[Bug 20183] New: Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Summary: Black & White 2: texture corruption on higher
                    resolutions
           Product: Wine
           Version: 1.1.29
          Platform: PC-x86-64
        OS/Version: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: directx-d3d
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

Description: The game 'Black & White 2' suffers from texture corruption when
using higher resolutions. All resolutions above 1024x768 cause texture
corruption including but not limited to: edgy black surfaces, stripes, terrain
rendered in the wrong place. Wine keeps spitting out the following message ad
nauseam:

fixme:d3d:state_pscale >>>>>>>>>>>>>>>>> GL_INVALID_VALUE (0x501) from
glPointSize(...); @ state.c / 1489

Operating system: Gentoo GNU/Linux ~amd64
Graphics card: NVIDIA GeForce GTX 280
Graphics card driver version: 190.25
Wine version: 1.1.29
Native DLLs: d3dx9_25.dll, d3dx9_30.dll, d3dx9_36.dll

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Vitaliy Margolen  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |DUPLICATE
              Alias|Itzamna                     |

--- Comment #1 from Vitaliy Margolen   2009-09-27 11:30:07 ---
Duplicate

*** This bug has been marked as a duplicate of bug 14939 ***

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Vitaliy Margolen  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #2 from Vitaliy Margolen   2009-09-27 11:30:15 ---
Closing

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Itzamna  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|CLOSED                      |UNCONFIRMED
         Resolution|DUPLICATE                   |

--- Comment #3 from Itzamna   2009-09-27 11:35:34 ---
This game is NOT a duplicate of bug 14939.

First of all: this game is Black & White 2, not Black & White 1.

Second: Black & White 1 suffers texture corruption because of Wine's lack of
S3TC compression:

fixme:d3d_surface:surface_convert_format Cannot find a conversion function from
format WINED3DFMT_DXT1 to WINED3DFMT_A4R4G4B4

While Black & White 2 clearly gives a different error:

fixme:d3d:state_pscale >>>>>>>>>>>>>>>>> GL_INVALID_VALUE (0x501) from
glPointSize(...); @ state.c / 1489

This bug was fixed around wine 1.1.25, but regressed later.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Ben Klein  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #4 from Ben Klein   2009-09-27 18:14:23 ---
(In reply to comment #3)
> This game is NOT a duplicate of bug 14939.
>
> First of all: this game is Black & White 2, not Black & White 1.

Agreed.
> Second: Black & White 1 suffers texture corruption because of Wine's lack of
> S3TC compression:

Actually, it's the lack of DEcompression.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

--- Comment #5 from Itzamna   2009-09-28 03:27:55 ---
This bug is caused by invalid pixel point size scaling, which makes sense, as
there are two characteristics:

1) the bug gets more severe at higher resolutions
2) the bug itself looks like pieces of terrain are unfolded and enlarged while
it shouldn't be

The responsible code is part of function 'state_pscale' in
./dlls/wined3d/state.c:

static void state_pscale(DWORD state, IWineD3DStateBlockImpl *stateblock,
struct wined3d_context *context)
{
    /* TODO: Group this with the viewport */
    /*
     * POINTSCALEENABLE controls how point size value is treated. If set to
     * true, the point size is scaled with respect to height of viewport.
     * When set to false point size is in pixels.
     */

/* Default values */
    GLfloat att[3] = {1.0f, 0.0f, 0.0f};
    union {
        DWORD d;
        float f;
    } pointSize, A, B, C;

pointSize.d = stateblock->renderState[WINED3DRS_POINTSIZE];
    A.d = stateblock->renderState[WINED3DRS_POINTSCALE_A];
    B.d = stateblock->renderState[WINED3DRS_POINTSCALE_B];
    C.d = stateblock->renderState[WINED3DRS_POINTSCALE_C];

if(stateblock->renderState[WINED3DRS_POINTSCALEENABLE]) {
        GLfloat scaleFactor;
        float h = stateblock->viewport.Height;

if(pointSize.f < GL_LIMITS(pointsizemin)) {
            /*
             * Minimum valid point size for OpenGL is driver specific. For
Direct3D it is
             * 0.0f. This means that OpenGL will clamp really small point sizes
to the
             * driver minimum. To correct for this we need to multiply by the
scale factor when sizes
             * are less than 1.0f. scale_factor =  1.0f / point_size.
             */
            scaleFactor = pointSize.f / GL_LIMITS(pointsizemin);
            /* Clamp the point size, don't rely on the driver to do it. MacOS
says min point size
             * is 1.0, but then accepts points below that and draws too small
points
             */
            pointSize.f = GL_LIMITS(pointsizemin);
        } else if(pointSize.f > GL_LIMITS(pointsize)) {
            /* gl already scales the input to glPointSize,
             * d3d scales the result after the point size scale.
             * If the point size is bigger than the max size, use the
             * scaling to scale it bigger, and set the gl point size to max
             */
            scaleFactor = pointSize.f / GL_LIMITS(pointsize);
            TRACE("scale: %f\n", scaleFactor);
            pointSize.f = GL_LIMITS(pointsize);
        } else {
            scaleFactor = 0.0f;
        }
        scaleFactor = pow(h * scaleFactor, 2);

