[T3D 1.1 Beta 3] Grid appears - LOGGED

#1

Quote:DirectX SDK : February 2010

If i recall correctly, this happens because you need to use DirectX SDK June 2010 instead.
Someone should really update the docs...

#2 Interestingly enough I used to get this artifact but it stopped with beta 3. I did update my DXSDK late in '10, so that might have been it.

#3 Is that from a stock T3D install? Try updating your DX SDK to June 2010. That was the version 1.1 Beta 3 was built and shipped on.

#4 I have posted about this issue several times. I use the latest DX, DXSDK and ATI drivers, and the issue has been present for me in every T3D version so far.

see here and here and here.

#5 +1

#6 @Scott Burns
Yes it's from stock T3D install, i updating DX SDK to June 2010 but the problem is always display. I try on my NVIDIA 8600 GS and there is no the bug, it seems to be link to ATI Card only.

Thanks all for the responses. ;)

#7 Thanks for the update Jean.

@Guy
Thanks for linking the older reports about it. The original ticket for it is still in the system and its on the list of tickets we're reviewing currently. According to the tickets this appears to be specific to lower range ATI cards only.

@Alfio
I'm afraid I don't follow what you mean. Are you saying you're experiencing this as well?

#8 Those "grid" lines almost appear to be the skirts on the generated terrain...
Is it possible this is some Z-buffering issue? Especially if it's lower range ATI only...is there some shader precision bug going on?

#9 @Scott
Yes, i have the same problems. With 1.0 and the 1.1 B3

Radeon HD 2600 Pro, on Windows 7 64 bit. DX June 2010 (sdk and redist)

#10 FYI, still a problem even now with an HD7770. Note, This is also being compiled against the Direct X June 2010 SDK.

EDIT: Further testing:
Simple answer,no easy way out of it. Not even disabling catalyst fixes it. Side note, TGEA works fine with blending though it's rather simplistic.. Will see if I can find a version of T3d that this isn't happening with.

If you comment out

if ( pass->layerTexConst->isValid() )
      desc.samplers[pass->layerTexConst->getSamplerRegister()] = GFXSamplerStateDesc::getClampPoint();

inside of terrain/terrCellMaterial.cpp in function TerrainCellMaterial::_createPass

You will end up with a result that shows the areas that are not being blended. The pattern I get on the terrain after this shows the problem areas that just aren't being covered by the blend. Which would actually match up fine, if there wasn't an edge being rendered.

I don't know enough about how the terrain is setup with it's shader to make sense of it, going to spend time familiarizing myself with it.

I had to agree with Bryan that it appeared to be a skirt issue. Though I've found that it doesn't just happen at the skirts.

I've removed skirts from the terrain.. The skirt artifacts are gone now, these though persist. Not to mention that the ones I'm seeing aren't near the skirt at all.

#11 Just recompiled current v3.0, with visual studio 2008, on windows 8 with dx9.0c .. I can say with the ATI HD7770 it's working now. Though only for T3D MIT v3.0. No driver changes.

Which proves to me it's an environmental issue between the OS, DX and Drivers. I don't believe there is much we can do in this case. I don't see any errors in calculations in the code, nor do I see range issues with the UVs (all I've checked have been within the acceptable min, max ranges). I can't find any inconsistencies in UV Mapping, Mip Mapping, or blending that should / or do cause this issue.
ATI Catalyst Version is 13.4