mirror of
https://github.com/tangerineturtle1/NX2navi.git
synced 2026-11-15 19:19:58 -04:00
eed3e6b786
The post pass stores the square root of each channel, so the stand-in for Ocarina's untranslatable upscaler now filters those values and squares them in its compute pass (shaders/fallback_tonemap.spvasm, vk_tonemap.c) instead of decoding them with the sRGB curve, which darkened mid-tones and lifted the darkest values; the pass is compiled in again, its round-24 device loss having been the push descriptor call outside the recorder. NX2NAVI_FALLBACK_EXPOSURE is gone. AmpereLowering emitted SURED as 0xeb48, which matches no Maxwell instruction, so the translator skipped every image reduction; with 0xeb58 the edge AA's blend-weight pass writes its image. The translator's own warnings go to res.json notes, and shader dumps are named by program serial so they match the frame dumps' program numbers. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
93 lines
4.4 KiB
Plaintext
93 lines
4.4 KiB
Plaintext
; Scale and decode for the untranslatable scene upscaler (runtime/vk_exec_draw.c untranslated_dispatch_fallback).
|
|
; Its input is the post pass's picture with each channel stored as the square root of its linear value (prog 102 packs
|
|
; sqrt(composite) into RGB9E5), and its output is linear (the HUD is drawn over it, the frame's last pass encodes to
|
|
; sRGB). So: filter in the stored (sqrt) values, as the game's upscaler does, then square. The pass replaced a
|
|
; decode with the sRGB curve, which darkened mid-tones and lifted the darkest values.
|
|
; Written in SPIR-V assembly (spirv-as; no GLSL compiler in the build). As GLSL:
|
|
; layout(local_size_x = 8, local_size_y = 8) in;
|
|
; layout(binding = 0) uniform sampler2D src; // the stored picture (linear filtering)
|
|
; layout(binding = 1, rgba16f) uniform writeonly image2D dst; // the pass output (vk_tonemap.c), blitted into the target
|
|
; layout(push_constant) uniform P { vec2 inv; float unused; float pad; ivec2 size; } pc;
|
|
; void main() {
|
|
; ivec2 p = ivec2(gl_GlobalInvocationID.xy);
|
|
; if (p.x >= pc.size.x || p.y >= pc.size.y) return;
|
|
; vec3 s = textureLod(src, (vec2(p) + 0.5) * pc.inv, 0.0).rgb;
|
|
; imageStore(dst, p, vec4(s * s, 1.0));
|
|
; }
|
|
OpCapability Shader
|
|
OpMemoryModel Logical GLSL450
|
|
OpEntryPoint GLCompute %main "main" %gid %src %dst %pc
|
|
OpExecutionMode %main LocalSize 8 8 1
|
|
OpDecorate %gid BuiltIn GlobalInvocationId
|
|
OpDecorate %src DescriptorSet 0
|
|
OpDecorate %src Binding 0
|
|
OpDecorate %dst DescriptorSet 0
|
|
OpDecorate %dst Binding 1
|
|
OpDecorate %dst NonReadable
|
|
OpDecorate %P Block
|
|
OpMemberDecorate %P 0 Offset 0
|
|
OpMemberDecorate %P 1 Offset 8
|
|
OpMemberDecorate %P 2 Offset 12
|
|
OpMemberDecorate %P 3 Offset 16
|
|
%void = OpTypeVoid
|
|
%fnvoid = OpTypeFunction %void
|
|
%float = OpTypeFloat 32
|
|
%int = OpTypeInt 32 1
|
|
%uint = OpTypeInt 32 0
|
|
%bool = OpTypeBool
|
|
%v2float = OpTypeVector %float 2
|
|
%v3float = OpTypeVector %float 3
|
|
%v4float = OpTypeVector %float 4
|
|
%v2int = OpTypeVector %int 2
|
|
%v3uint = OpTypeVector %uint 3
|
|
%v2uint = OpTypeVector %uint 2
|
|
%ptr_in_v3uint = OpTypePointer Input %v3uint
|
|
%gid = OpVariable %ptr_in_v3uint Input
|
|
%img2d = OpTypeImage %float 2D 0 0 0 1 Unknown
|
|
%simg2d = OpTypeSampledImage %img2d
|
|
%ptr_uc_simg = OpTypePointer UniformConstant %simg2d
|
|
%src = OpVariable %ptr_uc_simg UniformConstant
|
|
%stimg = OpTypeImage %float 2D 0 0 0 2 Rgba16f
|
|
%ptr_uc_stimg = OpTypePointer UniformConstant %stimg
|
|
%dst = OpVariable %ptr_uc_stimg UniformConstant
|
|
%P = OpTypeStruct %v2float %float %float %v2int
|
|
%ptr_pc_P = OpTypePointer PushConstant %P
|
|
%pc = OpVariable %ptr_pc_P PushConstant
|
|
%ptr_pc_v2float = OpTypePointer PushConstant %v2float
|
|
%ptr_pc_v2int = OpTypePointer PushConstant %v2int
|
|
%int_0 = OpConstant %int 0
|
|
%int_3 = OpConstant %int 3
|
|
%float_0 = OpConstant %float 0
|
|
%float_1 = OpConstant %float 1
|
|
%float_half = OpConstant %float 0.5
|
|
%v2half = OpConstantComposite %v2float %float_half %float_half
|
|
%v2bool = OpTypeVector %bool 2
|
|
%main = OpFunction %void None %fnvoid
|
|
%entry = OpLabel
|
|
%g3 = OpLoad %v3uint %gid
|
|
%gu = OpVectorShuffle %v2uint %g3 %g3 0 1
|
|
%p = OpBitcast %v2int %gu
|
|
%psize = OpAccessChain %ptr_pc_v2int %pc %int_3
|
|
%size = OpLoad %v2int %psize
|
|
%ge = OpSGreaterThanEqual %v2bool %p %size
|
|
%out = OpAny %bool %ge
|
|
OpSelectionMerge %body None
|
|
OpBranchConditional %out %done %body
|
|
%done = OpLabel
|
|
OpReturn
|
|
%body = OpLabel
|
|
%pf = OpConvertSToF %v2float %p
|
|
%pfh = OpFAdd %v2float %pf %v2half
|
|
%pinv = OpAccessChain %ptr_pc_v2float %pc %int_0
|
|
%inv = OpLoad %v2float %pinv
|
|
%uv = OpFMul %v2float %pfh %inv
|
|
%si = OpLoad %simg2d %src
|
|
%s4 = OpImageSampleExplicitLod %v4float %si %uv Lod %float_0
|
|
%s3 = OpVectorShuffle %v3float %s4 %s4 0 1 2
|
|
%tc = OpFMul %v3float %s3 %s3
|
|
%o4 = OpCompositeConstruct %v4float %tc %float_1
|
|
%di = OpLoad %stimg %dst
|
|
OpImageWrite %di %p %o4
|
|
OpReturn
|
|
OpFunctionEnd
|