From 92d49d8a0c3e60894765bc2c332fdc936c87fc45 Mon Sep 17 00:00:00 2001 From: Duck912 <350528965+Duck912@users.noreply.github.com> Date: Sun, 1 Nov 2026 21:43:28 -0500 Subject: video_core: temporal upscale stand-in for Fortune's Weave's DLSS The one-frame blit left 3D edges aliased and shaking: the game jitters its camera every frame for DLSS to accumulate. Fire Emblem: Fortune's Weave's first DLSS pass (also rejected) has the motion vectors (cbuf0+0x290, R16G16) and the frame's jitter (+0x1b0/+0x1b4, in input pixels). Kepler compute keeps them, and the stand-in for the last network pass now draws temporal_upscale.frag: input samples placed where the jitter put them and weighted by distance, blended with the previous result reprojected along the motion vectors (Catmull-Rom) and clipped to the 3x3 neighbourhood in YCoCg. The history is copied from the result right after drawing it, so post-processing the game later applies to its output does not feed back. EDEN_NX2_TEMPORAL_UPSCALE=0 restores the blit. Assisted-by: Claude Opus 5.5 (debugging, implementation and drafting) Signed-off-by: Duck912 <350528965+Duck912@users.noreply.github.com> --- src/video_core/texture_cache/texture_cache.h | 67 +++++++++++++++++++++++ src/video_core/texture_cache/texture_cache_base.h | 7 +++ 2 files changed, 74 insertions(+) (limited to 'src/video_core/texture_cache') diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h index 33ca051482..32775b8719 100644 --- a/src/video_core/texture_cache/texture_cache.h +++ b/src/video_core/texture_cache/texture_cache.h @@ -956,6 +956,73 @@ bool TextureCache

::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst, return true; } +template +bool TextureCache

::TemporalUpscale(const Tegra::Engines::Fermi2D::Surface& dst, + const Tegra::Engines::Fermi2D::Surface& src, + const Tegra::Engines::Fermi2D::Surface& motion_vectors, + std::array jitter, std::array motion_scale) { + if constexpr (FRAMEBUFFER_BLITS) { + return false; + } else { + const Tegra::Engines::Fermi2D::Config copy{ + .operation = Tegra::Engines::Fermi2D::Operation::SrcCopy, + .filter = Tegra::Engines::Fermi2D::Filter::Bilinear, + .must_accelerate = true, + .dst_x0 = 0, + .dst_y0 = 0, + .dst_x1 = static_cast(dst.width), + .dst_y1 = static_cast(dst.height), + .src_x0 = 0, + .src_y0 = 0, + .src_x1 = static_cast(src.width), + .src_y1 = static_cast(src.height), + }; + const auto result = GetBlitImages(dst, src, copy); + if (!result) { + return false; + } + const ImageInfo motion_info(motion_vectors); + has_deleted_images = false; + const ImageId motion_id = + FindOrInsertImage(motion_info, motion_vectors.Address(), RelaxedOptions{}); + if (has_deleted_images || !motion_id) { + return false; // the colour and output ids may be stale; the caller blits instead + } + const ImageId dst_id = result->dst_id; + const ImageId src_id = result->src_id; + PrepareImage(src_id, false, false); + PrepareImage(motion_id, false, false); + PrepareImage(dst_id, true, false); + for (const ImageId id : {src_id, motion_id, dst_id}) { + if (True(slot_images[id].flags & ImageFlagBits::Rescaled) || + slot_images[id].info.num_samples != 1) { + return false; + } + } + const auto view = [&](ImageId id, const Tegra::Engines::Fermi2D::Surface& surface, + PixelFormat format) -> std::optional> { + const std::optional base = slot_images[id].TryFindBase(surface.Address()); + if (!base) { + return std::nullopt; + } + const SubresourceRange range{.base = *base, .extent = {1, 1}}; + return RenderTargetFromImage(id, ImageViewInfo(ImageViewType::e2D, format, range)); + }; + const auto src_view = view(src_id, src, result->src_format); + const auto motion_view = view(motion_id, motion_vectors, motion_info.format); + const auto dst_view = view(dst_id, dst, result->dst_format); + if (!src_view || !motion_view || !dst_view) { + return false; + } + // After the views: creating them may reallocate the framebuffer slots. + Framebuffer* const dst_framebuffer = &slot_framebuffers[dst_view->first]; + return runtime.TemporalUpscale(dst_framebuffer, slot_image_views[dst_view->second], + slot_image_views[src_view->second], + slot_image_views[motion_view->second], jitter, + motion_scale); + } +} + template std::pair TextureCache

::TryFindFramebufferImageView( const Tegra::FramebufferConfig& config, DAddr cpu_addr) { diff --git a/src/video_core/texture_cache/texture_cache_base.h b/src/video_core/texture_cache/texture_cache_base.h index 141e5b76d5..d4b0f597ae 100644 --- a/src/video_core/texture_cache/texture_cache_base.h +++ b/src/video_core/texture_cache/texture_cache_base.h @@ -276,6 +276,13 @@ public: const Tegra::Engines::Fermi2D::Surface& src, const Tegra::Engines::Fermi2D::Config& copy); + /// Stand-in for a DLSS evaluation: temporal upscale of src into dst (see the runtime's + /// TemporalUpscale). False when the images cannot take it; the caller then blits. + bool TemporalUpscale(const Tegra::Engines::Fermi2D::Surface& dst, + const Tegra::Engines::Fermi2D::Surface& src, + const Tegra::Engines::Fermi2D::Surface& motion_vectors, + std::array jitter, std::array motion_scale); + /// Try to find a cached image view in the given CPU address [[nodiscard]] std::pair TryFindFramebufferImageView( const Tegra::FramebufferConfig& config, DAddr cpu_addr); -- cgit v1.2.3