Emulator

Could you try to narrow down what revision between r5991 and r6115 broke it?

Yeah those were just two revisions I had on hand. I figured I should post as soon as I noticed it, but I won't have a lot of time until this weekend. I'll try and narrow it down then.

Fwiw, check whether the issue occurs in r5991 if you enable EFB access.

EFBAccessEnable = True is in both gfx_dx9.ini's but thanks for the thought.

Anyway I tried r6057, r6085, r6095, r6099, r6102 and none of them had the issue, r6103 had the issue, so it would seem that's the culprit, especially since it has to do with lighting calculations.

Thanks a bunch, check r6267, Tino committed a possible fix for this issue.

Unfortunately the issue remains, though nothing seems worse than before. If there are any dumps or some such that would be helpful in troubleshooting, let me know.

r6267 fixes the flicker of purple waves, for the most part (XFB enable)

Before: XFB: flicker of the waves
real XFB: strong flickering of the waves

After: XFB: slight flicker of the waves
real XFB: no flicker of the waves

but the color is the same

I have test DX11 and it is ok.
I've noticed, if I Enable/disable the "Safe texture cache" of a DX-Plugin, then a pattern in the middle. This changes if again to Enable/disable the "Safe texture cache"of a DX-Plugin.

lucas: enabling/disabling the safe texture cache can cause those artifacts if you do it while the game is running.

Also, any luck finding the actual revision at fault?

godisgovernment: as previously mentioned, r6103 is almost certainly the revision which caused this. r6267 was supposed to have a possible fix, but the behavior was unfortunately the same afterwards (though not any worse). I have not checked revisions past maybe r6273, so it may have been fixed somewhere in between as a side effect; I'll check sometime this weekend.

Issue 3474 has been merged into this issue.

Just confirming that the issue is still present in r6430.

i have it tested and have not more the problem, i my the update of MS-DX 9.0c (junni 2010) have fix it, but i can not testing it because i can not downgrading or kill DX-Runtime.

Can anyone confirm this to be fixed?

Sorry I didn't respond earlier. I don't have access to a machine that can run dolphin at the moment, so I don't have build numbers. But for the latest builds, as late as the 22nd, I confirmed that the issue is NOT fixed. I even tested this on another machine. Water highlights in this portion of dragon roost remain pink for the DX9 plugin. Opengl still renders correctly. I cannot test DX11.

Because this open against I'd write what I know now ..
This problem was solved with the Nvidia update XXXXX for Nvidia users.
I think an update from ATI or a workaround can help right now in Dolphin for DX9. As I got to test only a GeForce GTX 260 for, I can not say if this applies to all Nvidia users, but I'm assuming.
I think a workaround in Dolphin would have to figure out which revision caused this problem because I know that it's not used to this problem has been. But probably will soon be a patch from ATI. Tests on other graphics hardware would certainly make sense.
So I hope I could help here.

A brief summary of the current state of this issue, as the issue has now extended to the OpenGL plugin

DX9 Plugin: r6103 made the water highlights in this area pink instead of light blue. This has not been fixed as of r7311.

OpenGL Plugin: Water highlights remained blue until r6954. By r6960 the water highlights were pink. Mamario doesn't have builds between r6954 and r6960 so I don't know exactly which commit extended this issue. This has not been fixed as of r7311

DX11 Plugin: No idea, don't have a DX11 card.

Just to reiterate I am running on Windows x64 builds, I don't know if the issue extends to Linux or OSX OpenGL plugins, but I would suspect that it does.

Hopefully these build numbers can narrow down what causes the issue, and maybe we can see a fix for both the DX9 and OpenGL plugin. And if possible, the title of this issue should probably change to reflect that this is no longer DX9 specific.

Well I just got a new graphics card (Radeon HD 6870) and the issue has disappeared for the DX9 plugin (and is not present with the DX11 plugin) even for older revisions, such as r7311. The status of the issue remains the same for OpenGL (present in revisions r6960 and newer).

So clearly it's some combination of hardware/drivers and revisions that cause this issue. The only hardware I have is ATI/AMD, but as mentioned previously I did test this on a friend's machine last December, and the issue was present in DX9 with his nvidia 8800GTX.

Confirm this issue is still exist with DX9 backend

I'm on Nvidia, and i have the same issue

Autoran, can you zip up a memory card save game and instructions to reproduce?

skidau, save №1
move back, away from Dragon Roost Cavern, and you will be in the room with pink water

Looks fine here. Windows 7 64-Bit. 3.0-496. HD6650M

http://i.imgur.com/7AGJ3.jpg

r6103 PixelShaderGen.cpp
partial revert patch fix

Yo, unfortunately that fix isn't valid - the overflow checks very much belong there.

Well, it's just a revert, you're a coder not me

Yeah, of course. I was just mentioning it for reference ;)

Dragon Roost, has a lot of environmental problems, i don't think they will ever fix it. There is that wind and fire that dolphin isn't working well with.

Tev_fixes branch has a few problems here and there, but it actually fixes this issue

fifo log: http://markus.members.selfnet.de/dolphin/tww.7z

I've been playing this game all week and tried all backends and I couldn't reproduce the pink water, I think it's been fixed unless someone else wants to verify.

It's only at night.

Reported to be fixed in tev_fixes_new.