 |
Bump. This got moved from the DR section. Has anyone tried the test map? I want to establish whether or not this is reproducible. If it is, there's several places where these spawnargs may be failing: the engine does not respect them, something in the material isn't set right, or it's shader-side. |
 |
Your .map file has the correct "nospecular" etc spawnargs set, so this is a TDM problem not a DR problem. I'm surprised no-one has reported this before, but I guess these spawnargs are little used in real maps. Even ambient lights have some specular in our lighting model. |
|
|
Ah, it's nice I'm not just going crazy. Use case scenarios for lights without specular would be where you want an ambient look, but also want to use light_center to point your light in a direction. Lights with the AmbientLight keyword don't have directionality through light_center, not to mention they don't cast shadows. This is how I came across this bug. A nodiffuse light could be used to fake Fresnel highlights I suppose, but it's hard to figure out without testing it first and seeing how it looks. |
 |
As I recall, there is a minimal hardcoded specular level in the light shader but the engine shouldn't be passing the specular texture if nospecular is set. One thing I notice is that UpdateLightDef (in RenderWorld.cpp ) doesn't have any nospecular criteria whereas it has other attribute criteria: if ( light ) { // if the shape of the light stays the same, we don't need to dump // any of our derived data, because shader parms are calculated every frame if ( rlight->axis == light->parms.axis && rlight->end == light->parms.end && rlight->lightCenter == light->parms.lightCenter && rlight->lightRadius == light->parms.lightRadius && rlight->noShadows == light->parms.noShadows && rlight->origin == light->parms.origin && rlight->parallel == light->parms.parallel && rlight->pointLight == light->parms.pointLight && rlight->right == light->parms.right && rlight->start == light->parms.start && rlight->target == light->parms.target && rlight->up == light->parms.up && rlight->shader == light->lightShader && rlight->prelightModel == light->parms.prelightModel ) { justUpdate = true; } else { // if we are updating shadows, the prelight model is no longer valid light->lightHasMoved = true; R_FreeLightDefDerivedData( light ); Edit: Adding noSpecular to the above has no effect on this bug... |
|
|
tr_render.cpp // Serp - This check is more expensive than just making extra calls, I am not sure if the calls might result in // things which should not be lit, being lit. Welp, we'll see. Reverting this does not cure the issue. |
|
|
That's all I can think of before looking into branches for the ARB bind. program.env[1] is already being used in the shader to determine whether substitute specular should be created... so I would need to create another new program.env variable if I were to go that route. |
|
|
Fix for nospecular in tr_render.cpp: case SL_SPECULAR: { // ignore stage that fails the condition if ( !surfaceRegs[ surfaceStage->conditionRegister ] ) { break; } // nbohr1more: 0004292 nospecular and nodiffuse fix else if ( backEnd.vLight->lightDef->parms.noSpecular ) { break; } else if ( inter.specularImage ) { RB_SubmittInteraction( &inter, DrawInteraction ); } R_SetDrawInteraction( surfaceStage, surfaceRegs, &inter.specularImage, inter.specularMatrix, inter.specularColor.ToFloatPtr() ); inter.specularColor[0] *= lightColor[0]; inter.specularColor[1] *= lightColor[1]; inter.specularColor[2] *= lightColor[2]; inter.specularColor[3] *= lightColor[3]; inter.vertexColor = surfaceStage->vertexColor; break; } nodiffuse and nobump will be much larger tasks. |
|
|
Src revision 6616 Binary 14584 |
- Anonymous
