Static light leaks through the wall

On my level the wall and ceiling are drawn like this:


Shot at 2007-09-02

There is light to be seen on the ceiling at the other side of the wall:


Shot at 2007-09-02

Now I know this can happen with dynamic lights, but with static lights and where the wall is between the two ceilings?
The light goes through the wall.
And yes the snap is on in WED, so there is space between the wall and ceiling.
Anybody an idea about this issue?

Edit: In A6 this is not happening, (this level is made with A7)

Ok why you don't check the aling of the blocks of the walls and the ceiling sees that solves your problem.If not let me know.

This is called "light bleeding" and happens with automatic atlas-mapping. I think this can happen if you compile your level into a single mesh with one uv-map. The borders of the single faces in the uv-map are probably too small and there is no room for overlapping of some colors over the borders of their uv-faces.

The Atlas-Mapper in Lightwave takes all that into account but I think the Gamestudio lightmapper still has those issues. I think only Conitec can help in this case. Mr. Stolipin has to re-work the light-mapper to prevent such problems.

It would not happen, if there where a correctly working automatic vertex-snap tool.

The walls are simply either overlapping, or having a small crack between,
since the vertecies are not on "integer" numbers.

The light-mapper / calculation of the static lights works actually
quite good, if the level is build properly.
But the way WED lets you edit blocks in the "standard" way will lead
to unclean alignment.
You can best check that by opening the level-file in a texteditor
and have a look at the vertex positions, they are very often
on numbers like 45.4403 or so, it should be 45 when modeled correctly.

Its partly to blame on WED and partly on the level-designer.

Damocles:
He said it happens in A7 but not in A6. So it does not come from cracks or snapping. It is because of the new map-compiler. I explained it in detail above.

But I agree: The integer issue also is well-known problem of WED.

Ok, never worked with A7. But the issue is also present on A6.

Frits: can you zoom really close to one of the corners of the wall, and
make a screenshot? Then you will see what I mean with the issue.

This is the topview, as you can see there are no gaps between the walls.
The light goes THROUGH the wall and not through a gap.



And this is the frontview, same here, light goes THROUGH the wall and no gaps.
Snap in WED is on with a value of 1.



The walls are 8 quants thick. The only way I can prevend this to happen is to set the HiRes in the properties from the lightsource.
In Map Properties; Mesh size = 0.
And like I said this is NOT happening in version 6, so I think that Frank has a point here.

Yes. HiRes-maps can help a little since a bigger uv-map with bigger light map texture is more precise at the uv-borders.

But I think that the mapping algorithm should be improved. Maybe we should inform Conitec about that issue.

Thanks Frank, but I had discribed this issue already in Ask Conitec (Static light, 7 sept.)
Now I have tested some more;


Shot at 2007-09-22

Here is the HiRes on for the light and the 'ceiling' left box is 8 quants and the 'ceiling' of the right box is 4 quants.

Has this also to do with the shadows?
I get a stange shadow when I move thw static light a little;


Shot at 2007-09-22
HiRes=on.


Shot at 2007-09-22
HiRes=off.

Another strange thing;


Shot at 2007-09-22

This is a model with cast=on but there should not be a shadow because the suncolor is set to 0,0,0.