Commit Graph

181 Commits

Author SHA1 Message Date
sideprojectslab 6924249bcf GS: Reverted field order in MAD shader updated weave shader to same field order as MAD and merged two Weave shaders into one
GS: corrected refactoring errors in interlace shaders
2022-11-18 21:29:37 +00:00
sideprojectslab 21fa2ee87b GS: Improved MAD compatibility with other graphics settings 2022-11-18 21:29:37 +00:00
sideprojectslab 64f6bf52b0 GS: Cleaned up and commented Motiond Adaptive Deinterlacing shader code
GS: fixed compile error in Metal interlace shader
GS fixed more compile errors in Metal interlace shader
GS: Adjusted code indentation in interlace shader for all renderers
2022-11-18 21:29:37 +00:00
sideprojectslab f10e7f4ab7 GS: Implemented Motion Adaptive Deinterlacing for all renderers
implemented FastMAD motion-adaptive deinterlacing for OpenGL renderer, other renderers will crash. FastMAD is replacing blend (either mode) so select blend to activate MAD under the hood

fixed an assert and assessed that one MUST select mode Blend bottom field first to enable MAD

removed forced mode 2, added separate motion thresholds for current field and alternate field motion and optimized MAD for Top-FIeld-First mode

committing kind-of broken status for review

the algorithm works well on most games, but somehow Kingdom Hearts works at half resolution

completely fixed weird artifacts on MAD, I only need to fix a 1-line offset that is causing the top of the screen to flicker

fixed flicker on first line, I still need to fine-tune some coefficients

solved all nastiness by realizing that MAD MUST work on an even resolution, so odd resolutions are rounded up. Now all games I tried look great

made MAD sensitivity adjustable inside GDDevice.h and passed to shaders as a parameters. For this purpose ZrH is now a vec4 to hold more parameters conveniently

ported MAD to DX11 and DX12

removed rounding of texture size to closest multiple of 2 and fized odd number of lines inside the shaders by also passing the vertical resolution as a parameter

improved compatibility of upper buffer offset adjustment for odd resolutions

added Vulkan support
2022-11-18 21:29:37 +00:00
lightningterror cdb886a454 GS-hw: Implement missing st_int shader bit.
Was causing bad shader errors in GT4.
2022-10-24 16:22:44 +02:00
lightningterror abcbdc4a25 GS-hw: Further improve how we handle blend mix 1.
Compensate slightly for Cd*(As + 1) - Cs*As.
Try to compensate a bit with substracting 1 (0.00392) * (Alpha + 1) from Cs.
The initial factor we chose is 1 (0.00392) as that is the minimum color Cd can be, then we multiply by alpha to get the minimum blended value it can be.
2022-10-18 15:20:41 +02:00
Connor McLaughlin c2823aae39 GS: Fix fractional scaling in GL/Vulkan 2022-10-17 15:26:01 +01:00
TellowKrinkle 1ad6605dfb GS:HW: Remove comment on rgba to 8i only-blue optimization
Doesn't seem to improve anything anymore (GPUs do this with csel now anyways, it's like 5 instructions with no divergence)
2022-10-17 10:28:27 +02:00
TellowKrinkle c6add663f8 GS:HW: Clean up ps_convert_rgba_8i 2022-10-17 10:28:27 +02:00
Connor McLaughlin 6c17f7ad49 GS: Support fractional upscale 2022-10-15 17:37:35 +01:00
TellowKrinkle 9b5dd92dad GS:HW: Use 16-bit unorm for HDR 2022-10-11 18:17:27 +02:00
refractionpcsx2 92cd6e8435 GS: Fix DX shader error 2022-10-09 04:53:18 +01:00
TellowKrinkle 0637682eb0 GS:HW: Properly handle fbmask of negative values
Previously was possible with blending and colclip, but now more common with the new hdr algorithm
2022-10-09 04:06:59 +01:00
TellowKrinkle 73ae9f8879 GS:HW: Make HDR algorithm more float-precision-friendly
- Use whole numbers instead of 255ths
- Use range -128 - 127 instead of 0 - 255 for accumulation blends
2022-10-07 20:44:53 +02:00
TellowKrinkle 1b34eb14f8 GS: Pass through primitive ID in geometry shaders in DX12 and VK
Fixes primid destination alpha on sprites
2022-09-06 21:07:56 +02:00
TellowKrinkle 412480b326 GS: Don't add 10 to date flag to indicate initialization
We already have 1 and 2 dedicated to the job
2022-09-06 21:07:56 +02:00
TellowKrinkle 62f9b16139 GS:OGL: Use blend min for primid destination alpha 2022-09-06 21:07:56 +02:00
TellowKrinkle ea81b74e3b GS:OGL: Add fetch_rt function to shader 2022-09-04 23:02:44 -05:00
lightningterror 6279ae63ce GS-hw: Remove remaining 24bit destination checks for Ad. 2022-09-01 10:28:42 +02:00
lightningterror edc82d77a5 GS-hw: Improve how we handle blending on 24bit with Ad factor in renderer. 2022-08-31 20:34:51 +02:00
TellowKrinkle 91601e5647 GS: Manually do bilinear sampling when converting RGBA to depth
Shader bilinear doesn't properly handle the case where r overflows into g (or g overflows into b, etc)
2022-08-20 10:12:24 +01:00
TellowKrinkle 882c09b870 Vulkan: Format convert.glsl 2022-08-20 10:12:24 +01:00
refractionpcsx2 5447da0588 GS: Don't do blend_mix on HDR 2022-08-16 20:54:55 +01:00
refractionpcsx2 1b18e02fe0 GS: Add constant adjustment in blend mix when reverse subtracting. 2022-08-16 20:54:55 +01:00
TellowKrinkle cb64e8d504 GS:HW: More accurate blend equation for blend mix 2022-08-13 00:01:23 +02:00
lightningterror 6b48cf574d GS-hw: Adjust how we handle specific blend mix cases.
Replace Cs*As + Cd*(1 - As) with Cs*As - Cd*(As - 1).
Replace Cs*F + Cd*(1 - F) with Cs*F - Cd*(F - 1).
As - 1 or F - 1 subtraction is only done for the dual source output (hw blending part)
since we are changing the equation.
Af will be replaced with As in shader and send it to dual source output.

