Commit Graph

82 Commits

Author SHA1 Message Date
refractionpcsx2 256babd337 GS/HW: Allow forcing Dither to 32bit for 16bit draws 2024-05-25 02:13:33 +01:00
Stenzek aa48256010 GS/HW: Round alpha values for indexed sample of RTA 2024-04-28 17:40:20 +02:00
lightningterror d824ae6e0c GS/HW: Extend blend second pass to more blend formulas.
`Cs + Cd*Ad, Cs - Cd*Ad, Cd*(1 - Ad), Cs*(1 + Ad), Cs*(1 - Ad).`
2024-04-23 02:00:02 +02:00
lightningterror 7dd7345e08 GS/HW: Minor shader optimization.
Use saturate instead of min max, saturate is faster than min max.
2024-04-21 15:10:53 +02:00
refractionpcsx2 30f4e77b31 GS/HW: Rearrange color on shuffle if SW Blend or TFX 2024-04-08 12:09:30 +01:00
lightningterror b1f4f67130 GS/HW: Add support for blend second pass.
Allows us to blend Cd with full alpha range of 0-2 bypassing hw blend limitations.
Not all Cd cases are covered, but it's a good start.

Also allows us to do Ad cases where we can double the blend to get the
proper blend result since Ad range is 0-1 instead of 0-2.
2024-04-07 20:04:50 +02:00
Stenzek a317e9c038 GS/HW: Improve accuracy of RGB_ONLY AFAIL
Instead of breaking the draw into two passes, which breaks when
fragments overlap each other and blending is enabled, use blending to
leave the value of Ad intact when a pixel fails the alpha test.

In the case of DATE being enabled, prefer PrimID over stencil, as since
we are changing Ad on a per-fragment basis, with some fragments not
being modified, stencil DATE will become desynchronized with the value
of Ad.
2024-03-28 11:46:20 +10:00
Stenzek 0917d49a01 GS: Remove separate alpha pass fallback
This is just wrong when overlap and depth writes are involved.
2024-03-28 11:46:20 +10:00
refractionpcsx2 059ec49389 GS/HW: RTA improvements and enhancements 2024-03-25 17:58:12 +00:00
refractionpcsx2 4ba43b8496 GS/HW: Improve channel + texture shuffle detection and processing 2024-03-24 11:38:33 +00:00
refractionpcsx2 bfef8397d6 GS/HW: Allow source is rt to use corrected alpha in most cases. 2024-03-17 17:24:39 +01:00
lightningterror 58628b8dd3 GS/HW: Make sure we don't multiply Cs if Ad is corrected.
Another potential to avoid copies.
2024-03-17 17:24:39 +01:00
lightningterror 71376ff4e6 GS/HW: Decorrect Ad on sw blend in tfx shader.
Otherwise if we do decorrection in covert shader we add more copies.

Also make sure to decorrect for fbmask, tex is fb.
2024-03-17 17:24:39 +01:00
lightningterror 6c9f132093 GS/HW: Add support for Ad (RTA) correction.
The idea is to adjust the alpha destination for more
accurate hw blending which will work on all renderers.

Old behavior has Ad in range within 0-1 whereas for blending 0-2 is needed.

copy rt -> adjust the alpha -> copy back the adjusted alpha-> restore old alpha after blending is done
2024-03-17 17:24:39 +01:00
lightningterror a7c5eebf99 GS/HW: Adjust blend mix for impossible blend.
Since we can't do Cd*(Alpha + 1) - Cs*Alpha in hw blend what we can do is adjust the Cs value that will be subtracted,
this way we can get a better result in hw blend. Result is still wrong but less wrong than before.
2024-03-14 16:16:11 +01:00
lightningterror c0e47767b7 GS/HW: Adjust dither on Blend Mix when Cs-Cd * Af blend. 2024-03-10 23:41:18 +00:00
refractionpcsx2 65649b3cbb GS/HW: Adjust dither on Blend Mix when Cs-Cd * As blend. 2024-03-10 23:41:18 +00:00
refractionpcsx2 4b88256df2 GS/HW: Allow palette lookups from depth and deswizzle manual deswizzles 2024-02-01 00:13:39 +00:00
refractionpcsx2 badede2e2d GS/HW: ignore lower 3 bits of 16bit color for AEM check 2024-01-15 14:45:00 +00:00
refractionpcsx2 01842a3c6b GS/HW: Allow blending on normal shuffles 2024-01-09 13:17:52 +00:00
Stenzek d9abe10308 Misc: Remove explicit PCH include, switch to SPDX 2023-12-24 14:03:14 +10:00
lightningterror f96e7a9156 GS/HW: Rename dfmt to dst_fmt.
It was confusing whenever the d meant depth or destination, this should make the code easier to read.
2023-10-19 19:25:13 +02:00
lightningterror 7fadb6ed05 GS/HW: Cleanup texture shuffle shaders. 2023-08-11 15:23:50 +01:00
lightningterror 713bd132dc GS/D3D: Add initial support for complex shuffles. 2023-08-11 13:14:19 +01:00
Stenzek 784118e9c1 GS/HW: Don't drop fractional colour before modulating
Fixes banding in Xenosaga, Tales of the Abyss, Beyond Good and Evil.
2023-07-28 16:31:35 +01:00
lightningterror 3dce969a5e GS-shaders: Add missing copyright headers. 2023-07-21 10:24:15 +02:00
Stenzek 7f7dd60587 GS/HW: VS expand instead of GS for DX/GL/Vulkan 2023-04-13 11:12:11 +01:00
Stenzek f569ad0970 GS/HW: Allow translating targets for depth sources
Fixes broken half screen in Miami Vice and 10 Pin - Champions Alley.
2023-04-09 17:56:49 +01:00
Stenzek faecc6913b GS/HW: Texture cache improvements
GS/HW: Only use temporary source for recursive draw

