[d3d9] Fixed function fix a bunch of wine tests with FF texture coord…

[K0bin]

Fixes #3117

Master:

visual.c:28422: Driver string: "nvd3dum.dll"
visual.c:28423: Description string: "NVIDIA GeForce RTX 3090"
visual.c:28426: Device name string: "\.\DISPLAY1"
visual.c:28428: Driver version 32767.65535.65535.65535
visual.c:5709: Test failed: quad 4 has color 00000000, expected 0x0033cc00
visual.c:5375: Test failed: D3DTTFF_COUNT3 | D3DTTFF_PROJECTED (draws non-projected) - top right: pixel (341, 211) has color ff000000, expected ffffffff.
visual.c:5375: Test failed: D3DTTFF_COUNT3 | D3DTTFF_PROJECTED (draws non-projected) - top right: pixel (341, 224) has color ff000000, expected ffffffff.
visual.c:5375: Test failed: D3DTTFF_COUNT3 | D3DTTFF_PROJECTED (draws non-projected) - top right: pixel (359, 211) has color ff000000, expected ffffffff.
visual.c:5375: Test failed: D3DTTFF_COUNT3 | D3DTTFF_PROJECTED (draws non-projected) - top right: pixel (359, 224) has color ff000000, expected ffffffff.
visual.c:5375: Test failed: D3DTTFF_PROJECTED (like COUNT3 | PROJECTED, texcoord has only 3 components) - bottom right: pixel (401, 361) has color ff000000, expected ffffffff.
visual.c:5375: Test failed: D3DTTFF_PROJECTED (like COUNT3 | PROJECTED, texcoord has only 3 components) - bottom right: pixel (401, 419) has color ff000000, expected ffffffff.
visual.c:5375: Test failed: D3DTTFF_PROJECTED (like COUNT3 | PROJECTED, texcoord has only 3 components) - bottom right: pixel (479, 361) has color ff000000, expected ffffffff.
visual.c:5375: Test failed: D3DTTFF_PROJECTED (like COUNT3 | PROJECTED, texcoord has only 3 components) - bottom right: pixel (479, 419) has color ff000000, expected ffffffff.
visual.c:5375: Test failed: D3DTTFF_COUNT1 (draws non-projected) - top right: pixel (341, 209) has color ffffffff, expected ff000000.
visual.c:5375: Test failed: D3DTTFF_COUNT1 (draws non-projected) - top right: pixel (341, 226) has color ffffffff, expected ff000000.
visual.c:5375: Test failed: D3DTTFF_COUNT1 (draws non-projected) - top right: pixel (359, 209) has color ffffffff, expected ff000000.
visual.c:5375: Test failed: D3DTTFF_COUNT1 (draws non-projected) - top right: pixel (359, 226) has color ffffffff, expected ff000000.
visual.c:5979: Test failed: quad 3 has color 00000000, expected 0x000000ff
visual.c:5981: Test failed: quad 4 has color 00000000, expected 0x00ffffff
visual.c:6068: Test failed: quad 3 has color 00ffffff, expected 0x00ff0000
45d4:visual: 460 tests executed (0 marked as todo, 0 as flaky, 16 failures), 0 skipped.

PR:

visual.c:28422: Driver string: "nvd3dum.dll"
visual.c:28423: Description string: "NVIDIA GeForce RTX 3090"
visual.c:28426: Device name string: "\.\DISPLAY1"
visual.c:28428: Driver version 32767.65535.65535.65535
a4d0:visual: 460 tests executed (0 marked as todo, 0 as flaky, 0 failures), 0 skipped.

I'd appreciate testing for regressions with:
#3293 (maybe @mrdeathjr28) I still had the Apitrace of Coldfear and that is fine with the PR.
#3628 (@Blisto91) I still had the Project Snowblind Apitrace and that also renders fine with the PR.
and Sims 2 (maybe @tannisroot)

[tannisroot]

Didn't see any regressions with this build in Sims 2 compared to 2.3.1.

[nolrinale]

Hello guys, i've tested the latest changes in PSOBB and the texture movement issue is fixed everywhere!!, theres a little detail with the refraction effect used in some monsters and for the splashing effects in water that seems to be a little bit off, not sure if its related to this texture fix but I've attached here a small video showcasing it.

@K0bin @doitsujin Moreover I wanted to mention that at least in the case of PSOBB, this game is actually a DX8 game not a DX9, so to be able to see the improvements i had to use d3d8to9 + your latest d3d9.dll as overrides not sure if that also might affect the outcome either

2024-05-21.22-52-30.mp4

[K0bin]

@nolrinale Can you make an apitrace of the part that's not quite correct now?

[nolrinale]

@nolrinale Can you make an apitrace of the part that's not quite correct now?

To get the apitrace all I have to do is to override all the DLL's in the x32 folder and run the game as usual right?

[K0bin]

https://github.com/doitsujin/dxvk/wiki/Using-Apitrace

@nolrinale

[nolrinale]

@K0bin
I'm having troubles setting it up with your fixed d3d9.dll from the artifacts, and is not loading correctly for me. I'm getting the old versionw ithout the fixes instead....

However since you seem to have the game up and running from the previous main issue #3117, create an Episode 2 Party and use the /warp command between areas 6-12 all these areas have water where you can splash around and see the issue

[nolrinale]

@K0bin I tried again to get you an apitrace of the affected area, but it looks like it wont get me a trace if I load the game with the d3d8to9's d3d8.dll.... and I require it to be able to elevate the game to DX9 to be able to see the fixed changes.

If i load the game with the default d3d8.dll included in the apitrace package it does work and I can get a trace file but then I'm stuck with the old graphic issues, however in this specific case the water ripples/splash textures and refraction works fine (minus the broken moving textures) so I guess there was a regression somewhere in this PR that affected this functionality.

[K0bin]

@nolrinale You don't have to use the latest fixed DXVK version. Apitrace records the D3D9 commands and those should be identical regardless of whether you use a DXVK build with my recent fixes or an older one.

[nolrinale]

@K0bin oh lol I had no clue about that!! well here you have it
https://drive.google.com/file/d/1UEU3vOvYtv4FbaFd7TFeebuBnnvP2jH8/view?usp=sharing
The recording goes up to a little debug room at the end with lots of water where im walking in circles showing you the effect

[K0bin]

Nice job!