28 files changed, 698 insertions, 130 deletions

diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index 91a30fef7..6a6325e38 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -3,6 +3,7 @@ add_subdirectory(host_shaders)
 if(LIBVA_FOUND)
     set_source_files_properties(command_classes/codecs/codec.cpp
         PROPERTIES COMPILE_DEFINITIONS LIBVA_FOUND=1)
+    list(APPEND FFmpeg_LIBRARIES ${LIBVA_LIBRARIES})
 endif()
 
 add_library(video_core STATIC
diff --git a/src/video_core/command_classes/codecs/codec.cpp b/src/video_core/command_classes/codecs/codec.cpp
index 916277811..2a532b883 100644
--- a/src/video_core/command_classes/codecs/codec.cpp
+++ b/src/video_core/command_classes/codecs/codec.cpp
@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
+#include <cstdio>
 #include <fstream>
 #include <vector>
 #include "common/assert.h"
@@ -15,12 +17,28 @@
 
 extern "C" {
 #include <libavutil/opt.h>
+#ifdef LIBVA_FOUND
+// for querying VAAPI driver information
+#include <libavutil/hwcontext_vaapi.h>
+#endif
 }
 
 namespace Tegra {
 namespace {
 constexpr AVPixelFormat PREFERRED_GPU_FMT = AV_PIX_FMT_NV12;
 constexpr AVPixelFormat PREFERRED_CPU_FMT = AV_PIX_FMT_YUV420P;
+constexpr std::array PREFERRED_GPU_DECODERS = {
+    AV_HWDEVICE_TYPE_CUDA,
+#ifdef _WIN32
+    AV_HWDEVICE_TYPE_D3D11VA,
+    AV_HWDEVICE_TYPE_DXVA2,
+#elif defined(__linux__)
+    AV_HWDEVICE_TYPE_VAAPI,
+    AV_HWDEVICE_TYPE_VDPAU,
+#endif
+    // last resort for Linux Flatpak (w/ NVIDIA)
+    AV_HWDEVICE_TYPE_VULKAN,
+};
 
 void AVPacketDeleter(AVPacket* ptr) {
     av_packet_free(&ptr);
@@ -59,46 +77,50 @@ Codec::~Codec() {
     av_buffer_unref(&av_gpu_decoder);
 }
 
+// List all the currently available hwcontext in ffmpeg
+static std::vector<AVHWDeviceType> ListSupportedContexts() {
+    std::vector<AVHWDeviceType> contexts{};
+    AVHWDeviceType current_device_type = AV_HWDEVICE_TYPE_NONE;
+    do {
+        current_device_type = av_hwdevice_iterate_types(current_device_type);
+        contexts.push_back(current_device_type);
+    } while (current_device_type != AV_HWDEVICE_TYPE_NONE);
+    return contexts;
+}
+
 bool Codec::CreateGpuAvDevice() {
-#if defined(LIBVA_FOUND)
-    static constexpr std::array<const char*, 3> VAAPI_DRIVERS = {
-        "i915",
-        "iHD",
-        "amdgpu",
-    };
-    AVDictionary* hwdevice_options = nullptr;
-    av_dict_set(&hwdevice_options, "connection_type", "drm", 0);
-    for (const auto& driver : VAAPI_DRIVERS) {
-        av_dict_set(&hwdevice_options, "kernel_driver", driver, 0);
-        const int hwdevice_error = av_hwdevice_ctx_create(&av_gpu_decoder, AV_HWDEVICE_TYPE_VAAPI,
-                                                          nullptr, hwdevice_options, 0);
-        if (hwdevice_error >= 0) {
-            LOG_INFO(Service_NVDRV, "Using VA-API with {}", driver);
-            av_dict_free(&hwdevice_options);
-            av_codec_ctx->pix_fmt = AV_PIX_FMT_VAAPI;
-            return true;
-        }
-        LOG_DEBUG(Service_NVDRV, "VA-API av_hwdevice_ctx_create failed {}", hwdevice_error);
-    }
-    LOG_DEBUG(Service_NVDRV, "VA-API av_hwdevice_ctx_create failed for all drivers");
-    av_dict_free(&hwdevice_options);
-#endif
     static constexpr auto HW_CONFIG_METHOD = AV_CODEC_HW_CONFIG_METHOD_HW_DEVICE_CTX;
-    static constexpr std::array GPU_DECODER_TYPES{
-        AV_HWDEVICE_TYPE_CUDA,
-#ifdef _WIN32
-        AV_HWDEVICE_TYPE_D3D11VA,
-#else
-        AV_HWDEVICE_TYPE_VDPAU,
-#endif
-    };
-    for (const auto& type : GPU_DECODER_TYPES) {
+    static const auto supported_contexts = ListSupportedContexts();
+    for (const auto& type : PREFERRED_GPU_DECODERS) {
+        if (std::none_of(supported_contexts.begin(), supported_contexts.end(),
+                         [&type](const auto& context) { return context == type; })) {
+            LOG_DEBUG(Service_NVDRV, "{} explicitly unsupported", av_hwdevice_get_type_name(type));
+            continue;
+        }
         const int hwdevice_res = av_hwdevice_ctx_create(&av_gpu_decoder, type, nullptr, nullptr, 0);
         if (hwdevice_res < 0) {
             LOG_DEBUG(Service_NVDRV, "{} av_hwdevice_ctx_create failed {}",
                       av_hwdevice_get_type_name(type), hwdevice_res);
             continue;
         }
+#ifdef LIBVA_FOUND
+        if (type == AV_HWDEVICE_TYPE_VAAPI) {
+            // we need to determine if this is an impersonated VAAPI driver
+            AVHWDeviceContext* hwctx =
+                static_cast<AVHWDeviceContext*>(static_cast<void*>(av_gpu_decoder->data));
+            AVVAAPIDeviceContext* vactx = static_cast<AVVAAPIDeviceContext*>(hwctx->hwctx);
+            const char* vendor_name = vaQueryVendorString(vactx->display);
+            if (strstr(vendor_name, "VDPAU backend")) {
+                // VDPAU impersonated VAAPI impl's are super buggy, we need to skip them
+                LOG_DEBUG(Service_NVDRV, "Skipping vdapu impersonated VAAPI driver");
+                continue;
+            } else {
+                // according to some user testing, certain vaapi driver (Intel?) could be buggy
+                // so let's log the driver name which may help the developers/supporters
+                LOG_DEBUG(Service_NVDRV, "Using VAAPI driver: {}", vendor_name);
+            }
+        }
+#endif
         for (int i = 0;; i++) {
             const AVCodecHWConfig* config = avcodec_get_hw_config(av_codec, i);
             if (!config) {
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index ab7c21a49..8788f5148 100644
--- a/src/video_core/gpu.cpp
+++ b/src/video_core/gpu.cpp
@@ -185,16 +185,6 @@ struct GPU::Impl {
         return *dma_pusher;
     }
 
-    /// Returns a reference to the GPU CDMA pusher.
-    [[nodiscard]] Tegra::CDmaPusher& CDmaPusher() {
-        return *cdma_pusher;
-    }
-
-    /// Returns a const reference to the GPU CDMA pusher.
-    [[nodiscard]] const Tegra::CDmaPusher& CDmaPusher() const {
-        return *cdma_pusher;
-    }
-
     /// Returns a reference to the underlying renderer.
     [[nodiscard]] VideoCore::RendererBase& Renderer() {
         return *renderer;
@@ -338,25 +328,27 @@ struct GPU::Impl {
     }
 
     /// Push GPU command buffer entries to be processed
-    void PushCommandBuffer(Tegra::ChCommandHeaderList& entries) {
+    void PushCommandBuffer(u32 id, Tegra::ChCommandHeaderList& entries) {
         if (!use_nvdec) {
             return;
         }
 
-        if (!cdma_pusher) {
-            cdma_pusher = std::make_unique<Tegra::CDmaPusher>(gpu);
+        if (!cdma_pushers.contains(id)) {
+            cdma_pushers.insert_or_assign(id, std::make_unique<Tegra::CDmaPusher>(gpu));
         }
 
         // SubmitCommandBuffer would make the nvdec operations async, this is not currently working
         // TODO(ameerj): RE proper async nvdec operation
         // gpu_thread.SubmitCommandBuffer(std::move(entries));
-
-        cdma_pusher->ProcessEntries(std::move(entries));
+        cdma_pushers[id]->ProcessEntries(std::move(entries));
     }
 
     /// Frees the CDMAPusher instance to free up resources
-    void ClearCdmaInstance() {
-        cdma_pusher.reset();
+    void ClearCdmaInstance(u32 id) {
+        const auto iter = cdma_pushers.find(id);
+        if (iter != cdma_pushers.end()) {
+            cdma_pushers.erase(iter);
+        }
     }
 
     /// Swap buffers (render frame)
@@ -659,7 +651,7 @@ struct GPU::Impl {
     Core::System& system;
     std::unique_ptr<Tegra::MemoryManager> memory_manager;
     std::unique_ptr<Tegra::DmaPusher> dma_pusher;
-    std::unique_ptr<Tegra::CDmaPusher> cdma_pusher;
+    std::map<u32, std::unique_ptr<Tegra::CDmaPusher>> cdma_pushers;
     std::unique_ptr<VideoCore::RendererBase> renderer;
     VideoCore::RasterizerInterface* rasterizer = nullptr;
     const bool use_nvdec;
@@ -811,14 +803,6 @@ const Tegra::DmaPusher& GPU::DmaPusher() const {
     return impl->DmaPusher();
 }
 
-Tegra::CDmaPusher& GPU::CDmaPusher() {
-    return impl->CDmaPusher();
-}
-
-const Tegra::CDmaPusher& GPU::CDmaPusher() const {
-    return impl->CDmaPusher();
-}
-
 VideoCore::RendererBase& GPU::Renderer() {
     return impl->Renderer();
 }
@@ -887,12 +871,12 @@ void GPU::PushGPUEntries(Tegra::CommandList&& entries) {
     impl->PushGPUEntries(std::move(entries));
 }
 
