[d3d9] Kane & Lynch: Dead Men (8080) Unhandled render state D3DRS_ADAPTIVETESS_Y

[pchome]

The FOURCC value is KanL. I doubt it used somewhere in drivers, it looks like it was either for debugging purposes or to be treated as D3D9Format::Unknown (Edit: I mean "and applied with D3DRS_ADAPTIVETESS_Y". Name likely was "invented" to be unsupported format.).

Out of curiosity I created a hack to accept this format (where D3D9Format::ATOC check is in current code)

dxvk/src/d3d9/d3d9_device.cpp

Lines 1695 to 1697 in b4f2094

	// NV's driver hack for ATOC.
	if (unlikely(State == D3DRS_ADAPTIVETESS_Y)) {
	constexpr uint32_t AlphaToCoverageEnable = uint32_t(D3D9Format::ATOC);

and there is indeed noticeable difference (see screenshots). Actually I don't know what I'm doing, so this bug report is FYI.

Before

After

Software information

• Native dlls: d3dx9_31, d3dcompiler_47, d3dx9_34.dll (present in game's installdir)
• GFWL: xliveless (Bundled version failed to install)
• Maxed settings (upped to AA:8/AF:16).

System information

• Wine version: Proton-4.11-11 (local build)

Apitrace file(s)

• 'll try to do on request

Log files

• d3d9.log: kaneandlynch_d3d9.log

[Joshua-Ashton]

What you linked to was just my code for ADAPTIVENESS_Y and ATOC?

Please don't link random values up to ATOC... That can break things.

[Joshua-Ashton]

IDK what KanL is meant to do, I imagine its meant to signify to the driver that this is Kane and Lynch for driver hacks.

Is there an actual problem with the game btw?

[pchome]

What you linked to was just my code for ADAPTIVENESS_Y and ATOC?

Yes, for context.

Is there an actual problem with the game btw?

No, the game works good.

[Joshua-Ashton]

This isn't a bug then.

If there's any difference on native drivers that's noticable, let me know.
(ie. does it look like ATOC or another feature is enabled?)