Also check if A*Alpha in the shader overflows, if it does then adjust the
alpha that is sent for HW blending further to compensate.
2022-08-07 14:08:04 +02:00
refractionpcsx2 42b334efcd GS-HW: Use correct alphas when AA1 is enabled + ABE disabled. 2022-07-18 09:40:21 +01:00
Stuart Kenny f7d76ebf1d GS: Add lottes crt to present shader. 2022-06-22 16:26:19 +02:00
lightningterror d2904c1fd5 GS: Fix up wave filter shaders. 2022-06-14 21:25:47 +02:00
Connor McLaughlin a9819542d4 GS: Split convert and present shaders 2022-06-05 21:27:16 +01:00
Stuart Kenny 0b2536dd3e GS/HW: Fix typos in wave filter shader 2022-05-31 17:36:16 +01:00
refractionpcsx2 89d44a5f60 GS: Offset interlace when upscaling 2022-05-24 18:03:07 +01:00
Connor McLaughlin b0f20c9973 GS/DX11: Fix MMOD=1 merge background color 2022-04-18 04:29:19 +01:00
Connor McLaughlin 1993203d26 GS: Add Direct3D 12 renderer 2022-04-15 12:56:41 +01:00
TellowKrinkle 5ecaa9459d GS: Add Metal renderer 2022-03-31 23:51:27 -05:00
Connor McLaughlin d32d8cc0ec Shaders/FXAA: Fix warnings with Vulkan compiler 2022-03-28 09:24:51 +02:00
Connor McLaughlin 0cb9655523 GS/Vulkan: Remove unused color attribute
Would likely be getting optimized out anyway.
2022-03-28 09:24:51 +02:00
Connor McLaughlin 3067bef82d GS/Vulkan: Implement FXAA 2022-03-28 09:24:51 +02:00
Connor McLaughlin 486764be03 GS/Vulkan: Implement shadeboost 2022-03-28 09:24:51 +02:00
Connor McLaughlin 4a5180bc0a GS: Put shadeboost params in uniforms
Compiling variants for these is silly.
2022-03-28 09:24:51 +02:00
Connor McLaughlin 1927896442 GS: Fix some edge cases with fbfetch
Second alpha pass with A masked, DATE enabled, etc.
2022-03-21 20:24:09 +00:00
Connor McLaughlin 45682c382f GS: Add NO_COLOR flag to PS (depth-only) 2022-03-20 23:41:33 +00:00
Connor McLaughlin 19d310475b GS: Draw alpha pass when dual source blend is missing 2022-03-20 23:41:33 +00:00
Connor McLaughlin a8b9df3952 GS: Utilize GL_EXT_framebuffer_fetch where available 2022-03-20 23:41:33 +00:00
Connor McLaughlin bb75c78c1a GS: Add option to disable texture barriers/geometry shaders 2022-03-20 23:41:33 +00:00
Connor McLaughlin f73398ea84 GS/Vulkan: Fix warning when compiling RGBA8->RGB5A1 shader
Also gets rid of the program info log length check, since that's been
long removed in glslang.
2022-02-20 14:03:33 +00:00
lightningterror b8236b1b1c GS-hw: Partially revert #5540
Allow to wrap on colclip.
For real tho, ideal solution is sw blend, doing any clamp/wrap is wrong on blend mix.
2022-02-19 00:38:13 +01:00
lightningterror 1a3d77b2c0 GS-hw: Fix d3d11 depth sample shader.
Also adjust some comments to match updated shader names.
2022-02-18 19:21:01 +01:00
lightningterror 06766ddb98 GS-hw: Don't wrap on 16bit destination format for blend mix.
Helps Dog's Life blending, we shouldn't wrap as we do more blend math afterward which will be wrong if we wrapped before.
2022-02-18 19:00:54 +01:00
lightningterror 7151848406 GS-hw: Rename alpha_clamp to blend_mix. 2022-02-18 19:00:54 +01:00
Connor McLaughlin dfe4bc199f GS/HW: Use integers for depth conversion shaders
Fixes z-fighting in reflections in DBZ BT3, maybe others?
2022-02-15 10:41:37 +01:00
refractionpcsx2 d4b1d9abe5 GS-hw: Increase 32->16bit conversion accuracy
Improves Dogs life (no longer goes completely black right away, but has decal problems)
Vastly improves texture quality in Spider-Man 3 when using Framebuffer Conversion to fix the textures
2022-02-14 14:50:21 +00:00
refractionpcsx2 ad415945a6 GS-hw: Apply clamp/wrap when FBMask enabled 2022-02-11 19:33:38 +01:00
Connor McLaughlin 2684fd6b62 GS: Add depth copy convert shader 2022-02-03 15:20:35 +00:00
lightningterror 3ca4272230 GS-hw: Implement hw, hw/sw, sw blending on Ad when alpha write is masked.
Idea is to replace Ad with As when alpha write is masked,
then expand/let blend mix, accumulation blend non recursive blend or hw clr blend to
do the blending with Ad swapped as As.