-void GPU::PushCommandBuffer(Tegra::ChCommandHeaderList& entries) {
-    impl->PushCommandBuffer(entries);
+void GPU::PushCommandBuffer(u32 id, Tegra::ChCommandHeaderList& entries) {
+    impl->PushCommandBuffer(id, entries);
 }
 
-void GPU::ClearCdmaInstance() {
-    impl->ClearCdmaInstance();
+void GPU::ClearCdmaInstance(u32 id) {
+    impl->ClearCdmaInstance(id);
 }
 
 void GPU::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index 05e5c94f3..500411176 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -83,6 +83,7 @@ enum class DepthFormat : u32 {
     S8_UINT_Z24_UNORM = 0x14,
     D24X8_UNORM = 0x15,
     D24S8_UNORM = 0x16,
+    S8_UINT = 0x17,
     D24C8_UNORM = 0x18,
     D32_FLOAT_S8X24_UINT = 0x19,
 };
@@ -241,10 +242,10 @@ public:
     void PushGPUEntries(Tegra::CommandList&& entries);
 
     /// Push GPU command buffer entries to be processed
-    void PushCommandBuffer(Tegra::ChCommandHeaderList& entries);
+    void PushCommandBuffer(u32 id, Tegra::ChCommandHeaderList& entries);
 
     /// Frees the CDMAPusher instance to free up resources
-    void ClearCdmaInstance();
+    void ClearCdmaInstance(u32 id);
 
     /// Swap buffers (render frame)
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer);
diff --git a/src/video_core/host_shaders/CMakeLists.txt b/src/video_core/host_shaders/CMakeLists.txt
index d779a967a..fd3e41434 100644
--- a/src/video_core/host_shaders/CMakeLists.txt
+++ b/src/video_core/host_shaders/CMakeLists.txt
@@ -10,6 +10,8 @@ set(SHADER_FILES
     astc_decoder.comp
     block_linear_unswizzle_2d.comp
     block_linear_unswizzle_3d.comp
+    convert_abgr8_to_d24s8.frag
+    convert_d24s8_to_abgr8.frag
     convert_depth_to_float.frag
     convert_float_to_depth.frag
     full_screen_triangle.vert
diff --git a/src/video_core/host_shaders/convert_abgr8_to_d24s8.frag b/src/video_core/host_shaders/convert_abgr8_to_d24s8.frag
new file mode 100644
index 000000000..ea055ddad
--- /dev/null
+++ b/src/video_core/host_shaders/convert_abgr8_to_d24s8.frag
@@ -0,0 +1,18 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 450
+#extension GL_ARB_shader_stencil_export : require
+
+layout(binding = 0) uniform sampler2D color_texture;
+
+void main() {
+    ivec2 coord = ivec2(gl_FragCoord.xy);
+    uvec4 color = uvec4(texelFetch(color_texture, coord, 0).abgr * (exp2(8) - 1.0f));
+    uvec4 bytes = color << uvec4(24, 16, 8, 0);
+    uint depth_stencil_unorm = bytes.x | bytes.y | bytes.z | bytes.w;
+
+    gl_FragDepth = float(depth_stencil_unorm & 0x00FFFFFFu) / (exp2(24.0) - 1.0f);
+    gl_FragStencilRefARB = int(depth_stencil_unorm >> 24);
+}
diff --git a/src/video_core/host_shaders/convert_d24s8_to_abgr8.frag b/src/video_core/host_shaders/convert_d24s8_to_abgr8.frag
new file mode 100644
index 000000000..94368fb59
--- /dev/null
+++ b/src/video_core/host_shaders/convert_d24s8_to_abgr8.frag
@@ -0,0 +1,23 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 450
+
+layout(binding = 0) uniform sampler2D depth_tex;
+layout(binding = 1) uniform isampler2D stencil_tex;
+
+layout(location = 0) out vec4 color;
+
+void main() {
+    ivec2 coord = ivec2(gl_FragCoord.xy);
+    uint depth = uint(textureLod(depth_tex, coord, 0).r * (exp2(24.0) - 1.0f));
+    uint stencil = uint(textureLod(stencil_tex, coord, 0).r);
+
+    highp uint depth_val =
+        uint(textureLod(depth_tex, coord, 0).r * (exp2(32.0) - 1.0));
+    lowp uint stencil_val = textureLod(stencil_tex, coord, 0).r;
+    highp uvec4 components =
+        uvec4(stencil_val, (uvec3(depth_val) >> uvec3(24u, 16u, 8u)) & 0x000000FFu);
+    color.abgr = vec4(components) / (exp2(8.0) - 1.0);
+}
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index 6956535e5..14e6522f2 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -9,6 +9,7 @@
 
 #include <glad/glad.h>
 
+#include "common/bit_util.h"
 #include "common/literals.h"
 #include "common/settings.h"
 #include "video_core/renderer_opengl/gl_device.h"
@@ -148,6 +149,8 @@ GLenum AttachmentType(PixelFormat format) {
     switch (const SurfaceType type = VideoCore::Surface::GetFormatType(format); type) {
     case SurfaceType::Depth:
         return GL_DEPTH_ATTACHMENT;
+    case SurfaceType::Stencil:
+        return GL_STENCIL_ATTACHMENT;
     case SurfaceType::DepthStencil:
         return GL_DEPTH_STENCIL_ATTACHMENT;
     default:
@@ -317,13 +320,12 @@ void AttachTexture(GLuint fbo, GLenum attachment, const ImageView* image_view) {
     }
 }
 
-OGLTexture MakeImage(const VideoCommon::ImageInfo& info, GLenum gl_internal_format) {
+OGLTexture MakeImage(const VideoCommon::ImageInfo& info, GLenum gl_internal_format,
+                     GLsizei gl_num_levels) {
     const GLenum target = ImageTarget(info);
     const GLsizei width = info.size.width;
     const GLsizei height = info.size.height;
     const GLsizei depth = info.size.depth;
-    const int max_host_mip_levels = std::bit_width(info.size.width);
-    const GLsizei num_levels = std::min(info.resources.levels, max_host_mip_levels);
     const GLsizei num_layers = info.resources.layers;
     const GLsizei num_samples = info.num_samples;
 
@@ -335,10 +337,10 @@ OGLTexture MakeImage(const VideoCommon::ImageInfo& info, GLenum gl_internal_form
     }
     switch (target) {
     case GL_TEXTURE_1D_ARRAY:
-        glTextureStorage2D(handle, num_levels, gl_internal_format, width, num_layers);
+        glTextureStorage2D(handle, gl_num_levels, gl_internal_format, width, num_layers);
         break;
     case GL_TEXTURE_2D_ARRAY:
-        glTextureStorage3D(handle, num_levels, gl_internal_format, width, height, num_layers);
+        glTextureStorage3D(handle, gl_num_levels, gl_internal_format, width, height, num_layers);
         break;
     case GL_TEXTURE_2D_MULTISAMPLE_ARRAY: {
         // TODO: Where should 'fixedsamplelocations' come from?
@@ -348,10 +350,10 @@ OGLTexture MakeImage(const VideoCommon::ImageInfo& info, GLenum gl_internal_form
         break;
     }
     case GL_TEXTURE_RECTANGLE:
-        glTextureStorage2D(handle, num_levels, gl_internal_format, width, height);
+        glTextureStorage2D(handle, gl_num_levels, gl_internal_format, width, height);
         break;
     case GL_TEXTURE_3D:
-        glTextureStorage3D(handle, num_levels, gl_internal_format, width, height, depth);
+        glTextureStorage3D(handle, gl_num_levels, gl_internal_format, width, height, depth);
         break;
     case GL_TEXTURE_BUFFER:
         UNREACHABLE();
@@ -397,9 +399,6 @@ OGLTexture MakeImage(const VideoCommon::ImageInfo& info, GLenum gl_internal_form
     return GL_R32UI;
 }
 
-[[nodiscard]] u32 NextPow2(u32 value) {
-    return 1U << (32U - std::countl_zero(value - 1U));
-}
 } // Anonymous namespace
 
 ImageBufferMap::~ImageBufferMap() {
@@ -526,8 +525,8 @@ void TextureCacheRuntime::CopyImage(Image& dst_image, Image& src_image,
     }
 }
 
-void TextureCacheRuntime::ConvertImage(Image& dst, Image& src,
-                                       std::span<const VideoCommon::ImageCopy> copies) {
+void TextureCacheRuntime::ReinterpretImage(Image& dst, Image& src,
+                                           std::span<const VideoCommon::ImageCopy> copies) {
     LOG_DEBUG(Render_OpenGL, "Converting {} to {}", src.info.format, dst.info.format);
     format_conversion_pass.ConvertImage(dst, src, copies);
 }
@@ -696,7 +695,9 @@ Image::Image(TextureCacheRuntime& runtime_, const VideoCommon::ImageInfo& info_,
         gl_format = tuple.format;
         gl_type = tuple.type;
     }
-    texture = MakeImage(info, gl_internal_format);
+    const int max_host_mip_levels = std::bit_width(info.size.width);
+    gl_num_levels = std::min(info.resources.levels, max_host_mip_levels);
+    texture = MakeImage(info, gl_internal_format, gl_num_levels);
     current_texture = texture.handle;
     if (runtime->device.HasDebuggingToolAttached()) {
         const std::string name = VideoCommon::Name(*this);
@@ -724,6 +725,9 @@ void Image::UploadMemory(const ImageBufferMap& map,
     u32 current_image_height = std::numeric_limits<u32>::max();
 
     for (const VideoCommon::BufferImageCopy& copy : copies) {
+        if (copy.image_subresource.base_level >= gl_num_levels) {
+            continue;
+        }
         if (current_row_length != copy.buffer_row_length) {
             current_row_length = copy.buffer_row_length;
             glPixelStorei(GL_UNPACK_ROW_LENGTH, current_row_length);
@@ -753,6 +757,9 @@ void Image::DownloadMemory(ImageBufferMap& map,
     u32 current_image_height = std::numeric_limits<u32>::max();
 
