == TEX0.TBW && src->m_from_target && src->m_from_target == t;
+				const bool was_used_last_draw = t->m_last_draw == (GSState::s_n - 1);
 				// if it's a shuffle, some games tend to offset back by a page, such as Tomb Raider, for no disernable reason, but it then causes problems.
 				// This can also happen horizontally (Catwoman moves everything one page left with shuffles), but this is too messy to deal with right now.
 				const bool overlaps = t->Overlaps(bp, TEX0.TBW, TEX0.PSM, min_rect) || (is_shuffle && src && GSLocalMemory::m_psm[src->m_TEX0.PSM].bpp == 8 && t->Overlaps(bp, TEX0.TBW, TEX0.PSM, min_rect + GSVector4i(0, 0, 0, 32)));
-				if (source_match || (no_target_or_newer && is_aligned_ok && width_match && overlaps && (is_shuffle || ds_offset || is_double_buffer)))
+				if (source_match || (no_target_or_newer && is_aligned_ok && width_match && overlaps && (is_shuffle || ds_offset || is_double_buffer || was_used_last_draw)))
 				{
 					const GSLocalMemory::psm_t& s_psm = GSLocalMemory::m_psm[TEX0.PSM];
 
@@ -2355,9 +2410,10 @@ GSTextureCache::Target* GSTextureCache::LookupTarget(GIFRegTEX0 TEX0, const GSVe
 					// 896 is just 448 * 2,just gives the buffer chance to be larger than normal, in case they do something like 640x640, or something ridiculous.
 					if (!is_shuffle && (ds && offset == 0 && (t->m_valid.w >= 896) && ((((t->m_end_block + 1) - t->m_TEX0.TBP0) >> 1) + t->m_TEX0.TBP0) <= bp))
 					{
-						t->m_valid.w /= 2;
-						t->m_end_block = ((((t->m_end_block + 1) - t->m_TEX0.TBP0) >> 1) + t->m_TEX0.TBP0) - 1;
-						continue;
+						u32 offset = (((bp - t->m_TEX0.TBP0) >> 5) / std::max(t->m_TEX0.TBW, 1U)) * s_psm.pgs.y;
+						dst = CreateTarget(TEX0, GSVector2i(t->m_valid.z, t->m_valid.w - offset), GSVector2i(t->m_valid.z, t->m_valid.w - offset), scale, type, true, fbmask, false, false, preserve_rgb || preserve_alpha, GSVector4i::zero(), src);
+						dst->m_32_bits_fmt |= (psm_s.bpp != 16);
+						break;
 					}
 
 					// I know what you're thinking, and I hate the guy who wrote it too (me). Project Snowblind, Tomb Raider etc decide to offset where they're drawing using a channel shuffle, and this gets messy, so best just to kill the old target.
@@ -3359,7 +3415,7 @@ bool GSTextureCache::PreloadTarget(GIFRegTEX0 TEX0, const GSVector2i& size, cons
 				Target* t = *j;
 
 				if (dst != t && t->m_TEX0.PSM == dst->m_TEX0.PSM && t->Overlaps(dst->m_TEX0.TBP0, dst->m_TEX0.TBW, dst->m_TEX0.PSM, dst_valid) &&
-					static_cast<int>(((t->m_TEX0.TBP0 - dst->m_TEX0.TBP0) / 32) % std::max(dst->m_TEX0.TBW, 1U)) <= std::max(0, static_cast<int>(dst->m_TEX0.TBW - t->m_TEX0.TBW)))
+					((std::abs(static_cast<int>(t->m_TEX0.TBP0 - dst->m_TEX0.TBP0)) >> 5) % std::max(static_cast<int>(dst->m_TEX0.TBW), 1)) <= std::max(0, static_cast<int>(dst->m_TEX0.TBW - t->m_TEX0.TBW)))
 				{
 					const u32 buffer_width = std::max(1U, dst->m_TEX0.TBW);
 