.. and don't insert it into the page map.

GS/HW: Lookup page list for depth sources

GS/HW: Avoid target copies by using shader sampling

GS/HW: Make texture cache a global pointer

GS/HW: Remove GetID() from GSTexture

It only made sense for OpenGL, was always zero in Vulkan.

GS/HW: Rewrite texture sampling hazard detection

Also avoid redundant channel shuffle setup.

GS/HW: Turn Haunting Ground CRC into an OI fix
2023-03-30 21:55:50 +01:00
Stenzek 435e73d838 GS/HW: Texture cache improvements
Change texture scale from vector to scalar

 - Independent X and Y scaling is long gone.
 - Also separate size and scale in TC lookup

Move clear value for texture to base class

Align heights to page size

 - Since FRAME and Z are in page units, we can't have two targets
 - overlapping within the same page.
 - Stops some small resizes too.
 - Test cases: Genji and Spider-Man 2 shadows.

Don't modify target TEX0 on shuffle/clear

Move upscale multiplier to uniform

Make P8 conversion page-aware

Fix incorrect depth preload shader

Improve HLE of texture shuffles

When a texture shuffle is split into two half-screen draws, we skip the
first, and draw the whole thing in the second, taking care of when both
the texture and framebuffer are offset.
2023-03-13 20:29:05 +00:00
lightningterror 06aed8491c GS-hw: Separate the Alpha masked Ad case from blend hw bit.
Allows for cleaner code.
2023-03-11 20:44:11 +01:00
lightningterror 62497b9300 GS-hw: Rename PS_CLR_HW to PS_BLEND_HW shader macro. 2023-03-11 20:44:11 +01:00
lightningterror 59aba9f757 GS-hw: Support Ad masked alpha on blend mix 1 and 2 clr cases. 2023-03-09 21:08:52 +01:00
lightningterror 911d35e800 GS-hw: Fix invert rounding for accumulation blend on gl.
Also make the checks consistent for all renderers.
2023-03-06 23:36:54 +01:00
lightningterror e8cf4822b1 GS-d3d: Automatically adjust color compensation based on rgb value for hw blend clr3 case.
Auto adjust when any color is higher than 128 ( 1.0f) to get more accurate color results.
2023-03-06 19:49:27 +01:00
lightningterror f478b3959c GS-d3d: Add clr3 case for blend mix 2.
When Cs*(Alpha + 1) overflows compensate with adjusting Alpha output for Cd*Alpha.
2023-03-03 16:35:10 +01:00
TellowKrinkle 7781907f0e GS: Blend truncation and dither goes the other way when subtracting
Truncation happens after subtraction, so it's the equivalent of rounding the value to subtract *up* instead of down
2023-02-26 07:22:32 +01:00
lightningterror 7f7950cd6b GS-d3d: Check each channel individually if it overflows and do corrections. 2023-02-24 02:06:00 +01:00
refractionpcsx2 925e874ada GS-HW: Fix real 16bit value shuffles 2023-02-17 19:06:35 +00:00
Stenzek 7d08a54ad9 GS/HW: Optimize TC source size based on CLAMP 2023-02-09 10:51:06 +00:00
Stenzek 0619555232 GS/HW: Use texel fetch/load instead of sampler for palettes
Saves a multiply in the shaders in some cases, and frees up one
descriptor slot.
2023-01-11 23:37:43 +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 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