     for (const VideoCommon::BufferImageCopy& copy : copies) {
+        if (copy.image_subresource.base_level >= gl_num_levels) {
+            continue;
+        }
         if (current_row_length != copy.buffer_row_length) {
             current_row_length = copy.buffer_row_length;
             glPixelStorei(GL_PACK_ROW_LENGTH, current_row_length);
@@ -792,7 +799,7 @@ GLuint Image::StorageHandle() noexcept {
         }
         store_view.Create();
         glTextureView(store_view.handle, ImageTarget(info), current_texture, GL_RGBA8, 0,
-                      info.resources.levels, 0, info.resources.layers);
+                      gl_num_levels, 0, info.resources.layers);
         return store_view.handle;
     default:
         return current_texture;
@@ -907,6 +914,8 @@ void Image::Scale(bool up_scale) {
             return GL_COLOR_ATTACHMENT0;
         case SurfaceType::Depth:
             return GL_DEPTH_ATTACHMENT;
+        case SurfaceType::Stencil:
+            return GL_STENCIL_ATTACHMENT;
         case SurfaceType::DepthStencil:
             return GL_DEPTH_STENCIL_ATTACHMENT;
         default:
@@ -920,8 +929,10 @@ void Image::Scale(bool up_scale) {
             return GL_COLOR_BUFFER_BIT;
         case SurfaceType::Depth:
             return GL_DEPTH_BUFFER_BIT;
+        case SurfaceType::Stencil:
+            return GL_STENCIL_BUFFER_BIT;
         case SurfaceType::DepthStencil:
-            return GL_STENCIL_BUFFER_BIT | GL_DEPTH_BUFFER_BIT;
+            return GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT;
         default:
             UNREACHABLE();
             return GL_COLOR_BUFFER_BIT;
@@ -933,8 +944,10 @@ void Image::Scale(bool up_scale) {
             return 0;
         case SurfaceType::Depth:
             return 1;
-        case SurfaceType::DepthStencil:
+        case SurfaceType::Stencil:
             return 2;
+        case SurfaceType::DepthStencil:
+            return 3;
         default:
             UNREACHABLE();
             return 0;
@@ -956,7 +969,7 @@ void Image::Scale(bool up_scale) {
         auto dst_info = info;
         dst_info.size.width = scaled_width;
         dst_info.size.height = scaled_height;
-        upscaled_backup = MakeImage(dst_info, gl_internal_format);
+        upscaled_backup = MakeImage(dst_info, gl_internal_format, gl_num_levels);
     }
     const u32 src_width = up_scale ? original_width : scaled_width;
     const u32 src_height = up_scale ? original_height : scaled_height;
@@ -1264,10 +1277,20 @@ Framebuffer::Framebuffer(TextureCacheRuntime& runtime, std::span<ImageView*, NUM
     }
 
     if (const ImageView* const image_view = depth_buffer; image_view) {
-        if (GetFormatType(image_view->format) == SurfaceType::DepthStencil) {
+        switch (GetFormatType(image_view->format)) {
+        case SurfaceType::Depth:
+            buffer_bits |= GL_DEPTH_BUFFER_BIT;
+            break;
+        case SurfaceType::Stencil:
+            buffer_bits |= GL_STENCIL_BUFFER_BIT;
+            break;
+        case SurfaceType::DepthStencil:
             buffer_bits |= GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT;
-        } else {
+            break;
+        default:
+            UNREACHABLE();
             buffer_bits |= GL_DEPTH_BUFFER_BIT;
+            break;
         }
         const GLenum attachment = AttachmentType(image_view->format);
         AttachTexture(handle, attachment, image_view);
@@ -1308,7 +1331,7 @@ void FormatConversionPass::ConvertImage(Image& dst_image, Image& src_image,
         const u32 copy_size = region.width * region.height * region.depth * img_bpp;
         if (pbo_size < copy_size) {
             intermediate_pbo.Create();
-            pbo_size = NextPow2(copy_size);
+            pbo_size = Common::NextPow2(copy_size);
             glNamedBufferData(intermediate_pbo.handle, pbo_size, nullptr, GL_STREAM_COPY);
         }
         // Copy from source to PBO
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h
index 578f8d523..37d5e6a6b 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.h
+++ b/src/video_core/renderer_opengl/gl_texture_cache.h
@@ -84,9 +84,13 @@ public:
 
     u64 GetDeviceLocalMemory() const;
 
+    bool ShouldReinterpret([[maybe_unused]] Image& dst, [[maybe_unused]] Image& src) {
+        return true;
+    }
+
     void CopyImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
 
-    void ConvertImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
+    void ReinterpretImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
 
     void ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view, bool rescaled) {
         UNIMPLEMENTED();
@@ -164,8 +168,8 @@ private:
 
     std::array<GLuint, Shader::NUM_TEXTURE_TYPES> null_image_views{};
 
-    std::array<OGLFramebuffer, 3> rescale_draw_fbos;
-    std::array<OGLFramebuffer, 3> rescale_read_fbos;
+    std::array<OGLFramebuffer, 4> rescale_draw_fbos;
+    std::array<OGLFramebuffer, 4> rescale_read_fbos;
     const Settings::ResolutionScalingInfo& resolution;
 };
 
@@ -221,6 +225,7 @@ private:
     GLenum gl_internal_format = GL_NONE;
     GLenum gl_format = GL_NONE;
     GLenum gl_type = GL_NONE;
+    GLsizei gl_num_levels{};
     TextureCacheRuntime* runtime{};
     GLuint current_texture{};
 };
@@ -338,7 +343,6 @@ struct TextureCacheParams {
     static constexpr bool FRAMEBUFFER_BLITS = true;
     static constexpr bool HAS_EMULATED_COPIES = true;
     static constexpr bool HAS_DEVICE_MEMORY_INFO = true;
-    static constexpr bool HAS_PIXEL_FORMAT_CONVERSIONS = true;
 
     using Runtime = OpenGL::TextureCacheRuntime;
     using Image = OpenGL::Image;
diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h
index 39158aa3e..daba42ed9 100644
--- a/src/video_core/renderer_opengl/maxwell_to_gl.h
+++ b/src/video_core/renderer_opengl/maxwell_to_gl.h
@@ -108,6 +108,7 @@ constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> FORMAT_TAB
     {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV},                // E5B9G9R9_FLOAT
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT},            // D32_FLOAT
     {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT},    // D16_UNORM
+    {GL_STENCIL_INDEX8, GL_STENCIL, GL_UNSIGNED_BYTE},                // S8_UINT
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8},    // D24_UNORM_S8_UINT
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8},    // S8_UINT_D24_UNORM
     {GL_DEPTH32F_STENCIL8, GL_DEPTH_STENCIL,
diff --git a/src/video_core/renderer_vulkan/blit_image.cpp b/src/video_core/renderer_vulkan/blit_image.cpp
index b3884a4f5..9a38b6b34 100644
--- a/src/video_core/renderer_vulkan/blit_image.cpp
+++ b/src/video_core/renderer_vulkan/blit_image.cpp
@@ -4,6 +4,8 @@
 
 #include <algorithm>
 
+#include "video_core/host_shaders/convert_abgr8_to_d24s8_frag_spv.h"
+#include "video_core/host_shaders/convert_d24s8_to_abgr8_frag_spv.h"
 #include "video_core/host_shaders/convert_depth_to_float_frag_spv.h"
 #include "video_core/host_shaders/convert_float_to_depth_frag_spv.h"
 #include "video_core/host_shaders/full_screen_triangle_vert_spv.h"
@@ -354,6 +356,8 @@ BlitImageHelper::BlitImageHelper(const Device& device_, VKScheduler& scheduler_,
       blit_color_to_color_frag(BuildShader(device, VULKAN_BLIT_COLOR_FLOAT_FRAG_SPV)),
       convert_depth_to_float_frag(BuildShader(device, CONVERT_DEPTH_TO_FLOAT_FRAG_SPV)),
       convert_float_to_depth_frag(BuildShader(device, CONVERT_FLOAT_TO_DEPTH_FRAG_SPV)),
+      convert_abgr8_to_d24s8_frag(BuildShader(device, CONVERT_ABGR8_TO_D24S8_FRAG_SPV)),
+      convert_d24s8_to_abgr8_frag(BuildShader(device, CONVERT_D24S8_TO_ABGR8_FRAG_SPV)),
       linear_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_LINEAR>)),
       nearest_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_NEAREST>)) {
     if (device.IsExtShaderStencilExportSupported()) {
@@ -448,6 +452,22 @@ void BlitImageHelper::ConvertR16ToD16(const Framebuffer* dst_framebuffer,
     Convert(*convert_r16_to_d16_pipeline, dst_framebuffer, src_image_view, up_scale, down_shift);
 }
 
+void BlitImageHelper::ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer,
+                                          ImageView& src_image_view, u32 up_scale, u32 down_shift) {
+    ConvertPipelineDepthTargetEx(convert_abgr8_to_d24s8_pipeline, dst_framebuffer->RenderPass(),
+                                 convert_abgr8_to_d24s8_frag, true);
+    ConvertColor(*convert_abgr8_to_d24s8_pipeline, dst_framebuffer, src_image_view, up_scale,
+                 down_shift);
+}
+
+void BlitImageHelper::ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer,
+                                          ImageView& src_image_view, u32 up_scale, u32 down_shift) {
+    ConvertPipelineColorTargetEx(convert_d24s8_to_abgr8_pipeline, dst_framebuffer->RenderPass(),
+                                 convert_d24s8_to_abgr8_frag, false);
+    ConvertDepthStencil(*convert_d24s8_to_abgr8_pipeline, dst_framebuffer, src_image_view, up_scale,
+                        down_shift);
+}
+
 void BlitImageHelper::Convert(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
                               const ImageView& src_image_view, u32 up_scale, u32 down_shift) {
     const VkPipelineLayout layout = *one_texture_pipeline_layout;
@@ -495,6 +515,101 @@ void BlitImageHelper::Convert(VkPipeline pipeline, const Framebuffer* dst_frameb
     scheduler.InvalidateState();
 }
 