@@ -3382,7 +3438,7 @@ bool GSTextureCache::PreloadTarget(GIFRegTEX0 TEX0, const GSVector2i& size, cons
 					}
 					// If the two targets are misaligned, it's likely a relocation, so we can just kill the old target.
 					// Kill targets that are overlapping new targets, but ignore the copy if the old target is dirty  because we favour GS memory.
-					if (((((t->m_TEX0.TBP0 - dst->m_TEX0.TBP0) >> 5) % buffer_width) != 0) && !t->m_dirty.empty())
+					if (((std::abs(static_cast<int>(t->m_TEX0.TBP0 - dst->m_TEX0.TBP0) >> 5) % buffer_width) != 0) && !t->m_dirty.empty())
 					{
 						InvalidateSourcesFromTarget(t);
 						i = list.erase(j);
@@ -3396,6 +3452,23 @@ bool GSTextureCache::PreloadTarget(GIFRegTEX0 TEX0, const GSVector2i& size, cons
 					{
 						GSVector4i new_valid = t->m_valid;
 						new_valid.w /= 2;
+						if (preserve_target && t->m_scale == dst->m_scale && dst->m_type == t->m_type && dst->m_dirty.empty() && !t->m_drawn_since_read.rintersect(new_valid).eq(t->m_drawn_since_read))
+						{
+							// Clamp the copy inside the source and destination.
+							const GSVector4i copy_rect = GSVector4i(GSVector4((new_valid + GSVector4i(0, new_valid.w).xyxy()).rintersect(t->m_drawn_since_read).rintersect(GSVector4i(0, 0, dst->m_unscaled_size.x, new_valid.w + dst->m_unscaled_size.y))) * dst->m_scale);
+							// Copy over the double buffer data, in case we need it.
+							// Clear the dirty first
+							t->Update();
+
+							dst->m_valid_rgb = t->m_valid_rgb;
+							dst->m_valid_alpha_low = t->m_valid_alpha_low;
+							dst->m_valid_alpha_high = t->m_valid_alpha_high;
+							dst->m_alpha_max = t->m_alpha_max;
+							dst->m_alpha_min = t->m_alpha_min;
+							dst->m_rt_alpha_scale = t->m_rt_alpha_scale;
+
+							g_gs_device->CopyRect(t->m_texture, dst->m_texture, copy_rect, 0, 0);
+						}
 						GL_INS("TC: RT resize buffer for FBP 0x%x, %dx%d => %d,%d", t->m_TEX0.TBP0, t->m_valid.width(), t->m_valid.height(), new_valid.width(), new_valid.height());
 						t->ResizeValidity(new_valid);
 						return hw_clear.value_or(false);
@@ -3545,11 +3618,11 @@ bool GSTextureCache::PreloadTarget(GIFRegTEX0 TEX0, const GSVector2i& size, cons
 							continue;
 						}
 
-						const int height_adjust = (((dst_end_block - t->m_TEX0.TBP0) >> 5) / std::max(t->m_TEX0.TBW, 1U)) * GSLocalMemory::m_psm[t->m_TEX0.PSM].pgs.y;
+						const int height_adjust = ((((dst_end_block + 31) - t->m_TEX0.TBP0) >> 5) / std::max(t->m_TEX0.TBW, 1U)) * GSLocalMemory::m_psm[t->m_TEX0.PSM].pgs.y;
 
 						if (height_adjust < t->m_unscaled_size.y)
 						{
-							t->m_TEX0.TBP0 = dst_end_block;
+							t->m_TEX0.TBP0 = GSLocalMemory::GetStartBlockAddress(t->m_TEX0.TBP0, t->m_TEX0.TBW, t->m_TEX0.PSM, GSVector4i(0, height_adjust, t->m_valid.z, t->m_valid.w));
 							t->m_valid.w -= height_adjust;
 							t->ResizeValidity(t->m_valid);
 
@@ -3805,6 +3878,9 @@ void GSTextureCache::ScaleTargetForDisplay(Target* t, const GIFRegTEX0& dispfb,
 	{
 		const GSVector4i right(old_width, 0, preload_width, needed_height);
 		const GSVector4i bottom(0, old_height, old_width, needed_height);
+
+		t->UpdateValidity(right.rintersect(bottom));
+
 		AddDirtyRectTarget(t, right, t->m_TEX0.PSM, t->m_TEX0.TBW, rgba);
 		AddDirtyRectTarget(t, bottom, t->m_TEX0.PSM, t->m_TEX0.TBW, rgba);
 	}
@@ -3813,6 +3889,9 @@ void GSTextureCache::ScaleTargetForDisplay(Target* t, const GIFRegTEX0& dispfb,
 		const GSVector4i newrect = GSVector4i((old_height < new_height) ? 0 : old_width,
 			(old_width < preload_width) ? 0 : old_height,
 			preload_width, needed_height);
+
+		t->UpdateValidity(newrect);
+
 		AddDirtyRectTarget(t, newrect, t->m_TEX0.PSM, t->m_TEX0.TBW, rgba);
 	}
 
