 |
Rev 15536. Partially fixed by adding translucent keyword (acts as an exclusion flag). |
 |
Please revert this change. shadowcaulk is supposed to be a sealing texture (i.e. use of forceOpaque keyword). Adding translucent to this texture overrides the effect of forceOpaque, and anyone (me!) trying to dmap a map using shadowcaulk will now see leaks. Need a different solution to the problem. |
|
|
Will do that tonight if you don't do it first. (Changes were reverted.) |
 |
1. I would need a test map 2. Someone needs to investigate why this happens with shadowcaulk and not regular caulk. DMAP? 3. What is shadowcaulk anyway and how is it different to regular caulk? Presume I'm clueless about DR and mapping (I am) 4. Try face-flipping your shadowcaulk surface. |
|
|
1. I can provide a "leak-test" shadowcaulk test map. 3. Regular caulk allows light to pass through, which can cause problems if the models and/or other textures on the far side of the caulk brush aren't creating shadows. Shadowcaulk acts like regular caulk, but any faces painted with shadowcaulk will stop the light (cast shadows). 4. "face-flipping" ??? Whatever it is, we can't ask existing missions to go back in and face-flip their surfaces. |
|
|
Without delving deeply, I think the original request is implying that we exclude shadowcaulk and common/shadows brushes from fog painting. I recall attempting to use the noFog keyword and finding that it was only a dmap flag so we don't really have a "don't fog on this" keyword? |
|
|
Strangely, some code comments indicate this is not a dmap flag: // All non-translucent surfaces receive fog unless they are explicitly noFog bool ReceivesFog( void ) const { return ( IsDrawn() && !noFog && coverage != MC_TRANSLUCENT ); } |
|
|
Interaction.cpp // "invisible ink" lights and shaders if ( shader->Spectrum() != lightShader->Spectrum() ) { continue; } // Proposed fix if ( !shader->ReceivesFog() && lightShader-> IsFogLight() ) { continue; } |
|
|
Since IsDrawn is part of the ReceivesFog condition, the fix for 3662 should now fix this too. Indeed, some strange z-fighting in Heart of Lone Salvation seems to have been cured by this. |
|
|
Attached test map with shadowcaulk used for one worldspawn brush. This dmaps okay with the latest SVN, but please use it for a check when changing shadowcaulk, to make sure shadowcaulk still seals. |
|
|
Thanks! I think that my noFog changes bypass the issue originally reported here but it might be worthwhile for Duzenko to study how shadowcaulk works in practice in case there is something amiss in his shadow casting code. |
 |
At note 11276, ideally there should be parity between how the fog is rendered in shadow stencil and map mode. By the sound of this, it may not be or, rather, there will be parity but it sounds like shadow textures may "appear" different than eg 2.06, when fog is in the mix. I wager this is probably a good enough fix but it remains to be seen how models with common/shadow textures behave. |
|
|
I've not witnessed any variance in fog\shadow scenes between the shadow modes since this change. For both modes this should prevent tris owned by materials that are "ineligible for fogging" to be submitted the vertex pipeline. (Also an optimization, I guess.) grayman's test map works fine HOLS works fine, as well as all the fogged missions I tried. If you have an example map where you are concerned about the behavior of shadows and fog, I'd be happy to review it. |
- Anonymous