+void BlitImageHelper::ConvertColor(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
+                                   ImageView& src_image_view, u32 up_scale, u32 down_shift) {
+    const VkPipelineLayout layout = *one_texture_pipeline_layout;
+    const VkImageView src_view = src_image_view.ColorView();
+    const VkSampler sampler = *nearest_sampler;
+    const VkExtent2D extent{
+        .width = std::max((src_image_view.size.width * up_scale) >> down_shift, 1U),
+        .height = std::max((src_image_view.size.height * up_scale) >> down_shift, 1U),
+    };
+    scheduler.RequestRenderpass(dst_framebuffer);
+    scheduler.Record([pipeline, layout, sampler, src_view, extent, up_scale, down_shift,
+                      this](vk::CommandBuffer cmdbuf) {
+        const VkOffset2D offset{
+            .x = 0,
+            .y = 0,
+        };
+        const VkViewport viewport{
+            .x = 0.0f,
+            .y = 0.0f,
+            .width = static_cast<float>(extent.width),
+            .height = static_cast<float>(extent.height),
+            .minDepth = 0.0f,
+            .maxDepth = 0.0f,
+        };
+        const VkRect2D scissor{
+            .offset = offset,
+            .extent = extent,
+        };
+        const PushConstants push_constants{
+            .tex_scale = {viewport.width, viewport.height},
+            .tex_offset = {0.0f, 0.0f},
+        };
+        const VkDescriptorSet descriptor_set = one_texture_descriptor_allocator.Commit();
+        UpdateOneTextureDescriptorSet(device, descriptor_set, sampler, src_view);
+
+        // TODO: Barriers
+        cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
+        cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set,
+                                  nullptr);
+        cmdbuf.SetViewport(0, viewport);
+        cmdbuf.SetScissor(0, scissor);
+        cmdbuf.PushConstants(layout, VK_SHADER_STAGE_VERTEX_BIT, push_constants);
+        cmdbuf.Draw(3, 1, 0, 0);
+    });
+    scheduler.InvalidateState();
+}
+
+void BlitImageHelper::ConvertDepthStencil(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
+                                          ImageView& src_image_view, u32 up_scale, u32 down_shift) {
+    const VkPipelineLayout layout = *two_textures_pipeline_layout;
+    const VkImageView src_depth_view = src_image_view.DepthView();
+    const VkImageView src_stencil_view = src_image_view.StencilView();
+    const VkSampler sampler = *nearest_sampler;
+    const VkExtent2D extent{
+        .width = std::max((src_image_view.size.width * up_scale) >> down_shift, 1U),
+        .height = std::max((src_image_view.size.height * up_scale) >> down_shift, 1U),
+    };
+    scheduler.RequestRenderpass(dst_framebuffer);
+    scheduler.Record([pipeline, layout, sampler, src_depth_view, src_stencil_view, extent, up_scale,
+                      down_shift, this](vk::CommandBuffer cmdbuf) {
+        const VkOffset2D offset{
+            .x = 0,
+            .y = 0,
+        };
+        const VkViewport viewport{
+            .x = 0.0f,
+            .y = 0.0f,
+            .width = static_cast<float>(extent.width),
+            .height = static_cast<float>(extent.height),
+            .minDepth = 0.0f,
+            .maxDepth = 0.0f,
+        };
+        const VkRect2D scissor{
+            .offset = offset,
+            .extent = extent,
+        };
+        const PushConstants push_constants{
+            .tex_scale = {viewport.width, viewport.height},
+            .tex_offset = {0.0f, 0.0f},
+        };
+        const VkDescriptorSet descriptor_set = two_textures_descriptor_allocator.Commit();
+        UpdateTwoTexturesDescriptorSet(device, descriptor_set, sampler, src_depth_view,
+                                       src_stencil_view);
+        // TODO: Barriers
+        cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
+        cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set,
+                                  nullptr);
+        cmdbuf.SetViewport(0, viewport);
+        cmdbuf.SetScissor(0, scissor);
+        cmdbuf.PushConstants(layout, VK_SHADER_STAGE_VERTEX_BIT, push_constants);
+        cmdbuf.Draw(3, 1, 0, 0);
+    });
+    scheduler.InvalidateState();
+}
+
 VkPipeline BlitImageHelper::FindOrEmplaceColorPipeline(const BlitImagePipelineKey& key) {
     const auto it = std::ranges::find(blit_color_keys, key);
     if (it != blit_color_keys.end()) {
@@ -636,4 +751,44 @@ void BlitImageHelper::ConvertColorToDepthPipeline(vk::Pipeline& pipeline, VkRend
     });
 }
 
+void BlitImageHelper::ConvertPipelineEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
+                                        vk::ShaderModule& module, bool is_target_depth,
+                                        bool single_texture) {
+    if (pipeline) {
+        return;
+    }
+    const std::array stages = MakeStages(*full_screen_vert, *module);
+    pipeline = device.GetLogical().CreateGraphicsPipeline({
+        .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
+        .pNext = nullptr,
+        .flags = 0,
+        .stageCount = static_cast<u32>(stages.size()),
+        .pStages = stages.data(),
+        .pVertexInputState = &PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
+        .pInputAssemblyState = &PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
+        .pTessellationState = nullptr,
+        .pViewportState = &PIPELINE_VIEWPORT_STATE_CREATE_INFO,
+        .pRasterizationState = &PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
+        .pMultisampleState = &PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
+        .pDepthStencilState = is_target_depth ? &PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO : nullptr,
+        .pColorBlendState = &PIPELINE_COLOR_BLEND_STATE_GENERIC_CREATE_INFO,
+        .pDynamicState = &PIPELINE_DYNAMIC_STATE_CREATE_INFO,
+        .layout = single_texture ? *one_texture_pipeline_layout : *two_textures_pipeline_layout,
+        .renderPass = renderpass,
+        .subpass = 0,
+        .basePipelineHandle = VK_NULL_HANDLE,
+        .basePipelineIndex = 0,
+    });
+}
+
+void BlitImageHelper::ConvertPipelineColorTargetEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
+                                                   vk::ShaderModule& module, bool single_texture) {
+    ConvertPipelineEx(pipeline, renderpass, module, false, single_texture);
+}
+
+void BlitImageHelper::ConvertPipelineDepthTargetEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
+                                                   vk::ShaderModule& module, bool single_texture) {
+    ConvertPipelineEx(pipeline, renderpass, module, true, single_texture);
+}
+
 } // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/blit_image.h b/src/video_core/renderer_vulkan/blit_image.h
index d77f76678..b1a717090 100644
--- a/src/video_core/renderer_vulkan/blit_image.h
+++ b/src/video_core/renderer_vulkan/blit_image.h
@@ -56,10 +56,22 @@ public:
     void ConvertR16ToD16(const Framebuffer* dst_framebuffer, const ImageView& src_image_view,
                          u32 up_scale, u32 down_shift);
 
+    void ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer, ImageView& src_image_view,
+                             u32 up_scale, u32 down_shift);
+
+    void ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view,
+                             u32 up_scale, u32 down_shift);
+
 private:
     void Convert(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
                  const ImageView& src_image_view, u32 up_scale, u32 down_shift);
 
+    void ConvertColor(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
+                      ImageView& src_image_view, u32 up_scale, u32 down_shift);
+
+    void ConvertDepthStencil(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
+                             ImageView& src_image_view, u32 up_scale, u32 down_shift);
+
     [[nodiscard]] VkPipeline FindOrEmplaceColorPipeline(const BlitImagePipelineKey& key);
 
     [[nodiscard]] VkPipeline FindOrEmplaceDepthStencilPipeline(const BlitImagePipelineKey& key);
@@ -68,6 +80,15 @@ private:
 
     void ConvertColorToDepthPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass);
 
+    void ConvertPipelineEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
+                           vk::ShaderModule& module, bool is_target_depth, bool single_texture);
+
+    void ConvertPipelineColorTargetEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
+                                      vk::ShaderModule& module, bool single_texture);
+
+    void ConvertPipelineDepthTargetEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
+                                      vk::ShaderModule& module, bool single_texture);
+
     const Device& device;
     VKScheduler& scheduler;
     StateTracker& state_tracker;
@@ -83,6 +104,8 @@ private:
     vk::ShaderModule blit_depth_stencil_frag;
     vk::ShaderModule convert_depth_to_float_frag;
     vk::ShaderModule convert_float_to_depth_frag;
+    vk::ShaderModule convert_abgr8_to_d24s8_frag;
+    vk::ShaderModule convert_d24s8_to_abgr8_frag;
     vk::Sampler linear_sampler;
     vk::Sampler nearest_sampler;
 
@@ -94,6 +117,8 @@ private:
     vk::Pipeline convert_r32_to_d32_pipeline;
     vk::Pipeline convert_d16_to_r16_pipeline;
     vk::Pipeline convert_r16_to_d16_pipeline;
+    vk::Pipeline convert_abgr8_to_d24s8_pipeline;
+    vk::Pipeline convert_d24s8_to_abgr8_pipeline;
 };
 
 } // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
index 68a23b602..751e4792b 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
@@ -162,7 +162,7 @@ struct FormatTuple {
     {VK_FORMAT_UNDEFINED},                                     // R16_SINT
     {VK_FORMAT_R16G16_UNORM, Attachable | Storage},            // R16G16_UNORM
     {VK_FORMAT_R16G16_SFLOAT, Attachable | Storage},           // R16G16_FLOAT
-    {VK_FORMAT_UNDEFINED},                                     // R16G16_UINT
+    {VK_FORMAT_R16G16_UINT, Attachable | Storage},             // R16G16_UINT
     {VK_FORMAT_R16G16_SINT, Attachable | Storage},             // R16G16_SINT
     {VK_FORMAT_R16G16_SNORM, Attachable | Storage},            // R16G16_SNORM
     {VK_FORMAT_UNDEFINED},                                     // R32G32B32_FLOAT
@@ -176,8 +176,8 @@ struct FormatTuple {
     {VK_FORMAT_R32_UINT, Attachable | Storage},                // R32_UINT
     {VK_FORMAT_R32_SINT, Attachable | Storage},                // R32_SINT
     {VK_FORMAT_ASTC_8x8_UNORM_BLOCK},                          // ASTC_2D_8X8_UNORM
-    {VK_FORMAT_UNDEFINED},                                     // ASTC_2D_8X5_UNORM
-    {VK_FORMAT_UNDEFINED},                                     // ASTC_2D_5X4_UNORM
+    {VK_FORMAT_ASTC_8x5_UNORM_BLOCK},                          // ASTC_2D_8X5_UNORM
+    {VK_FORMAT_ASTC_5x4_UNORM_BLOCK},                          // ASTC_2D_5X4_UNORM
     {VK_FORMAT_B8G8R8A8_SRGB, Attachable},                     // B8G8R8A8_SRGB
     {VK_FORMAT_BC1_RGBA_SRGB_BLOCK},                           // BC1_RGBA_SRGB
     {VK_FORMAT_BC2_SRGB_BLOCK},                                // BC2_SRGB
@@ -208,6 +208,9 @@ struct FormatTuple {
     {VK_FORMAT_D32_SFLOAT, Attachable}, // D32_FLOAT
     {VK_FORMAT_D16_UNORM, Attachable},  // D16_UNORM
 
+    // Stencil formats
+    {VK_FORMAT_S8_UINT, Attachable}, // S8_UINT
+
     // DepthStencil formats
     {VK_FORMAT_D24_UNORM_S8_UINT, Attachable},  // D24_UNORM_S8_UINT
     {VK_FORMAT_D24_UNORM_S8_UINT, Attachable},  // S8_UINT_D24_UNORM (emulated)
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
index 407fd2a15..197cba8e3 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
@@ -8,6 +8,7 @@
 #include <vector>
 
 #include "common/bit_cast.h"
+#include "common/bit_util.h"
 #include "common/settings.h"
 
 #include "video_core/engines/fermi_2d.h"
@@ -102,6 +103,7 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
             usage |= VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT;
             break;
         case VideoCore::Surface::SurfaceType::Depth:
+        case VideoCore::Surface::SurfaceType::Stencil:
         case VideoCore::Surface::SurfaceType::DepthStencil:
             usage |= VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT;
             break;
@@ -173,6 +175,8 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
         return VK_IMAGE_ASPECT_COLOR_BIT;
     case VideoCore::Surface::SurfaceType::Depth:
         return VK_IMAGE_ASPECT_DEPTH_BIT;
+    case VideoCore::Surface::SurfaceType::Stencil:
+        return VK_IMAGE_ASPECT_STENCIL_BIT;
     case VideoCore::Surface::SurfaceType::DepthStencil:
         return VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT;
     default:
@@ -195,6 +199,8 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
     case PixelFormat::D16_UNORM:
     case PixelFormat::D32_FLOAT:
         return VK_IMAGE_ASPECT_DEPTH_BIT;
+    case PixelFormat::S8_UINT:
+        return VK_IMAGE_ASPECT_STENCIL_BIT;
     default:
         return VK_IMAGE_ASPECT_COLOR_BIT;
     }
@@ -308,6 +314,19 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
     };
 }
 
+[[nodiscard]] VkBufferImageCopy MakeBufferImageCopy(const VideoCommon::ImageCopy& copy, bool is_src,
+                                                    VkImageAspectFlags aspect_mask) noexcept {
+    return VkBufferImageCopy{
+        .bufferOffset = 0,
+        .bufferRowLength = 0,
+        .bufferImageHeight = 0,
+        .imageSubresource = MakeImageSubresourceLayers(
+            is_src ? copy.src_subresource : copy.dst_subresource, aspect_mask),
+        .imageOffset = MakeOffset3D(is_src ? copy.src_offset : copy.dst_offset),
+        .imageExtent = MakeExtent3D(copy.extent),
+    };
+}
+
 [[maybe_unused]] [[nodiscard]] std::vector<VkBufferCopy> TransformBufferCopies(
     std::span<const VideoCommon::BufferCopy> copies, size_t buffer_offset) {
     std::vector<VkBufferCopy> result(copies.size());
@@ -754,6 +773,173 @@ StagingBufferRef TextureCacheRuntime::DownloadStagingBuffer(size_t size) {
     return staging_buffer_pool.Request(size, MemoryUsage::Download);
 }
 
+bool TextureCacheRuntime::ShouldReinterpret(Image& dst, Image& src) {
+    if (VideoCore::Surface::GetFormatType(dst.info.format) ==
+            VideoCore::Surface::SurfaceType::DepthStencil &&
+        !device.IsExtShaderStencilExportSupported()) {
+        return true;
+    }
+    if (VideoCore::Surface::GetFormatType(src.info.format) ==
+            VideoCore::Surface::SurfaceType::DepthStencil &&
+        !device.IsExtShaderStencilExportSupported()) {
+        return true;
+    }
+    if (dst.info.format == PixelFormat::D32_FLOAT_S8_UINT ||
+        src.info.format == PixelFormat::D32_FLOAT_S8_UINT) {
+        return true;
+    }
+    return false;
+}
+
+VkBuffer TextureCacheRuntime::GetTemporaryBuffer(size_t needed_size) {
+    const auto level = (8 * sizeof(size_t)) - std::countl_zero(needed_size - 1ULL);
+    if (buffer_commits[level]) {
+        return *buffers[level];
+    }
+    const auto new_size = Common::NextPow2(needed_size);
+    static constexpr VkBufferUsageFlags flags =
+        VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT |
+        VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_TEXEL_BUFFER_BIT;
+    buffers[level] = device.GetLogical().CreateBuffer({
+        .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
+        .pNext = nullptr,
+        .flags = 0,
+        .size = new_size,
+        .usage = flags,
+        .sharingMode = VK_SHARING_MODE_EXCLUSIVE,
+        .queueFamilyIndexCount = 0,
+        .pQueueFamilyIndices = nullptr,
+    });
+    buffer_commits[level] = std::make_unique<MemoryCommit>(
+        memory_allocator.Commit(buffers[level], MemoryUsage::DeviceLocal));
+    return *buffers[level];
+}
+
+void TextureCacheRuntime::ReinterpretImage(Image& dst, Image& src,
+                                           std::span<const VideoCommon::ImageCopy> copies) {
+    std::vector<VkBufferImageCopy> vk_in_copies(copies.size());
+    std::vector<VkBufferImageCopy> vk_out_copies(copies.size());
+    const VkImageAspectFlags src_aspect_mask = src.AspectMask();
+    const VkImageAspectFlags dst_aspect_mask = dst.AspectMask();
+
+    std::ranges::transform(copies, vk_in_copies.begin(), [src_aspect_mask](const auto& copy) {
+        return MakeBufferImageCopy(copy, true, src_aspect_mask);
+    });
+    std::ranges::transform(copies, vk_out_copies.begin(), [dst_aspect_mask](const auto& copy) {
+        return MakeBufferImageCopy(copy, false, dst_aspect_mask);
+    });
+    const u32 img_bpp = BytesPerBlock(src.info.format);
+    size_t total_size = 0;
+    for (const auto& copy : copies) {
+        total_size += copy.extent.width * copy.extent.height * copy.extent.depth * img_bpp;
+    }
+    const VkBuffer copy_buffer = GetTemporaryBuffer(total_size);
+    const VkImage dst_image = dst.Handle();
+    const VkImage src_image = src.Handle();
+    scheduler.RequestOutsideRenderPassOperationContext();
+    scheduler.Record([dst_image, src_image, copy_buffer, src_aspect_mask, dst_aspect_mask,
+                      vk_in_copies, vk_out_copies](vk::CommandBuffer cmdbuf) {
+        RangedBarrierRange dst_range;
+        RangedBarrierRange src_range;
+        for (const VkBufferImageCopy& copy : vk_in_copies) {
+            src_range.AddLayers(copy.imageSubresource);
+        }
+        for (const VkBufferImageCopy& copy : vk_out_copies) {
+            dst_range.AddLayers(copy.imageSubresource);
+        }
+        static constexpr VkMemoryBarrier READ_BARRIER{
+            .sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = VK_ACCESS_MEMORY_WRITE_BIT,
+            .dstAccessMask = VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT,
+        };
+        static constexpr VkMemoryBarrier WRITE_BARRIER{
+            .sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+            .dstAccessMask = VK_ACCESS_MEMORY_READ_BIT | VK_ACCESS_MEMORY_WRITE_BIT,
+        };
+        const std::array pre_barriers{
+            VkImageMemoryBarrier{
+                .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+                .pNext = nullptr,
+                .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT | VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+                                 VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+                                 VK_ACCESS_TRANSFER_WRITE_BIT,
+                .dstAccessMask = VK_ACCESS_TRANSFER_READ_BIT,
+                .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+                .newLayout = VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL,
+                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .image = src_image,
+                .subresourceRange = src_range.SubresourceRange(src_aspect_mask),
+            },
+        };
+        const std::array middle_in_barrier{
+            VkImageMemoryBarrier{
+                .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+                .pNext = nullptr,
+                .srcAccessMask = 0,
+                .dstAccessMask = 0,
+                .oldLayout = VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL,
+                .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .image = src_image,
+                .subresourceRange = src_range.SubresourceRange(src_aspect_mask),
+            },
+        };
+        const std::array middle_out_barrier{
+            VkImageMemoryBarrier{
+                .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+                .pNext = nullptr,
+                .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT | VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+                                 VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+                                 VK_ACCESS_TRANSFER_WRITE_BIT,
+                .dstAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+                .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+                .newLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .image = dst_image,
+                .subresourceRange = dst_range.SubresourceRange(dst_aspect_mask),
+            },
+        };
+        const std::array post_barriers{
+            VkImageMemoryBarrier{
+                .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+                .pNext = nullptr,
+                .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
+                .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT |
+                                 VK_ACCESS_COLOR_ATTACHMENT_READ_BIT |
+                                 VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT |
+                                 VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_READ_BIT |
+                                 VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT |
+                                 VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT,
+                .oldLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+                .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .image = dst_image,
+                .subresourceRange = dst_range.SubresourceRange(dst_aspect_mask),
+            },
+        };
+        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT,
+                               0, {}, {}, pre_barriers);
+
+        cmdbuf.CopyImageToBuffer(src_image, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, copy_buffer,
+                                 vk_in_copies);
+        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_PIPELINE_STAGE_ALL_COMMANDS_BIT,
+                               0, WRITE_BARRIER, nullptr, middle_in_barrier);
+
+        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT,
+                               0, READ_BARRIER, {}, middle_out_barrier);
+        cmdbuf.CopyBufferToImage(copy_buffer, dst_image, VK_IMAGE_LAYOUT_GENERAL, vk_out_copies);
+        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_PIPELINE_STAGE_ALL_COMMANDS_BIT,
+                               0, {}, {}, post_barriers);
+    });
+}
+
 void TextureCacheRuntime::BlitImage(Framebuffer* dst_framebuffer, ImageView& dst, ImageView& src,
                                     const Region2D& dst_region, const Region2D& src_region,
                                     Tegra::Engines::Fermi2D::Filter filter,
@@ -881,6 +1067,11 @@ void TextureCacheRuntime::ConvertImage(Framebuffer* dst, ImageView& dst_view, Im
             return blit_image_helper.ConvertD16ToR16(dst, src_view, up_scale, down_shift);
         }
         break;
+    case PixelFormat::A8B8G8R8_UNORM:
+        if (src_view.format == PixelFormat::S8_UINT_D24_UNORM) {
+            return blit_image_helper.ConvertD24S8ToABGR8(dst, src_view, up_scale, down_shift);
+        }
+        break;
     case PixelFormat::R32_FLOAT:
         if (src_view.format == PixelFormat::D32_FLOAT) {
             return blit_image_helper.ConvertD32ToR32(dst, src_view, up_scale, down_shift);
@@ -891,6 +1082,9 @@ void TextureCacheRuntime::ConvertImage(Framebuffer* dst, ImageView& dst_view, Im
             return blit_image_helper.ConvertR16ToD16(dst, src_view, up_scale, down_shift);
         }
         break;
+    case PixelFormat::S8_UINT_D24_UNORM:
+        return blit_image_helper.ConvertABGR8ToD24S8(dst, src_view, up_scale, down_shift);
+        break;
     case PixelFormat::D32_FLOAT:
         if (src_view.format == PixelFormat::R32_FLOAT) {
             return blit_image_helper.ConvertR32ToD32(dst, src_view, up_scale, down_shift);
@@ -1386,6 +1580,14 @@ VkImageView ImageView::StencilView() {
     return *stencil_view;
 }
 