@@ -4341,7 +4420,9 @@ void GSTextureCache::InvalidateVideoMem(const GSOffset& off, const GSVector4i& r
 								InvalidateTemporaryZ();
 						}
 
-						if (GSLocalMemory::m_psm[psm].depth)
+						// If we're dealing with quadrant draws, we need to position them correctly (Final Fantasy X).
+						if (GSLocalMemory::m_psm[psm].depth &&
+							r.width() <= (GSLocalMemory::m_psm[psm].pgs.x >> 1) && r.height() <= (GSLocalMemory::m_psm[psm].pgs.y >> 1))
 							DirtyRectByPage(bp, psm, bw, t, r);
 						else
 							AddDirtyRectTarget(t, r, psm, bw, rgba);
@@ -4735,7 +4816,7 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 	// Not even going to go down the rabbit hole of palette formats on the GPU.. We shouldn't have any targets with P4/P8 anyway.
 	const GSLocalMemory::psm_t& spsm_s = GSLocalMemory::m_psm[SPSM];
 	const GSLocalMemory::psm_t& dpsm_s = GSLocalMemory::m_psm[DPSM];
-	if (SPSM != DPSM || ((spsm_s.pal + dpsm_s.pal) != 0 && !alpha_only))
+	if (GSLocalMemory::m_psm[SPSM].bpp != GSLocalMemory::m_psm[DPSM].bpp || ((spsm_s.pal + dpsm_s.pal) != 0 && !alpha_only))
 	{
 		GL_CACHE("TC: Skipping HW move from 0x%X to 0x%X with SPSM=%s DPSM=%s", SBP, DBP, psm_str(SPSM), psm_str(DPSM));
 		return false;
@@ -4751,6 +4832,12 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 			return false;
 	}
 
+	bool req_resize = false;
+
+	// Save for later in case of page copy.
+	const u32 start_SBP = SBP;
+	const u32 start_DBP = DBP;
+
 	if (m_expected_src_bp == static_cast<int>(SBP) && m_expected_dst_bp == static_cast<int>(DBP))
 	{
 		// Get the new position so we can work out the offset.
@@ -4761,17 +4848,18 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 		sy += rect_offset.y;
 		dx += rect_offset.x;
 		dy += rect_offset.y;
-
+		req_resize = true;
 		GL_INS("TC: Detected striped move, realigning from SBP %x->%x DBP %x->%x", SBP, m_remembered_src_bp, DBP, m_remembered_dst_bp);
 
 		SBP = m_remembered_src_bp;
 		DBP = m_remembered_dst_bp;
-
-		m_expected_src_bp = -1;
-		m_remembered_src_bp = -1;
-		m_expected_dst_bp = -1;
-		m_remembered_dst_bp = -1;
 	}
+
+	m_expected_src_bp = -1;
+	m_remembered_src_bp = -1;
+	m_expected_dst_bp = -1;
+	m_remembered_dst_bp = -1;
+
 	// Look for an exact match on the targets.
 	GSTextureCache::Target* src = GetExactTarget(SBP, SBW, spsm_s.depth ? DepthStencil : RenderTarget, SBP);
 	GSTextureCache::Target* dst = GetExactTarget(DBP, DBW, dpsm_s.depth ? DepthStencil : RenderTarget, DBP);
@@ -4806,15 +4894,18 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 		const GSVector2i target_size = GetTargetSize(DBP, DBW, DPSM, Common::AlignUpPow2(w, 64), h);
 		dst = LookupTarget(new_TEX0, target_size, src->m_scale, src->m_type);
 		if (!dst)
-			dst = CreateTarget(new_TEX0, target_size, target_size, src->m_scale, src->m_type);
+		{
+			dst = Target::Create(new_TEX0, target_size.x, target_size.y, src->m_scale, GSLocalMemory::m_psm[DPSM].depth, true);
+			if (!dst) [[unlikely]]
+				return false;
+		}
 		else // Expand if necessary (Silent hill 4 takes an old target which is smaller).
 		{
-			dst->ResizeTexture(std::max(dst->m_unscaled_size.x, target_size.x), std::max(dst->m_unscaled_size.y, target_size.y));
+			req_resize = true;
 
