Yes, that's because the ColorKey color is blitted into the panel with the switch and it becomes transparent when the panel is blitted into the backbuffer. This problem is fixed now. The implementation is a bit tricky, so, there may exist some hardware that can't do it.
The panels are outdated but they are working. The bigest problem is the high number of temporary surface. The first half of the main panel has 70 temporary surfaces in use which is causing most of the trouble because the modern hardware doesn't provide a proper sopport for color keying. Also, one pixel can go through three blitting operation before it reaches the back buffer. There are better methods to build a panel but it's not a critical issue at a moment.
I'll get that as well, it's caused by the system running out of memory. There is a major memory leak somewhere in the client.
What are the requirements to run the D3D9Client? When I activate it in Orbiter_ng it complains about d3dx9_42.dll is missing. Do I need to install a runtime package?
You need to upgrade your DirectX 9.0c to at least Feb 2010 version, or to the newest (Jun 2010 version).
---------------------------------------------------------------
>>> ERROR: Could not load vessel module: DeltaGlider
>>> [Vessel::LoadModule | .\Vessel.cpp | 5442]
---------------------------------------------------------------Can confirm the significant FPS boost here! Over 100 FPS with the very poly-heavy SSU on the launch pad! With the DX7 client it was a challenge to get it over 30.
The D3D9Client doesn't change anything from any DLL.
Yes, I can reproduce that one. It's been around from the begining. It has probably something to do with the implementation of in-surface blitting. DirectX versions from 9 to 11 doesn't support any blitting actions where the target and source surfaces are the one and the same surface. So, the D3D9Client will clone the surface.
Then I guess something in SSU is incompatible with the D3D9Client as no pad animations are working.
OutputVS MeshTechVS(float3 posL : POSITION0, float3 normalL : NORMAL0, float2 tex0: TEXCOORD0)
{
// Zero output.
OutputVS outVS = (OutputVS)0;
// Apply a mesh group transformation matrix
float3 posX = mul(float4(posL, 1.0f), gGrpT).xyz;
float3 nrmX = mul(float4(normalL, 0.0f), gGrpT).xyz;
float3 posW = mul(float4(posX, 1.0f), gW).xyz;
// Convert transformed vertex position into a "screen" space using a combined (World, View and Projection) Matrix
outVS.posH = mul(float4(posW, 1.0f), gVP);
// A vector from the vertex to the camera
outVS.toCamW = -posW;
// No texture coordinate processing. Pass through.
outVS.tex0 = tex0;
outVS.normalW = mul(float4(nrmX, 0.0f), gW).xyz;
return outVS;
}
); and still getting a CTD on quitting Orbiter. Oh, and the oapiDebugString message space isn't cleared between sessions (When quitting to launchpad and starting a new scenario immediately).