+VkImageView ImageView::ColorView() {
+    if (color_view) {
+        return *color_view;
+    }
+    color_view = MakeView(VK_FORMAT_R8G8B8A8_UNORM, VK_IMAGE_ASPECT_COLOR_BIT);
+    return *color_view;
+}
+
 VkImageView ImageView::StorageView(Shader::TextureType texture_type,
                                    Shader::ImageFormat image_format) {
     if (image_format == Shader::ImageFormat::Typeless) {
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h
index f5f8f9a74..753e3e8a1 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.h
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.h
@@ -61,6 +61,10 @@ public:
 
     void CopyImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
 
+    bool ShouldReinterpret(Image& dst, Image& src);
+
+    void ReinterpretImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
+
     void ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view, bool rescaled);
 
     bool CanAccelerateImageUpload(Image&) const noexcept {
@@ -82,6 +86,8 @@ public:
         return true;
     }
 
+    [[nodiscard]] VkBuffer GetTemporaryBuffer(size_t needed_size);
+
     const Device& device;
     VKScheduler& scheduler;
     MemoryAllocator& memory_allocator;
@@ -90,6 +96,10 @@ public:
     ASTCDecoderPass& astc_decoder_pass;
     RenderPassCache& render_pass_cache;
     const Settings::ResolutionScalingInfo& resolution;
+
+    constexpr static size_t indexing_slots = 8 * sizeof(size_t);
+    std::array<vk::Buffer, indexing_slots> buffers{};
+    std::array<std::unique_ptr<MemoryCommit>, indexing_slots> buffer_commits{};
 };
 
 class Image : public VideoCommon::ImageBase {
@@ -174,6 +184,8 @@ public:
 
     [[nodiscard]] VkImageView StencilView();
 
+    [[nodiscard]] VkImageView ColorView();
+
     [[nodiscard]] VkImageView StorageView(Shader::TextureType texture_type,
                                           Shader::ImageFormat image_format);
 
@@ -214,6 +226,7 @@ private:
     std::unique_ptr<StorageViews> storage_views;
     vk::ImageView depth_view;
     vk::ImageView stencil_view;
+    vk::ImageView color_view;
     VkImage image_handle = VK_NULL_HANDLE;
     VkImageView render_target = VK_NULL_HANDLE;
     VkSampleCountFlagBits samples = VK_SAMPLE_COUNT_1_BIT;
@@ -316,7 +329,6 @@ struct TextureCacheParams {
     static constexpr bool FRAMEBUFFER_BLITS = false;
     static constexpr bool HAS_EMULATED_COPIES = false;
     static constexpr bool HAS_DEVICE_MEMORY_INFO = true;
-    static constexpr bool HAS_PIXEL_FORMAT_CONVERSIONS = false;
 
     using Runtime = Vulkan::TextureCacheRuntime;
     using Image = Vulkan::Image;
diff --git a/src/video_core/shader_notify.cpp b/src/video_core/shader_notify.cpp
index dc6995b46..bcaf5f575 100644
--- a/src/video_core/shader_notify.cpp
+++ b/src/video_core/shader_notify.cpp
@@ -18,7 +18,7 @@ int ShaderNotify::ShadersBuilding() noexcept {
     const int now_complete = num_complete.load(std::memory_order::relaxed);
     const int now_building = num_building.load(std::memory_order::relaxed);
     if (now_complete == now_building) {
-        const auto now = std::chrono::high_resolution_clock::now();
+        const auto now = std::chrono::steady_clock::now();
         if (completed && num_complete == num_when_completed) {
             if (now - complete_time > TIME_TO_STOP_REPORTING) {
                 report_base = now_complete;
diff --git a/src/video_core/shader_notify.h b/src/video_core/shader_notify.h
index ad363bfb5..4d8d52071 100644
--- a/src/video_core/shader_notify.h
+++ b/src/video_core/shader_notify.h
@@ -28,6 +28,6 @@ private:
 
     bool completed{};
     int num_when_completed{};
-    std::chrono::high_resolution_clock::time_point complete_time;
+    std::chrono::steady_clock::time_point complete_time;
 };
 } // namespace VideoCore
diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp
index 58d262446..a36015c8c 100644
--- a/src/video_core/surface.cpp
+++ b/src/video_core/surface.cpp
@@ -82,6 +82,8 @@ PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
         return PixelFormat::D32_FLOAT;
     case Tegra::DepthFormat::D16_UNORM:
         return PixelFormat::D16_UNORM;
+    case Tegra::DepthFormat::S8_UINT:
+        return PixelFormat::S8_UINT;
     case Tegra::DepthFormat::D32_FLOAT_S8X24_UINT:
         return PixelFormat::D32_FLOAT_S8_UINT;
     default:
@@ -214,6 +216,11 @@ SurfaceType GetFormatType(PixelFormat pixel_format) {
     }
 
     if (static_cast<std::size_t>(pixel_format) <
+        static_cast<std::size_t>(PixelFormat::MaxStencilFormat)) {
+        return SurfaceType::Stencil;
+    }
+
+    if (static_cast<std::size_t>(pixel_format) <
         static_cast<std::size_t>(PixelFormat::MaxDepthStencilFormat)) {
         return SurfaceType::DepthStencil;
     }
diff --git a/src/video_core/surface.h b/src/video_core/surface.h
index 2ce7c7d33..33e8d24ab 100644
--- a/src/video_core/surface.h
+++ b/src/video_core/surface.h
@@ -110,8 +110,12 @@ enum class PixelFormat {
 
     MaxDepthFormat,
 
+    // Stencil formats
+    S8_UINT = MaxDepthFormat,
+    MaxStencilFormat,
+
     // DepthStencil formats
-    D24_UNORM_S8_UINT = MaxDepthFormat,
+    D24_UNORM_S8_UINT = MaxStencilFormat,
     S8_UINT_D24_UNORM,
     D32_FLOAT_S8_UINT,
 