 			// If it was matched to an old target, make sure to clear the other type and update its information.
 			if (dst->m_was_dst_matched)
 			{
-				g_texture_cache->InvalidateVideoMemType(GSTextureCache::DepthStencil - dst->m_type, dst->m_TEX0.TBP0);
 				dst->m_TEX0 = new_TEX0;
 			}
 		}
@@ -4829,6 +4920,24 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 	if (!src || !dst || src->m_scale != dst->m_scale)
 		return false;
 
+	// If we have an offset, adjust the base positions
+	if (src->m_TEX0.TBP0 != SBP)
+	{
+		const GSVector4i offset = TranslateAlignedRectByPage(src, SBP, SPSM, SBW, GSVector4i(0, 1).xxyy(), false);
+
+		sx += offset.x;
+		sy += offset.y;
+	}
+
+	if (dst->m_TEX0.TBP0 != DBP)
+	{
+		const GSVector4i offset = TranslateAlignedRectByPage(dst, DBP, DPSM, DBW, GSVector4i(0, 1).xxyy(), false);
+
+		dx += offset.x;
+		dy += offset.y;
+		req_resize = true;
+	}
+
 	// Scale coordinates.
 	const float scale = src->m_scale;
 	const int scaled_sx = static_cast<int>(sx * scale);
@@ -4842,6 +4951,11 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 	if ((scaled_sx + scaled_w) > src->m_texture->GetWidth() || (scaled_sy + scaled_h) > src->m_texture->GetHeight())
 		return false;
 
+	if (req_resize)
+	{
+		const GSVector2i target_size = GetTargetSize(DBP, DBW, DPSM, Common::AlignUpPow2(dx + w, 64), dx + h);
+		dst->ResizeTexture(std::max(dst->m_unscaled_size.x, target_size.x), std::max(dst->m_unscaled_size.y, target_size.y));
+	}
 	// We don't want to copy "old" data that the game has overwritten with writes,
 	// so flush any overlapping dirty area.
 	src->UpdateIfDirtyIntersects(GSVector4i(sx, sy, sx + w, sy + h));
@@ -4850,6 +4964,9 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 	// we're not going to be able to use it as a source.
 	dst->Update();
 
+	// Invalidate any opposite targets.
+	g_texture_cache->InvalidateVideoMemType(GSTextureCache::DepthStencil - dst->m_type, dst->m_TEX0.TBP0);
+
 	// Expand the target when we used a more conservative size.
 	const int required_dh = scaled_dy + scaled_h;
 	if ((scaled_dx + scaled_w) <= dst->m_texture->GetWidth() && required_dh > dst->m_texture->GetHeight())
@@ -4938,9 +5055,30 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 	{
 		if (SPSM == PSMT8H && SPSM == DPSM)
 		{
+			ShaderConvert shader = ShaderConvert::COPY;
+
 			const GSVector4 src_rect = GSVector4(scaled_sx, scaled_sy, scaled_sx + scaled_w, scaled_sy + scaled_h) / (GSVector4(src->m_texture->GetSize()).xyxy());
 			const GSVector4 dst_rect = GSVector4(scaled_dx, scaled_dy, (scaled_dx + scaled_w), (scaled_dy + scaled_h));
-			g_gs_device->StretchRect(src->m_texture, src_rect, dst->m_texture, dst_rect, false, false, false, true);
+			g_gs_device->StretchRect(src->m_texture, src_rect, dst->m_texture, dst_rect, false, false, false, true, shader);
+		}
+		else if (src->m_type != dst->m_type)
+		{
+			ShaderConvert shader = dst->m_type ? ShaderConvert::RGBA8_TO_FLOAT32 : ShaderConvert::FLOAT32_TO_RGBA8;
+
+			switch (dpsm_s.trbpp)
+			{
+				case 24:
+					shader = dst->m_type ? ShaderConvert::RGBA8_TO_FLOAT24 : ShaderConvert::FLOAT32_TO_RGB8;
+					break;
+				case 16:
+					shader = dst->m_type ? ShaderConvert::RGB5A1_TO_FLOAT16 : ShaderConvert::FLOAT16_TO_RGB5A1;
+					break;
+				default:
+					break;
+			}
+			const GSVector4 src_rect = GSVector4(scaled_sx, scaled_sy, scaled_sx + scaled_w, scaled_sy + scaled_h) / (GSVector4(src->m_texture->GetSize()).xyxy());
+			g_gs_device->StretchRect(src->m_texture, src_rect, dst->m_texture, GSVector4(scaled_sx, scaled_sy, scaled_sx + scaled_w, scaled_sy + scaled_h), shader, false);
+
 		}
 		else
 		{
@@ -4968,20 +5106,29 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 		dst->m_alpha_range |= src->m_alpha_range;
 	}
 