We are doing this to try to bring some originally higher blending modes to lower levels
where we can do the draws with less texture barriers instead (gl/vk),
as for d3d11 this allows to run blending on the draws since previously the cases weren't handled properly,
it will be slower on d3d11 since we will be reading the frame buffer but it's better than nothing.

D3D11: It is enabled on Medium blending or higher, if draw is fbmask then it will enable
it on basic blending too.

OpenGL/Vulkan:
It is enabled based on the previous blending modes:
accumulation blend -> either minimum or basic level, depending on colclamp.
non recursive blend -> either minimum or basic level, depending on colclamp.
blend mix -> basic and higher level.
hw clr blend -> minimum and higher level.

All:
Prefer full sw blend when primitives don't overlap, sw fbmask or full barrier is used, it is more accurate.
2022-02-03 01:11:08 +01:00
lightningterror ae14afd5f7 GS-hw: Combine BLEND_C_CLR2_AF and BLEND_C_CLR3_AS in one shader bit.
Free a shader bit.
2022-02-03 01:11:08 +01:00
TellowKrinkle 6d0b9b3747 GS:HW: Don't scale rt when converting to texture
Scale texture coordinates instead
2022-02-01 01:24:35 +00:00
Connor McLaughlin f33ee27f56 GS: Get rid of extra binding for channel shuffle
Having this binding was redundant, as there's no "normal" texture
sampled when we're doing a channel shuffle, and it caused issues in
Vulkan when the render target or depth buffer the source.

Also fixes the Urban Chaos HLE shader.