@@ -125,8 +129,9 @@ constexpr std::size_t MaxPixelFormat = static_cast<std::size_t>(PixelFormat::Max
 enum class SurfaceType {
     ColorTexture = 0,
     Depth = 1,
-    DepthStencil = 2,
-    Invalid = 3,
+    Stencil = 2,
+    DepthStencil = 3,
+    Invalid = 4,
 };
 
 enum class SurfaceTarget {
@@ -229,6 +234,7 @@ constexpr std::array<u32, MaxPixelFormat> BLOCK_WIDTH_TABLE = {{
     1,  // E5B9G9R9_FLOAT
     1,  // D32_FLOAT
     1,  // D16_UNORM
+    1,  // S8_UINT
     1,  // D24_UNORM_S8_UINT
     1,  // S8_UINT_D24_UNORM
     1,  // D32_FLOAT_S8_UINT
@@ -328,6 +334,7 @@ constexpr std::array<u32, MaxPixelFormat> BLOCK_HEIGHT_TABLE = {{
     1,  // E5B9G9R9_FLOAT
     1,  // D32_FLOAT
     1,  // D16_UNORM
+    1,  // S8_UINT
     1,  // D24_UNORM_S8_UINT
     1,  // S8_UINT_D24_UNORM
     1,  // D32_FLOAT_S8_UINT
@@ -427,6 +434,7 @@ constexpr std::array<u32, MaxPixelFormat> BITS_PER_BLOCK_TABLE = {{
     32,  // E5B9G9R9_FLOAT
     32,  // D32_FLOAT
     16,  // D16_UNORM
+    8,   // S8_UINT
     32,  // D24_UNORM_S8_UINT
     32,  // S8_UINT_D24_UNORM
     64,  // D32_FLOAT_S8_UINT
diff --git a/src/video_core/texture_cache/format_lookup_table.cpp b/src/video_core/texture_cache/format_lookup_table.cpp
index ddfb726fe..afa807d5d 100644
--- a/src/video_core/texture_cache/format_lookup_table.cpp
+++ b/src/video_core/texture_cache/format_lookup_table.cpp
@@ -139,6 +139,8 @@ PixelFormat PixelFormatFromTextureInfo(TextureFormat format, ComponentType red,
         return PixelFormat::D16_UNORM;
     case Hash(TextureFormat::S8D24, UINT, UNORM, UNORM, UNORM, LINEAR):
         return PixelFormat::S8_UINT_D24_UNORM;
+    case Hash(TextureFormat::S8D24, UINT, UNORM, UINT, UINT, LINEAR):
+        return PixelFormat::S8_UINT_D24_UNORM;
     case Hash(TextureFormat::R8G24, UINT, UNORM, UNORM, UNORM, LINEAR):
         return PixelFormat::S8_UINT_D24_UNORM;
     case Hash(TextureFormat::D32S8, FLOAT, UINT, UNORM, UNORM, LINEAR):
diff --git a/src/video_core/texture_cache/formatter.h b/src/video_core/texture_cache/formatter.h
index c6cf0583f..b2c81057b 100644
--- a/src/video_core/texture_cache/formatter.h
+++ b/src/video_core/texture_cache/formatter.h
@@ -194,6 +194,8 @@ struct fmt::formatter<VideoCore::Surface::PixelFormat> : fmt::formatter<fmt::str
                 return "D32_FLOAT";
             case PixelFormat::D16_UNORM:
                 return "D16_UNORM";
+            case PixelFormat::S8_UINT:
+                return "S8_UINT";
             case PixelFormat::D24_UNORM_S8_UINT:
                 return "D24_UNORM_S8_UINT";
             case PixelFormat::S8_UINT_D24_UNORM:
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index 241f71a91..5aaeb16ca 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -472,9 +472,10 @@ template <class P>
 void TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
                                 const Tegra::Engines::Fermi2D::Surface& src,
                                 const Tegra::Engines::Fermi2D::Config& copy) {
-    const BlitImages images = GetBlitImages(dst, src);
+    const BlitImages images = GetBlitImages(dst, src, copy);
     const ImageId dst_id = images.dst_id;
     const ImageId src_id = images.src_id;
+
     PrepareImage(src_id, false, false);
     PrepareImage(dst_id, true, false);
 
@@ -758,14 +759,18 @@ ImageId TextureCache<P>::FindImage(const ImageInfo& info, GPUVAddr gpu_addr,
             return ImageId{};
         }
     }
-    const bool broken_views = runtime.HasBrokenTextureViewFormats();
+    const bool broken_views =
+        runtime.HasBrokenTextureViewFormats() || True(options & RelaxedOptions::ForceBrokenViews);
     const bool native_bgr = runtime.HasNativeBgr();
-    ImageId image_id;
+    const bool flexible_formats = True(options & RelaxedOptions::Format);
+    ImageId image_id{};
+    boost::container::small_vector<ImageId, 1> image_ids;
     const auto lambda = [&](ImageId existing_image_id, ImageBase& existing_image) {
         if (True(existing_image.flags & ImageFlagBits::Remapped)) {
             return false;
         }
-        if (info.type == ImageType::Linear || existing_image.info.type == ImageType::Linear) {
+        if (info.type == ImageType::Linear || existing_image.info.type == ImageType::Linear)
+            [[unlikely]] {
             const bool strict_size = False(options & RelaxedOptions::Size) &&
                                      True(existing_image.flags & ImageFlagBits::Strong);
             const ImageInfo& existing = existing_image.info;
@@ -774,17 +779,27 @@ ImageId TextureCache<P>::FindImage(const ImageInfo& info, GPUVAddr gpu_addr,
                 IsPitchLinearSameSize(existing, info, strict_size) &&
                 IsViewCompatible(existing.format, info.format, broken_views, native_bgr)) {
                 image_id = existing_image_id;
-                return true;
+                image_ids.push_back(existing_image_id);
+                return !flexible_formats && existing.format == info.format;
             }
         } else if (IsSubresource(info, existing_image, gpu_addr, options, broken_views,
                                  native_bgr)) {
             image_id = existing_image_id;
-            return true;
+            image_ids.push_back(existing_image_id);
+            return !flexible_formats && existing_image.info.format == info.format;
         }
         return false;
     };
     ForEachImageInRegion(*cpu_addr, CalculateGuestSizeInBytes(info), lambda);
-    return image_id;
+    if (image_ids.size() <= 1) [[likely]] {
+        return image_id;
+    }
+    auto image_ids_compare = [this](ImageId a, ImageId b) {
+        auto& image_a = slot_images[a];
+        auto& image_b = slot_images[b];
+        return image_a.modification_tick < image_b.modification_tick;
+    };
+    return *std::ranges::max_element(image_ids, image_ids_compare);
 }
 
 template <class P>
@@ -1076,31 +1091,66 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA
 
 template <class P>
 typename TextureCache<P>::BlitImages TextureCache<P>::GetBlitImages(
-    const Tegra::Engines::Fermi2D::Surface& dst, const Tegra::Engines::Fermi2D::Surface& src) {
-    static constexpr auto FIND_OPTIONS = RelaxedOptions::Format | RelaxedOptions::Samples;
+    const Tegra::Engines::Fermi2D::Surface& dst, const Tegra::Engines::Fermi2D::Surface& src,
+    const Tegra::Engines::Fermi2D::Config& copy) {
+
+    static constexpr auto FIND_OPTIONS = RelaxedOptions::Samples;
     const GPUVAddr dst_addr = dst.Address();
     const GPUVAddr src_addr = src.Address();
     ImageInfo dst_info(dst);
     ImageInfo src_info(src);
+    const bool can_be_depth_blit =
+        dst_info.format == src_info.format && copy.filter == Tegra::Engines::Fermi2D::Filter::Point;
     ImageId dst_id;
     ImageId src_id;
+    RelaxedOptions try_options = FIND_OPTIONS;
+    if (can_be_depth_blit) {
+        try_options |= RelaxedOptions::Format;
+    }
     do {
         has_deleted_images = false;
-        dst_id = FindImage(dst_info, dst_addr, FIND_OPTIONS);
-        src_id = FindImage(src_info, src_addr, FIND_OPTIONS);
-        const ImageBase* const dst_image = dst_id ? &slot_images[dst_id] : nullptr;
+        src_id = FindImage(src_info, src_addr, try_options);
+        dst_id = FindImage(dst_info, dst_addr, try_options);
         const ImageBase* const src_image = src_id ? &slot_images[src_id] : nullptr;
-        DeduceBlitImages(dst_info, src_info, dst_image, src_image);
-        if (GetFormatType(dst_info.format) != GetFormatType(src_info.format)) {
-            continue;
+        if (src_image && src_image->info.num_samples > 1) {
+            RelaxedOptions find_options{FIND_OPTIONS | RelaxedOptions::ForceBrokenViews};
+            src_id = FindOrInsertImage(src_info, src_addr, find_options);
+            dst_id = FindOrInsertImage(dst_info, dst_addr, find_options);
+            if (has_deleted_images) {
+                continue;
+            }
+            break;
         }
-        if (!dst_id) {
-            dst_id = InsertImage(dst_info, dst_addr, RelaxedOptions{});
+        if (can_be_depth_blit) {
+            const ImageBase* const dst_image = dst_id ? &slot_images[dst_id] : nullptr;
+            DeduceBlitImages(dst_info, src_info, dst_image, src_image);
+            if (GetFormatType(dst_info.format) != GetFormatType(src_info.format)) {
+                continue;
+            }
         }
         if (!src_id) {
             src_id = InsertImage(src_info, src_addr, RelaxedOptions{});
         }
+        if (!dst_id) {
+            dst_id = InsertImage(dst_info, dst_addr, RelaxedOptions{});
+        }
     } while (has_deleted_images);
+    const ImageBase& src_image = slot_images[src_id];
+    const ImageBase& dst_image = slot_images[dst_id];
+    const bool native_bgr = runtime.HasNativeBgr();
+    if (GetFormatType(dst_info.format) != GetFormatType(dst_image.info.format) ||
+        GetFormatType(src_info.format) != GetFormatType(src_image.info.format) ||
+        !VideoCore::Surface::IsViewCompatible(dst_info.format, dst_image.info.format, false,
+                                              native_bgr) ||
+        !VideoCore::Surface::IsViewCompatible(src_info.format, src_image.info.format, false,
+                                              native_bgr)) {
+        // Make sure the images match the expected format.
+        do {
+            has_deleted_images = false;
+            src_id = FindOrInsertImage(src_info, src_addr, RelaxedOptions{});
+            dst_id = FindOrInsertImage(dst_info, dst_addr, RelaxedOptions{});
+        } while (has_deleted_images);
+    }
     return BlitImages{
         .dst_id = dst_id,
         .src_id = src_id,
@@ -1157,7 +1207,14 @@ template <class P>
 ImageViewId TextureCache<P>::FindRenderTargetView(const ImageInfo& info, GPUVAddr gpu_addr,
                                                   bool is_clear) {
     const auto options = is_clear ? RelaxedOptions::Samples : RelaxedOptions{};
-    const ImageId image_id = FindOrInsertImage(info, gpu_addr, options);
+    ImageId image_id{};
+    bool delete_state = has_deleted_images;
+    do {
+        has_deleted_images = false;
+        image_id = FindOrInsertImage(info, gpu_addr, options);
+        delete_state |= has_deleted_images;
+    } while (has_deleted_images);
+    has_deleted_images = delete_state;
     if (!image_id) {
         return NULL_IMAGE_VIEW_ID;
     }
@@ -1759,8 +1816,8 @@ void TextureCache<P>::CopyImage(ImageId dst_id, ImageId src_id, std::vector<Imag
     }
     UNIMPLEMENTED_IF(dst.info.type != ImageType::e2D);
     UNIMPLEMENTED_IF(src.info.type != ImageType::e2D);
-    if constexpr (HAS_PIXEL_FORMAT_CONVERSIONS) {
-        return runtime.ConvertImage(dst, src, copies);
+    if (runtime.ShouldReinterpret(dst, src)) {
+        return runtime.ReinterpretImage(dst, src, copies);
     }
     for (const ImageCopy& copy : copies) {
         UNIMPLEMENTED_IF(copy.dst_subresource.num_layers != 1);
@@ -1780,7 +1837,13 @@ void TextureCache<P>::CopyImage(ImageId dst_id, ImageId src_id, std::vector<Imag
         const SubresourceExtent src_extent{.levels = 1, .layers = 1};
         const SubresourceRange dst_range{.base = dst_base, .extent = dst_extent};
         const SubresourceRange src_range{.base = src_base, .extent = src_extent};
-        const ImageViewInfo dst_view_info(ImageViewType::e2D, dst.info.format, dst_range);
+        PixelFormat dst_format = dst.info.format;
+        if (GetFormatType(src.info.format) == SurfaceType::DepthStencil &&
+            GetFormatType(dst_format) == SurfaceType::ColorTexture &&
+            BytesPerBlock(dst_format) == 4) {
+            dst_format = PixelFormat::A8B8G8R8_UNORM;
+        }
+        const ImageViewInfo dst_view_info(ImageViewType::e2D, dst_format, dst_range);
         const ImageViewInfo src_view_info(ImageViewType::e2D, src.info.format, src_range);
         const auto [dst_framebuffer_id, dst_view_id] = RenderTargetFromImage(dst_id, dst_view_info);
         Framebuffer* const dst_framebuffer = &slot_framebuffers[dst_framebuffer_id];
diff --git a/src/video_core/texture_cache/texture_cache_base.h b/src/video_core/texture_cache/texture_cache_base.h
index a9504c0e8..7107887a6 100644
--- a/src/video_core/texture_cache/texture_cache_base.h
+++ b/src/video_core/texture_cache/texture_cache_base.h
@@ -59,8 +59,6 @@ class TextureCache {
     static constexpr bool HAS_EMULATED_COPIES = P::HAS_EMULATED_COPIES;
     /// True when the API can provide info about the memory of the device.
     static constexpr bool HAS_DEVICE_MEMORY_INFO = P::HAS_DEVICE_MEMORY_INFO;
-    /// True when the API provides utilities for pixel format conversions.
-    static constexpr bool HAS_PIXEL_FORMAT_CONVERSIONS = P::HAS_PIXEL_FORMAT_CONVERSIONS;
 
