Projected lights

[OrbWeaver]

I did some investigating into why projected lights don't render properly, and this is what I found.

 

1. Projected versus omni lights are detected correctly in OpenGLShaderPass::flushRenderables(), which is where the maths calculations happen.

2. The projected light is rendered as if it were stuck at the origin, and does not move with the actual light volume (although the brightness does change if the light volume is moved, suggesting that the falloff texture coordinates are still changing).

3. The rendering calculations seem to make use of the origin of the light's AABB. This makes some sense for non-rotated omni lights, but not for projected lights as far as I can see.

4. The actual AABB for projected lights is not correct for rendering -- it only includes the selectable points rather than the whole volume (which extends on BOTH sides, not just towards the point). This is fine for the selection system but doesn't seem appropriate for the renderer.

5. I am not certain what RendererLight::offset() is required to do, but it seems to be returning the vector of the target point relative to the origin point.

 

I am not sure what the exact problem is because I don't fully understand the maths, but it looks like there is some confusion at a code level of how the calculation should be done. Perhaps some thought is needed regarding what methods should be on RendererLight and what values they should calculate, so that the omni and projected light calculations both make sense.

[AluminumHaste]

It would be incredible if this could be fixed, as it stands, having to place projected lights in the editor and then having to go into Doom 3 and dmap and then go into the map to see if everything lines up right, is really frustrating. Take your time man, even as it is this is still one of the best editors I've ever used.

[Komag]

I don't mind having to check in-game mainly because I'm already used to checking in-game for all other lighting anyway!

 

But I know what you mean, at least having SOME sort of indication right in the editor would be welcome.

[greebo]

I was about to compare the code to the GtkRadiant one some time in the past, but failed due to some compilation failures (couldn't get GtkRadiant run in my debugger, but I guess I didn't try hard enough). The code in GtkRadiant works fine, if I recall correctly, so to fix this we could just investigate the differences.

 

I haven't looked into the maths at all yet, so I can't really comment on these things.