Fixes validation errors in GT4, NFS: Carbon, Urban Chaos, probably
others too.
2022-01-30 14:17:12 +00:00
lightningterror 3b691da8d1 GS-hw: Implement alternative hw blending for Cs*Ad, Cs*Ad + Cd, Cd - Cs*Ad.
Alpha destination value is wrong so let us try to compensate.

Multiply Cs by (255/128) in shader.
Will work best if Cs is 0.5 or less,  if it's higher than 0.5 then the closer to 1 the less accurate it gets.

Ofc it is best to use sw blending but it will help if sw blending is not present/selected for specific draw, will help d3d11 quite more.
2022-01-27 12:09:35 +01:00
lightningterror c9216e5625 GS-hw: Remove useless shader line from CLR As/Af case. 2022-01-24 11:42:54 +01:00
lightningterror 93af9656d0 GS-hw: Fix clr 1 case shader.
Forgot to rename.
2022-01-24 11:11:38 +01:00
lightningterror 4ab9d1a493 GS-hw: Implement alternative hw blending for Cd*As, Cd*F.
Do hw blend for Cd*As, Cd*F, expand the clear color blend method as it is almost the same.

Most helpful for dx11 ofc, for gl/vulkan will help with rendering on lower
levels of blending without needing a barrier.
2022-01-23 20:59:35 +01:00
lightningterror 3c92054a6e GS-hw: Move PABE shader bit to the top of sw blending.
Early return when there is no sw blending, no need to run the blend code.
Optimization.
2022-01-11 13:15:56 +01:00
Connor McLaughlin 0f5b17de0f GS/DX11: Implement automatic/manual LOD in pixel shader 2022-01-11 12:30:11 +01:00
Connor McLaughlin 68f18245a7 GS/DX11: Implement merge feedback write 2022-01-09 16:35:34 +01:00
Connor McLaughlin 1a8a5a5e8e GS: Add Vulkan renderer 2022-01-08 05:19:45 +00:00
Connor McLaughlin 0c36647506 GS: Rewrite presentation interface and OSD 2022-01-02 15:06:39 +00:00
Connor McLaughlin 2330d3a056 GS: Use hardware point/line expansion where supported in GL 2022-01-02 15:06:39 +00:00
Connor McLaughlin bce54a64ff GS: Use flat/nointerpolation for IIP=0 instead of GS 2022-01-02 15:06:39 +00:00
lightningterror 6a431daffc GS-hw: Extend Blend mix to work when alpha is higher than 1.
Clamp the alpha(As/Af) to 1 in the shader, it is already clamped to 1 in hw blend unit.
It should allow to better render the effects, still not fully accurate but it's closer to what we want.
Hopefully it helps D3D11 more.
2021-12-27 20:22:41 +01:00
Connor McLaughlin a2ae4ba441 GS/OpenGL: Get rid of interface blocks for convert shaders 2021-12-26 13:21:51 +00:00
Connor McLaughlin 4e1b2792c9 GS/OpenGL: Move some uniform buffers to program uniforms 2021-12-26 13:21:51 +00:00
Connor McLaughlin 6e064da536 GS/OpenGL: Use programs instead of separate pipelines 2021-12-26 13:21:51 +00:00
Sergeanur 21bc56a8ba GS: Simulate scan mask 2021-12-26 11:51:25 +00:00
Connor McLaughlin efed92b15c GS/OpenGL: Move scaling factor to shader constant 2021-12-22 11:41:57 +01:00
Connor McLaughlin 08c00eed1e GS: Unify constant buffers between renderers 2021-12-22 11:41:57 +01:00
lightningterror 9368e29d62 GS-d3d11:fx: Flip x and y dithering positions. 2021-12-19 22:24:30 +01:00
TellowKrinkle ae8751e9ca GS:DX11: Use GSHWDrawConfig selector structs 2021-12-19 22:21:10 +01:00
TellowKrinkle a11b8c4750 GS:OGL: Use GSHWDrawConfig selector structs 2021-12-19 22:21:10 +01:00
TellowKrinkle 962db4bfc4 Resources: Move fxaa.fx to out of dx11 directory
It was used by both DX and OGL
2021-12-12 06:10:47 -06:00
TellowKrinkle da651cabe4 Resources: Move shaders to shared shader/ directory 2021-12-12 06:10:47 -06:00