-	if (GSLocalMemory::IsPageAligned(src->m_TEX0.PSM, GSVector4i(sx, sy, sx + w, sy + h)) && (w == GSLocalMemory::m_psm[src->m_TEX0.PSM].pgs.x || h == GSLocalMemory::m_psm[src->m_TEX0.PSM].pgs.y))
+	u32 page_mask = GSLocalMemory::IsPageAlignedMasked(src->m_TEX0.PSM, GSVector4i(sx, sy, sx + w, sy + h));
+	if (((page_mask & 0x0f0f) == 0x0f0f || (page_mask & 0xf0f0) == 0xf0f0) && (w == GSLocalMemory::m_psm[src->m_TEX0.PSM].pgs.x || h == GSLocalMemory::m_psm[src->m_TEX0.PSM].pgs.y))
 	{
-		// Vertical Strips
-		if (w == GSLocalMemory::m_psm[src->m_TEX0.PSM].pgs.x)
+		// Page copy
+		if (w == GSLocalMemory::m_psm[src->m_TEX0.PSM].pgs.x && h == GSLocalMemory::m_psm[src->m_TEX0.PSM].pgs.y)
+		{
+			m_expected_src_bp = start_SBP + BLOCKS_PER_PAGE;
+			m_expected_dst_bp = start_DBP + BLOCKS_PER_PAGE;
+		}
+		// Vertical Strips.
+		else if (w == GSLocalMemory::m_psm[src->m_TEX0.PSM].pgs.x)
 		{
 			m_expected_src_bp = GSLocalMemory::GetStartBlockAddress(src->m_TEX0.TBP0, src->m_TEX0.TBW, src->m_TEX0.PSM, GSVector4i(sx + w, 0, sx + w + w, h));
 			m_expected_dst_bp = GSLocalMemory::GetStartBlockAddress(dst->m_TEX0.TBP0, dst->m_TEX0.TBW, dst->m_TEX0.PSM, GSVector4i(dx + w, 0, dx + w + w, h));
 		}
+		// Horizontal Strips.
 		else
 		{
 			m_expected_src_bp = GSLocalMemory::GetStartBlockAddress(src->m_TEX0.TBP0, src->m_TEX0.TBW, src->m_TEX0.PSM, GSVector4i(0, sy + h, w, sy + h + h));
 			m_expected_dst_bp = GSLocalMemory::GetStartBlockAddress(dst->m_TEX0.TBP0, dst->m_TEX0.TBW, dst->m_TEX0.PSM, GSVector4i(0, dy + h, w, dy + h + h));
 		}
 
+		// Only check the source, the destination might need expanding.
 		if (static_cast<u32>(m_expected_src_bp) < src->UnwrappedEndBlock() && static_cast<u32>(m_expected_src_bp) >= src->m_TEX0.TBP0)
 		{
 			m_remembered_src_bp = src->m_TEX0.TBP0;
@@ -5003,7 +5150,7 @@ bool GSTextureCache::Move(u32 SBP, u32 SBW, u32 SPSM, int sx, int sy, u32 DBP, u
 
 	// Invalidate any sources that overlap with the target (since they're now stale).
 	InvalidateVideoMem(g_gs_renderer->m_mem.GetOffset(DBP, DBW, DPSM), GSVector4i(dx, dy, dx + w, dy + h), false);
-
+	CombineAlignedInsideTargets(dst);
 	return true;
 }
 
@@ -5190,7 +5337,7 @@ GSTextureCache::Target* GSTextureCache::GetExactTarget(u32 BP, u32 BW, int type,
 	{
 		Target* t = *it;
 		const u32 tgt_bw = std::max(t->m_TEX0.TBW, 1U);
-		if ((t->m_TEX0.TBP0 == BP || (GSConfig.UserHacks_TextureInsideRt >= GSTextureInRtMode::InsideTargets && t->m_TEX0.TBP0 < BP && (((BP - t->m_TEX0.TBP0) >> 5) % tgt_bw) == 0)) && tgt_bw == BW && t->UnwrappedEndBlock() >= end_bp)
+		if ((t->m_TEX0.TBP0 == BP || (GSConfig.UserHacks_TextureInsideRt >= GSTextureInRtMode::InsideTargets && t->m_TEX0.TBP0 < BP && !(BP & 0x1f) && (((BP - t->m_TEX0.TBP0) >> 5) % tgt_bw) == 0)) && tgt_bw == BW && t->UnwrappedEndBlock() >= end_bp)
 		{
 			rts.MoveFront(it.Index());
 			return t;
@@ -5667,7 +5814,7 @@ GSTextureCache::Source* GSTextureCache::CreateSource(const GIFRegTEX0& TEX0, con
 		if (is_8bits)
 		{
 			GL_INS("TC: Reading RT as a packed-indexed 8 bits format");
-			shader = ShaderConvert::RGBA_TO_8I;
+			shader = GSLocalMemory::m_psm[dst->m_TEX0.PSM].bpp == 16 ? ShaderConvert::RGB5A1_TO_8I : ShaderConvert::RGBA_TO_8I;
 		}
 