att[0] = A.f / scaleFactor;
        att[1] = B.f / scaleFactor;
        att[2] = C.f / scaleFactor;
    }

if(GL_SUPPORT(ARB_POINT_PARAMETERS)) {
        GL_EXTCALL(glPointParameterfvARB)(GL_POINT_DISTANCE_ATTENUATION_ARB,
att);
        checkGLcall("glPointParameterfvARB(GL_DISTANCE_ATTENUATION_ARB, ...)");
    }
    else if(GL_SUPPORT(EXT_POINT_PARAMETERS)) {
        GL_EXTCALL(glPointParameterfvEXT)(GL_DISTANCE_ATTENUATION_EXT, att);
        checkGLcall("glPointParameterfvEXT(GL_DISTANCE_ATTENUATION_EXT, ...)");
    } else if(stateblock->renderState[WINED3DRS_POINTSCALEENABLE]) {
        WARN("POINT_PARAMETERS not supported in this version of opengl\n");
    }

glPointSize(pointSize.f);
    checkGLcall("glPointSize(...);");
}

I'm trying to fix it, but I do not know what the minimal / maximal pixel point
sizes allowed in OpenGL are. Can a developer please help me?

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

--- Comment #6 from Ben Klein   2009-09-28 04:00:44 ---
(In reply to comment #5)
> The responsible code is part of function 'state_pscale' in
> ./dlls/wined3d/state.c:

For future reference: copying large chunks of source into bug comments is
unnecessary. An attachment or even just a line number will suffice.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

--- Comment #7 from Itzamna   2009-09-28 09:19:24 ---
(In reply to comment #6)
> (In reply to comment #5)
> > The responsible code is part of function 'state_pscale' in
> > ./dlls/wined3d/state.c:
>
> For future reference: copying large chunks of source into bug comments is
> unnecessary. An attachment or even just a line number will suffice.

Are you going to say anything useful, or are you just going to pick on small
omissions / inconveniences?

If you don't know what the pixel point scaling factor should be, kindly refrain
from posting.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Itzamna  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|1.1.29                      |1.1.30

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Jeff Zaroyko  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|1.1.30                      |1.1.29

--- Comment #8 from Jeff Zaroyko   2009-09-28 16:54:05 ---
Please don't change the original reported version.  A comment to say it's
present in later versions is enough.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

--- Comment #9 from Ben Klein   2009-09-28 17:15:29 ---
(In reply to comment #7)
> Are you going to say anything useful, or are you just going to pick on small
> omissions / inconveniences?
>
> If you don't know what the pixel point scaling factor should be, kindly refrain
> from posting.

"Too much information" is not a "small omission". Quite the opposite. Proper
use and etiquette of Bugzilla is still important. This is not a tool that only
you use.

OpenGL docs should say what the minimum/maximum point sizes are.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

--- Comment #10 from Itzamna   2009-09-28 17:32:02 ---
@ Ben Klein: I am sorry, I am new to Bugzilla. Bear with me as I take you
through the problem.

@ Jeff Zaroyko: I'm sorry - I also confirmed this bug to be present in 1.1.30.

I also researched the etiology: apparently, glPointSize returns a value of
0x501 (1 * 16^0 + 5 * 16^2 = 1281 in decimal numbers), while antialiasing is
enabled. If antialiasing is enabled, OpenGL does not allow an infinite range of
point sizes (source:
http://www.opengl.org/documentation/specs/man_pages/hardcopy/GL/html/gl/poin...
).

I do not know the range of valid point sizes, as it is defined in the program.
The only constant about point size ranges is whether the driver allows point
sizes < 0, and that differs among drivers too.
I'm digging through the source code now to try and find the range of valid
point sizes.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

--- Comment #11 from Ben Klein   2009-09-28 17:37:25 ---
(In reply to comment #10)
> @ Ben Klein: I am sorry, I am new to Bugzilla. Bear with me as I take you
> through the problem.

Apology accepted. :)
> If antialiasing is enabled, OpenGL does not allow an infinite range of
> point sizes (source:
> http://www.opengl.org/documentation/specs/man_pages/hardcopy/GL/html/gl/poin...).
> From your link:

To query the range of supported sizes ... call glGet with arguments
GL_POINT_SIZE_RANGE ...

Maybe this will help.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

--- Comment #12 from Itzamna   2009-09-28 18:17:38 ---
I'll add:

glGetFloatv(GL_POINT_SIZE_RANGE, &min &max);
printf("%f %f", min, max);

and see what I can come up with.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Paul Bredbury  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Wylda  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |[e-mail]
         Resolution|                            |FIXED

--- Comment #13 from Wylda  2010-09-19 05:42:27 CDT ---

No texture corruptions here - works correctly in clean wine environment. Tested
under:
 * wine-1.3.3
 * nVidia GT240 v195.36.31
 * full screen
 * highest graphical details
 * resolution 1900x1200
 * Debian Squeeze 32bit
 * No need for coping d3dx9_25.dll and d3dx9_36.dll

Marking fixed. Also no replay for 1 year.

[Bug 20183] Black & White 2: texture corruption on higher resolutions

http://bugs.winehq.org/show_bug.cgi?id=20183

Alexandre Julliard  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #14 from Alexandre Julliard  2010-10-01 13:56:57 CDT ---
Closing bugs fixed in 1.3.4.