     static constexpr u64 DEFAULT_EXPECTED_MEMORY = 1_GiB;
     static constexpr u64 DEFAULT_CRITICAL_MEMORY = 2_GiB;
@@ -254,7 +252,8 @@ private:
 
     /// Return a blit image pair from the given guest blit parameters
     [[nodiscard]] BlitImages GetBlitImages(const Tegra::Engines::Fermi2D::Surface& dst,
-                                           const Tegra::Engines::Fermi2D::Surface& src);
+                                           const Tegra::Engines::Fermi2D::Surface& src,
+                                           const Tegra::Engines::Fermi2D::Config& copy);
 
     /// Find or create a sampler from a guest descriptor sampler
     [[nodiscard]] SamplerId FindSampler(const TSCEntry& config);
diff --git a/src/video_core/texture_cache/types.h b/src/video_core/texture_cache/types.h
index 5c274abdf..5ac27b3a7 100644
--- a/src/video_core/texture_cache/types.h
+++ b/src/video_core/texture_cache/types.h
@@ -54,6 +54,7 @@ enum class RelaxedOptions : u32 {
     Size = 1 << 0,
     Format = 1 << 1,
     Samples = 1 << 2,
+    ForceBrokenViews = 1 << 3,
 };
 DECLARE_ENUM_FLAG_OPERATORS(RelaxedOptions)
 
diff --git a/src/video_core/texture_cache/util.cpp b/src/video_core/texture_cache/util.cpp
index ddc9fb13a..7bd31b211 100644
--- a/src/video_core/texture_cache/util.cpp
+++ b/src/video_core/texture_cache/util.cpp
@@ -1151,6 +1151,7 @@ bool IsSubresource(const ImageInfo& candidate, const ImageBase& image, GPUVAddr
 
 void DeduceBlitImages(ImageInfo& dst_info, ImageInfo& src_info, const ImageBase* dst,
                       const ImageBase* src) {
+    const auto original_dst_format = dst_info.format;
     if (src && GetFormatType(src->info.format) != SurfaceType::ColorTexture) {
         src_info.format = src->info.format;
     }
@@ -1161,7 +1162,13 @@ void DeduceBlitImages(ImageInfo& dst_info, ImageInfo& src_info, const ImageBase*
         dst_info.format = src->info.format;
     }
     if (dst && GetFormatType(dst->info.format) != SurfaceType::ColorTexture) {
-        src_info.format = dst->info.format;
+        if (src) {
+            if (GetFormatType(src->info.format) == SurfaceType::ColorTexture) {
+                dst_info.format = original_dst_format;
+            }
+        } else {
+            src_info.format = dst->info.format;
+        }
     }
 }
 
diff --git a/src/video_core/video_core.cpp b/src/video_core/video_core.cpp
index e852c817e..329bf4def 100644
--- a/src/video_core/video_core.cpp
+++ b/src/video_core/video_core.cpp
@@ -55,10 +55,4 @@ std::unique_ptr<Tegra::GPU> CreateGPU(Core::Frontend::EmuWindow& emu_window, Cor
     }
 }
 
-float GetResolutionScaleFactor(const RendererBase& renderer) {
-    return Settings::values.resolution_info.active
-               ? Settings::values.resolution_info.up_factor
-               : renderer.GetRenderWindow().GetFramebufferLayout().GetScalingRatio();
-}
-
 } // namespace VideoCore
diff --git a/src/video_core/video_core.h b/src/video_core/video_core.h
index f86877e86..084df641f 100644
--- a/src/video_core/video_core.h
+++ b/src/video_core/video_core.h
@@ -25,6 +25,4 @@ class RendererBase;
 /// Creates an emulated GPU instance using the given system context.
 std::unique_ptr<Tegra::GPU> CreateGPU(Core::Frontend::EmuWindow& emu_window, Core::System& system);
 
-float GetResolutionScaleFactor(const RendererBase& renderer);
-
 } // namespace VideoCore
diff --git a/src/video_core/vulkan_common/vulkan_device.cpp b/src/video_core/vulkan_common/vulkan_device.cpp
index 95106f88f..7bf5b6578 100644
--- a/src/video_core/vulkan_common/vulkan_device.cpp
+++ b/src/video_core/vulkan_common/vulkan_device.cpp
@@ -21,6 +21,13 @@
 namespace Vulkan {
 namespace {
 namespace Alternatives {
+constexpr std::array STENCIL8_UINT{
+    VK_FORMAT_D16_UNORM_S8_UINT,
+    VK_FORMAT_D24_UNORM_S8_UINT,
+    VK_FORMAT_D32_SFLOAT_S8_UINT,
+    VK_FORMAT_UNDEFINED,
+};
+
 constexpr std::array DEPTH24_UNORM_STENCIL8_UINT{
     VK_FORMAT_D32_SFLOAT_S8_UINT,
     VK_FORMAT_D16_UNORM_S8_UINT,
@@ -74,6 +81,8 @@ void SetNext(void**& next, T& data) {
 
 constexpr const VkFormat* GetFormatAlternatives(VkFormat format) {
     switch (format) {
+    case VK_FORMAT_S8_UINT:
+        return Alternatives::STENCIL8_UINT.data();
     case VK_FORMAT_D24_UNORM_S8_UINT:
         return Alternatives::DEPTH24_UNORM_STENCIL8_UINT.data();
     case VK_FORMAT_D16_UNORM_S8_UINT:
@@ -121,6 +130,7 @@ std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties(vk::Physica
         VK_FORMAT_R16G16_UNORM,
         VK_FORMAT_R16G16_SNORM,
         VK_FORMAT_R16G16_SFLOAT,
+        VK_FORMAT_R16G16_UINT,
         VK_FORMAT_R16G16_SINT,
         VK_FORMAT_R16_UNORM,
         VK_FORMAT_R16_SNORM,
@@ -145,6 +155,7 @@ std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties(vk::Physica
         VK_FORMAT_R4G4B4A4_UNORM_PACK16,
         VK_FORMAT_D32_SFLOAT,
         VK_FORMAT_D16_UNORM,
+        VK_FORMAT_S8_UINT,
         VK_FORMAT_D16_UNORM_S8_UINT,
         VK_FORMAT_D24_UNORM_S8_UINT,
         VK_FORMAT_D32_SFLOAT_S8_UINT,