 #ifdef ENABLE_OGL_DEBUG
@@ -5737,7 +5884,11 @@ GSTextureCache::Source* GSTextureCache::CreateSource(const GIFRegTEX0& TEX0, con
 			{
 				// We're inside the target, so conversion needs to happen on the entire target so we can offset properly.
 				src->m_unscaled_size.x = dst->m_unscaled_size.x * 2;
-				src->m_unscaled_size.y = dst->m_unscaled_size.y * 2;
+				if (GSLocalMemory::m_psm[dst->m_TEX0.PSM].bpp == 32)
+					src->m_unscaled_size.y = dst->m_unscaled_size.y * 2;
+				else
+					src->m_unscaled_size.y = dst->m_unscaled_size.y;
+
 				new_size.x = src->m_unscaled_size.x;
 				new_size.y = src->m_unscaled_size.y;
 			}
@@ -5908,7 +6059,8 @@ GSTextureCache::Source* GSTextureCache::CreateSource(const GIFRegTEX0& TEX0, con
 
 					// Adjust to match a PSMT8 texture (coordinates are double C32, we shouldn't be converting from anything else).
 					x_offset *= 2;
-					y_offset *= 2;
+					if (GSLocalMemory::m_psm[dst->m_TEX0.PSM].bpp == 32) 
+						y_offset *= 2;
 
 					src->m_region.SetX(x_offset, x_offset + tw);
 					src->m_region.SetY(y_offset, y_offset + th);
@@ -6695,6 +6847,8 @@ void GSTextureCache::Read(Target* t, const GSVector4i& r)
 	{
 		case PSMCT32:
 		case PSMCT24:
+		case PSMZ32:
+		case PSMZ24:
 		{
 			// If we're downloading a depth buffer that's been reinterpreted as a color
 			// format, convert it to integer. The format/swizzle is likely wrong, but it's
@@ -6720,27 +6874,11 @@ void GSTextureCache::Read(Target* t, const GSVector4i& r)
 
 		case PSMCT16:
 		case PSMCT16S:
-		{
-			fmt = GSTexture::Format::UInt16;
-			ps_shader = is_depth ? ShaderConvert::FLOAT32_TO_16_BITS : ShaderConvert::RGBA8_TO_16_BITS;
-			dltex = &m_uint16_download_texture;
-		}
-		break;
-
-		case PSMZ32:
-		case PSMZ24:
-		{
-			fmt = GSTexture::Format::UInt32;
-			ps_shader = ShaderConvert::FLOAT32_TO_32_BITS;
-			dltex = &m_uint32_download_texture;
-		}
-		break;
-
 		case PSMZ16:
 		case PSMZ16S:
 		{
 			fmt = GSTexture::Format::UInt16;
-			ps_shader = ShaderConvert::FLOAT32_TO_16_BITS;
+			ps_shader = is_depth ? ShaderConvert::FLOAT32_TO_16_BITS : ShaderConvert::RGBA8_TO_16_BITS;
 			dltex = &m_uint16_download_texture;
 		}
 		break;
@@ -7953,10 +8091,11 @@ GSTextureCache::TempZAddress GSTextureCache::GetTemporaryZInfo()
 	return m_temporary_z_info;
 }
 
-void GSTextureCache::SetTemporaryZInfo(u32 address, u32 offset)
+void GSTextureCache::SetTemporaryZInfo(u32 address, u32 offset, u32 rt_offset)
 {
 	m_temporary_z_info.ZBP = address;
 	m_temporary_z_info.offset = offset;
+	m_temporary_z_info.rt_offset = rt_offset;
 	m_temporary_z_info.rect_since = GSVector4i::zero();
 }
 void GSTextureCache::SetTemporaryZInfo(TempZAddress address_info)